paperID
stringlengths
36
36
pwc_id
stringlengths
8
47
arxiv_id
stringlengths
6
16
nips_id
float64
url_abs
stringlengths
18
329
url_pdf
stringlengths
18
742
title
stringlengths
8
325
abstract
stringlengths
1
7.27k
authors
stringlengths
2
7.06k
published
stringlengths
10
10
conference
stringlengths
12
47
conference_url_abs
stringlengths
16
198
conference_url_pdf
stringlengths
27
199
proceeding
stringlengths
6
47
taskID
stringlengths
7
1.44k
areaID
stringclasses
688 values
embedding
stringlengths
9.26k
12.5k
umap_embedding
stringlengths
29
44
7d464839-1c82-4c17-b186-5939cf4d0834
convtexttm-an-explainable-convolutional
null
null
https://aclanthology.org/2022.lrec-1.401
https://aclanthology.org/2022.lrec-1.401.pdf
ConvTextTM: An Explainable Convolutional Tsetlin Machine Framework for Text Classification
Recent advancements in natural language processing (NLP) have reshaped the industry, with powerful language models such as GPT-3 achieving superhuman performance on various tasks. However, the increasing complexity of such models turns them into “black boxes”, creating uncertainty about their internal operation and decision-making. Tsetlin Machine (TM) employs human-interpretable conjunctive clauses in propositional logic to solve complex pattern recognition problems and has demonstrated competitive performance in various NLP tasks. In this paper, we propose ConvTextTM, a novel convolutional TM architecture for text classification. While legacy TM solutions treat the whole text as a corpus-specific set-of-words (SOW), ConvTextTM breaks down the text into a sequence of text fragments. The convolution over the text fragments opens up for local position-aware analysis. Further, ConvTextTM eliminates the dependency on a corpus-specific vocabulary. Instead, it employs a generic SOW formed by the tokenization scheme of the Bidirectional Encoder Representations from Transformers (BERT). The convolution binds together the tokens, allowing ConvTextTM to address the out-of-vocabulary problem as well as spelling errors. We investigate the local explainability of our proposed method using clause-based features. Extensive experiments are conducted on seven datasets, to demonstrate that the accuracy of ConvTextTM is either superior or comparable to state-of-the-art baselines.
['Lei Jiao', 'Ole-Christoffer Granmo', 'Bimal Bhattarai']
null
null
null
null
lrec-2022-6
['document-classification']
['natural-language-processing']
[ 1.96279019e-01 3.76703382e-01 -3.36888254e-01 -5.44920743e-01 -6.80800438e-01 -5.31262040e-01 5.80383778e-01 1.67422891e-01 -2.29765326e-01 5.74322462e-01 7.12968260e-02 -7.71200418e-01 1.65895477e-01 -9.25298214e-01 -9.82210577e-01 -4.98653382e-01 -1.66344270e-03 6.57391191e-01 2.14108899e-02 -2.58971542e-01 9.12166387e-02 -5.43231964e-02 -1.17688262e+00 1.07700241e+00 9.24601316e-01 1.23795140e+00 2.15898544e-01 3.73674303e-01 -8.78762841e-01 1.16429234e+00 -6.40461922e-01 -7.65839279e-01 4.30110209e-02 4.68735546e-02 -1.13986564e+00 -1.72451451e-01 4.06383187e-01 -1.33129045e-01 -2.74266154e-01 9.44630027e-01 -2.16063663e-01 -2.31308997e-01 3.18986326e-01 -1.26392770e+00 -8.27280998e-01 1.26371312e+00 -5.00917256e-01 -3.55632342e-02 2.43959785e-01 -6.93662688e-02 1.69382823e+00 -9.22363043e-01 4.16494787e-01 1.46719885e+00 5.99450350e-01 2.22230241e-01 -1.01573026e+00 -6.17507100e-01 4.76350218e-01 4.57726538e-01 -1.34804761e+00 -2.96954019e-03 5.03761172e-01 -1.40762761e-01 1.73435068e+00 2.27502838e-01 4.90313649e-01 1.07899523e+00 7.18246877e-01 1.30169761e+00 6.69950068e-01 -5.81726551e-01 4.56831530e-02 -6.21569157e-02 3.10411721e-01 8.35759401e-01 3.05321783e-01 -2.68610060e-01 -4.96281862e-01 3.74553986e-02 3.25221241e-01 1.56866927e-02 -1.96622983e-01 -2.31291130e-01 -1.21147108e+00 8.69573295e-01 5.85544825e-01 4.74916428e-01 -1.95154533e-01 3.91926110e-01 7.74425268e-01 2.34293163e-01 3.92171144e-01 4.89768505e-01 -9.31101024e-01 7.95740932e-02 -9.12380755e-01 3.37004006e-01 7.91231155e-01 1.34453201e+00 4.67150718e-01 -1.17264658e-01 -2.56351858e-01 5.35177886e-01 3.83889884e-01 3.09101343e-01 4.91724133e-01 -3.65742922e-01 1.12690926e+00 9.15428400e-01 -3.53871167e-01 -9.65919316e-01 -4.03750181e-01 -5.51212907e-01 -9.51342404e-01 -4.09334958e-01 2.80881971e-01 4.56894711e-02 -8.26834619e-01 1.49476254e+00 -7.60025829e-02 -1.85054943e-01 1.61156401e-01 6.44948602e-01 6.95151925e-01 9.98656988e-01 2.28987426e-01 1.51903152e-01 1.65169525e+00 -1.20144999e+00 -6.73939049e-01 -5.62172055e-01 9.38902736e-01 -4.22953874e-01 9.74618435e-01 6.04809105e-01 -1.01078951e+00 -3.15817207e-01 -1.03031981e+00 -5.60275853e-01 -5.16066551e-01 1.60562083e-01 8.68968725e-01 2.28049383e-01 -6.84652686e-01 4.08023983e-01 -7.32499957e-01 -1.38707563e-01 6.06937408e-01 5.15770197e-01 -4.42410171e-01 -1.29310533e-01 -1.50879776e+00 7.88477242e-01 7.99558759e-01 5.57328999e-01 -5.37730694e-01 -6.97389960e-01 -1.20302474e+00 5.28109133e-01 6.06233895e-01 -7.69095898e-01 1.46947062e+00 -1.06069601e+00 -1.28503788e+00 7.09943116e-01 -3.82310301e-01 -8.50906193e-01 3.56081784e-01 -2.95449197e-01 -3.11805338e-01 -3.40160280e-02 1.25993714e-01 6.25050426e-01 8.28487754e-01 -7.72571862e-01 -9.47760105e-01 -1.83832243e-01 2.17541307e-01 -6.59937188e-02 -1.02450766e-01 1.00637868e-01 -3.76750976e-01 -6.62339330e-01 1.87013239e-01 -6.71469748e-01 9.19136629e-02 -2.20906153e-01 -7.75332868e-01 -6.14667535e-01 6.83333278e-01 -4.54704493e-01 1.24712920e+00 -2.06170225e+00 5.80458678e-02 4.60860319e-02 3.95709455e-01 1.46743074e-01 -1.16983891e-01 5.50006688e-01 -2.44265154e-01 5.92141151e-01 -1.65920496e-01 -5.21360576e-01 3.54470313e-01 5.51701605e-01 -8.99398804e-01 2.27182344e-01 6.86778486e-01 1.15701354e+00 -7.80361950e-01 -6.89725757e-01 1.61355019e-01 1.52611494e-01 -6.13468647e-01 -1.73043475e-01 -7.46324658e-01 -1.41234994e-01 -6.34027004e-01 6.92933500e-01 7.11659074e-01 -3.14136624e-01 5.01407206e-01 1.24603556e-02 -7.87190199e-02 8.29518616e-01 -7.32056797e-01 1.60933089e+00 -5.03531277e-01 8.33728611e-01 -2.48049483e-01 -1.15907168e+00 6.40398800e-01 3.62118751e-01 -2.59098387e-03 -6.90858722e-01 2.35288024e-01 1.18231997e-01 1.02562860e-01 -3.75009537e-01 6.23020291e-01 -2.40870714e-01 -3.17075640e-01 2.21064895e-01 1.65947810e-01 4.96967696e-02 1.68165490e-01 2.27764770e-01 1.10059893e+00 1.11503772e-01 4.66100961e-01 -2.58356065e-01 7.06490338e-01 1.90856591e-01 4.62194473e-01 7.18114078e-01 5.11001423e-02 3.07281494e-01 8.08632255e-01 -8.11676979e-01 -8.69055808e-01 -7.70051062e-01 -2.05172777e-01 1.10820806e+00 -5.00647388e-02 -8.27778101e-01 -5.54408789e-01 -8.12301934e-01 1.63360521e-01 9.47366357e-01 -6.80144727e-01 6.49940129e-03 -7.64400899e-01 -4.70497608e-01 8.68401825e-01 7.68148661e-01 6.89779520e-01 -1.12527192e+00 -3.50960910e-01 2.04146087e-01 -4.94691640e-01 -1.25412154e+00 -1.64962992e-01 6.07571006e-01 -7.55734861e-01 -9.58690643e-01 1.42993540e-01 -9.52776074e-01 5.89613497e-01 8.97368565e-02 1.23883820e+00 1.98181391e-01 1.29635939e-02 -2.21205086e-01 -4.83654946e-01 -5.95344245e-01 -1.99652538e-01 2.82104880e-01 -2.23335832e-01 -1.80382341e-01 7.07960069e-01 -1.57877266e-01 -5.33223115e-02 1.28089208e-02 -9.36599493e-01 3.55518669e-01 6.98650360e-01 1.05173564e+00 5.16330004e-01 4.18131411e-01 1.01847418e-01 -1.02243066e+00 4.74996448e-01 -3.84971201e-01 -5.71492434e-01 3.94167960e-01 -3.76973659e-01 4.69164699e-01 1.08519912e+00 -1.99929282e-01 -1.04683805e+00 -2.68032461e-01 8.11893344e-02 -2.05669090e-01 3.46709304e-02 8.53749633e-01 -2.14922190e-01 3.82738650e-01 2.30013922e-01 3.15600038e-01 -4.53895569e-01 -2.09602490e-01 3.47744823e-01 6.31201267e-01 4.39847976e-01 -8.46269667e-01 6.36519551e-01 2.85557240e-01 -1.91266388e-01 -4.98850405e-01 -1.13646924e+00 -2.20426083e-01 -5.18427610e-01 4.20560539e-01 7.82615960e-01 -7.97152460e-01 -1.03278458e+00 1.84668720e-01 -1.57625270e+00 -1.97924480e-01 -1.04342110e-01 1.28765821e-01 -3.47903699e-01 2.55736440e-01 -7.39094615e-01 -6.30709469e-01 -5.54438353e-01 -1.03366542e+00 1.28036439e+00 -2.06019267e-01 -3.20239395e-01 -9.93800342e-01 -3.71546924e-01 4.42517191e-01 1.13798998e-01 -1.21713303e-01 1.50925195e+00 -9.67815280e-01 -6.86789453e-01 -2.67605126e-01 -3.14180344e-01 2.66080528e-01 -3.08065325e-01 -2.48655044e-02 -1.03669727e+00 -8.52804109e-02 -3.77060995e-02 -3.37800473e-01 9.15803909e-01 2.30188817e-01 1.52500367e+00 -5.84466159e-01 -4.35291886e-01 5.88157475e-01 1.32886100e+00 -9.35538765e-03 7.19426751e-01 6.82942629e-01 6.57940865e-01 4.96141165e-01 4.88860130e-01 1.64737597e-01 5.24459183e-01 2.87366927e-01 6.17724419e-01 5.53553030e-02 2.90065199e-01 -5.90305567e-01 3.93250585e-01 7.10732400e-01 2.24601537e-01 -6.77009404e-01 -1.12305903e+00 3.40398908e-01 -2.00016356e+00 -9.57539380e-01 -2.03184009e-01 1.68760502e+00 7.99622178e-01 4.45994288e-01 -6.46370530e-01 2.56673306e-01 4.95803088e-01 8.82685483e-02 -2.01706171e-01 -8.09844196e-01 -1.62253767e-01 3.93812001e-01 4.79381740e-01 2.82028586e-01 -1.25706232e+00 1.24337900e+00 5.92026949e+00 9.24846947e-01 -1.20574498e+00 -1.81855798e-01 5.53026855e-01 3.00856531e-02 -2.30005860e-01 7.73400292e-02 -8.97953689e-01 1.83614552e-01 7.46421099e-01 -9.55233872e-02 3.03264827e-01 8.24958622e-01 1.09610725e-02 9.89174005e-03 -1.51095450e+00 7.01584280e-01 -2.74212770e-02 -1.48936129e+00 5.21431208e-01 -7.94018805e-02 3.68885875e-01 1.24134675e-01 2.98784729e-02 8.19052219e-01 1.44631132e-01 -1.30742192e+00 1.15173578e+00 8.05237889e-02 5.73892355e-01 -6.92859113e-01 1.10342705e+00 4.44744647e-01 -1.31536591e+00 -3.65784079e-01 -5.45941889e-01 -3.30958426e-01 -1.41571388e-01 4.17064548e-01 -1.01724434e+00 7.28771746e-01 7.34783232e-01 7.06292748e-01 -4.36224133e-01 4.32940215e-01 -5.69452345e-01 6.32527411e-01 -1.89599678e-01 -2.38633588e-01 6.70200884e-01 1.17884323e-01 3.47711027e-01 1.39040792e+00 -9.40862228e-04 8.02677646e-02 -6.97365822e-03 1.30468905e+00 -4.39454228e-01 1.03501827e-02 -3.86772156e-01 -3.41678739e-01 3.26132834e-01 1.12893927e+00 -3.82775128e-01 -5.06420314e-01 -5.01024485e-01 7.67380595e-01 5.82399786e-01 2.74654359e-01 -8.94562602e-01 -4.77915823e-01 5.16662717e-01 -7.49823898e-02 6.26838088e-01 -1.25375643e-01 -4.59187001e-01 -1.27506208e+00 4.48218971e-01 -9.55075264e-01 4.05736446e-01 -7.72875488e-01 -1.24550736e+00 6.54774249e-01 2.67180093e-02 -9.55009818e-01 -1.30152449e-01 -1.05943656e+00 -5.58001578e-01 7.33237803e-01 -1.68667781e+00 -1.39312494e+00 7.54537582e-02 3.83888721e-01 7.75006175e-01 -6.67872503e-02 8.21689427e-01 4.39797528e-02 -7.13329136e-01 7.18589723e-01 -4.40347530e-02 5.12671947e-01 5.06710768e-01 -1.37592173e+00 6.46723032e-01 7.13102341e-01 2.04396456e-01 1.09216452e+00 5.39355934e-01 -4.52020913e-01 -1.63469803e+00 -1.21320617e+00 1.42514908e+00 -4.08611417e-01 9.49614227e-01 -7.54268289e-01 -9.18474853e-01 1.12259114e+00 2.98641920e-01 -7.40827397e-02 5.15532672e-01 3.97191674e-01 -6.63089037e-01 -8.59956816e-02 -8.41687977e-01 5.14026105e-01 7.26453602e-01 -6.45607591e-01 -1.03049076e+00 3.61552984e-01 1.03066397e+00 -5.19477844e-01 -6.10097408e-01 1.10710435e-01 5.55490613e-01 -6.69829667e-01 6.03022099e-01 -8.98950338e-01 9.68769610e-01 -3.08080494e-01 -1.46086693e-01 -1.00466812e+00 -3.09223294e-01 -4.54141676e-01 -6.72661513e-02 9.41303611e-01 7.23202229e-01 -6.28457665e-01 6.43005013e-01 6.03847742e-01 -4.98162121e-01 -9.14520264e-01 -9.41731036e-01 -5.35443723e-01 3.46844614e-01 -9.10157084e-01 7.50696599e-01 9.59444463e-01 6.41182840e-01 5.31587064e-01 -3.77225690e-02 1.47066668e-01 2.86755681e-01 2.53524035e-01 4.32639152e-01 -1.04698646e+00 -2.23084986e-01 -5.34483254e-01 -2.72851050e-01 -1.24658048e+00 6.26688719e-01 -1.27770531e+00 9.74401385e-02 -1.50747418e+00 2.84192950e-01 -2.86130726e-01 -1.98832959e-01 1.03724420e+00 1.15789801e-01 -2.89394945e-01 2.10615963e-01 5.45227602e-02 -6.24147773e-01 5.03659129e-01 1.15276837e+00 -7.00890660e-01 1.55334845e-01 -4.20527279e-01 -7.29176283e-01 7.17042267e-01 7.06658244e-01 -5.03574848e-01 -2.33913228e-01 -1.02064276e+00 6.62273169e-01 -1.79850012e-01 3.58072102e-01 -6.67782962e-01 3.84876549e-01 -1.18899286e-01 3.78948092e-01 -7.02411294e-01 -6.40596822e-02 -1.04502416e+00 -3.25460643e-01 3.56524706e-01 -5.42310059e-01 1.15581125e-01 5.00694454e-01 5.12289464e-01 -4.66157496e-01 -1.76549882e-01 3.47676367e-01 -1.36707529e-01 -5.56571841e-01 1.00099422e-01 -5.12344122e-01 -1.10759772e-03 7.45456100e-01 -7.50534087e-02 -5.81677735e-01 -1.65464878e-02 -2.41237670e-01 5.78090847e-01 5.80416471e-02 4.58722264e-01 4.92800832e-01 -1.00457311e+00 -4.57286865e-01 2.49419630e-01 2.76924193e-01 5.02449512e-01 -3.45754400e-02 8.02869201e-01 -7.00549781e-01 1.14528179e+00 1.15423121e-01 -5.20755649e-01 -1.02769315e+00 6.33136749e-01 4.44193274e-01 -6.59129381e-01 -7.22516477e-01 8.97470355e-01 4.98847276e-01 -5.29258490e-01 3.25871080e-01 -1.39523721e+00 8.78084823e-02 -3.79031211e-01 5.37768066e-01 -2.78956860e-01 2.32257918e-01 -2.12546095e-01 -5.15301526e-01 1.30088180e-01 -3.94719094e-01 3.14734131e-01 1.28970766e+00 1.20180771e-01 -4.71762598e-01 3.42358351e-01 1.10674024e+00 -1.71819195e-01 -5.33926904e-01 -3.00414354e-01 2.23942071e-01 -8.08537900e-02 1.48716658e-01 -8.70671928e-01 -7.94161558e-01 1.06324148e+00 -2.72985071e-01 1.33740395e-01 8.61700714e-01 -1.11523844e-01 1.01700032e+00 9.31646705e-01 3.90799373e-01 -8.62251639e-01 -3.26263428e-01 1.08799851e+00 7.83755958e-01 -1.14000607e+00 -1.17482647e-01 -4.99959350e-01 -5.73810160e-01 1.45128608e+00 5.25117099e-01 8.51535499e-02 2.90603787e-01 3.82779062e-01 -2.69995689e-01 -2.54287004e-01 -1.21337807e+00 1.48266613e-01 2.84692496e-01 2.24443957e-01 6.80743992e-01 1.35697827e-01 -5.16560078e-02 1.10827255e+00 -4.49864149e-01 -9.75629613e-02 2.33744398e-01 9.19857323e-01 -3.66153270e-01 -1.13081908e+00 -2.98105568e-01 4.80035424e-01 -4.56075937e-01 -5.63117087e-01 -4.77518588e-01 1.07077074e+00 3.31949115e-01 8.58878851e-01 3.46233785e-01 -3.19951653e-01 2.06625655e-01 2.25658923e-01 3.43199819e-01 -6.80000544e-01 -8.31745923e-01 -2.48441145e-01 2.75590658e-01 -4.83116627e-01 -7.67873228e-02 -4.86533433e-01 -1.78098941e+00 -5.88341355e-01 -4.18216228e-01 2.15613618e-01 4.40811962e-01 1.35189474e+00 2.10312337e-01 7.36984551e-01 1.72433212e-01 -3.72669339e-01 -6.82461202e-01 -9.54325199e-01 -4.66368824e-01 1.33768022e-01 3.87160271e-01 -4.29376364e-01 -1.63352653e-01 -1.27270687e-02]
[9.62418270111084, 7.7499308586120605]
686daf99-f788-4700-9002-0289ed0653c3
automatic-face-reenactment
1602.02651
null
http://arxiv.org/abs/1602.02651v1
http://arxiv.org/pdf/1602.02651v1.pdf
Automatic Face Reenactment
We propose an image-based, facial reenactment system that replaces the face of an actor in an existing target video with the face of a user from a source video, while preserving the original target performance. Our system is fully automatic and does not require a database of source expressions. Instead, it is able to produce convincing reenactment results from a short source video captured with an off-the-shelf camera, such as a webcam, where the user performs arbitrary facial gestures. Our reenactment pipeline is conceived as part image retrieval and part face transfer: The image retrieval is based on temporal clustering of target frames and a novel image matching metric that combines appearance and motion to select candidate frames from the source video, while the face transfer uses a 2D warping strategy that preserves the user's identity. Our system excels in simplicity as it does not rely on a 3D face model, it is robust under head motion and does not require the source and target performance to be similar. We show convincing reenactment results for videos that we recorded ourselves and for low-quality footage taken from the Internet.
['Thorsten Thormaehlen', 'Pablo Garrido', 'Levi Valgaerts', 'Christian Theobalt', 'Patrick Perez', 'Ole Rehmsen']
2016-02-08
automatic-face-reenactment-1
http://openaccess.thecvf.com/content_cvpr_2014/html/Garrido_Automatic_Face_Reenactment_2014_CVPR_paper.html
http://openaccess.thecvf.com/content_cvpr_2014/papers/Garrido_Automatic_Face_Reenactment_2014_CVPR_paper.pdf
cvpr-2014-6
['face-reenactment', 'face-transfer']
['computer-vision', 'computer-vision']
[ 3.80287230e-01 -1.66393183e-02 7.58156851e-02 -5.91507673e-01 -7.74189353e-01 -6.16879940e-01 6.53173625e-01 -7.05513656e-01 -4.40095693e-01 2.87764370e-01 1.51659558e-02 2.70381600e-01 1.40006751e-01 -2.59777457e-01 -7.33579993e-01 -6.93918049e-01 1.72771066e-01 2.24016711e-01 2.79720306e-01 -2.55681604e-01 1.52418703e-01 8.22728395e-01 -1.86867738e+00 3.57622474e-01 1.11621246e-02 1.06214058e+00 -1.22506857e-01 7.24381387e-01 3.36247623e-01 6.30214036e-01 -5.32624006e-01 -6.15734518e-01 6.83359742e-01 -7.04140842e-01 -6.26753867e-01 5.20730793e-01 1.19419730e+00 -6.12860739e-01 -4.17906404e-01 9.95662153e-01 5.62553167e-01 1.09144427e-01 1.31548703e-01 -1.70069838e+00 -3.34572881e-01 -1.71779633e-01 -5.79806805e-01 -1.04530118e-01 8.02141130e-01 2.59350259e-02 4.41607237e-01 -9.73002911e-01 1.14104617e+00 1.45273745e+00 5.65882742e-01 8.67892325e-01 -1.25919390e+00 -1.11553931e+00 -2.31138855e-01 1.08657591e-01 -1.53759658e+00 -1.06301343e+00 7.98367143e-01 -2.32076615e-01 2.00376689e-01 2.95490146e-01 8.94538641e-01 1.18089628e+00 -8.00331533e-02 3.48547101e-01 8.50579083e-01 -4.06363577e-01 -2.96311341e-02 2.10928321e-01 -4.10997123e-01 8.55423093e-01 -2.88141042e-01 1.44875556e-01 -8.55430841e-01 -3.64357650e-01 1.04376709e+00 7.72225559e-02 -5.64332187e-01 -7.33467937e-01 -1.05533910e+00 3.97909850e-01 9.87681821e-02 4.33964491e-01 -2.26934522e-01 1.74654394e-01 8.90245587e-02 6.17874742e-01 1.62427410e-01 -2.24498250e-02 -2.20018744e-01 -1.08085185e-01 -1.38617861e+00 1.10932015e-01 6.96083307e-01 9.68478680e-01 7.50042379e-01 -1.41800359e-01 9.18222964e-02 5.51945210e-01 3.09374720e-01 5.43343365e-01 6.33161008e-01 -1.63593471e+00 -2.72192419e-01 4.00682062e-01 8.77156109e-02 -1.11723375e+00 1.32339209e-01 3.54739010e-01 -1.33347571e-01 5.57296336e-01 2.14471281e-01 -2.04529939e-03 -7.09589839e-01 1.85935116e+00 5.99361062e-01 5.06248653e-01 -1.94755748e-01 9.63752687e-01 8.38701308e-01 2.83146769e-01 -2.29674444e-01 -4.49796617e-01 1.20290923e+00 -7.35094249e-01 -5.79072595e-01 1.72645777e-01 1.60898685e-01 -1.02628922e+00 8.58563244e-01 3.67670506e-01 -1.34639275e+00 -6.69154704e-01 -8.66971493e-01 -4.03315248e-03 4.40952778e-02 -9.85195953e-03 1.11255057e-01 9.02660310e-01 -1.54827464e+00 6.52181089e-01 -5.34804225e-01 -8.07134032e-01 1.00541033e-01 7.08221436e-01 -1.24636364e+00 -2.13703816e-03 -7.72603810e-01 7.92699218e-01 2.21467510e-01 -2.25550026e-01 -8.50769281e-01 -5.39918065e-01 -7.82016516e-01 -1.81472287e-01 2.25220203e-01 -4.45768952e-01 1.24535394e+00 -2.04266763e+00 -1.90270960e+00 1.43157589e+00 -4.58117843e-01 2.68937601e-03 5.92467248e-01 -1.00212015e-01 -5.45451939e-01 7.22671270e-01 -2.91424673e-02 8.57695162e-01 1.58329701e+00 -1.11251068e+00 -5.15591085e-01 -4.05283630e-01 -1.39696345e-01 1.38689652e-01 -3.05662721e-01 4.88180518e-01 -9.87001896e-01 -6.65595889e-01 -4.40236777e-02 -1.23222721e+00 4.74837750e-01 5.60979009e-01 2.66595393e-01 2.05699578e-01 1.43687832e+00 -6.61914229e-01 6.55769229e-01 -2.61740351e+00 1.78784415e-01 3.43239576e-01 1.20593101e-01 2.60476977e-01 -4.31439489e-01 1.99867308e-01 -4.77263182e-01 -2.00843185e-01 7.68954977e-02 -4.02203470e-01 -4.03517723e-01 -1.26455426e-01 4.15372439e-02 8.38518858e-01 -6.75631687e-02 6.18938327e-01 -8.79022598e-01 -6.58052921e-01 9.18650031e-02 4.92394805e-01 -5.98035336e-01 5.41913509e-01 2.49432847e-01 4.42052960e-01 1.09599754e-01 6.88661397e-01 6.36613131e-01 2.64113009e-01 2.05215782e-01 -3.44712496e-01 2.67825257e-02 -3.75592649e-01 -1.09929574e+00 1.98526347e+00 4.14177626e-02 8.06825161e-01 5.89223266e-01 -4.25521135e-01 9.75624621e-01 5.53822339e-01 7.66885221e-01 -3.22830856e-01 3.61520886e-01 3.22140098e-01 -1.23329632e-01 -4.57504243e-01 2.88626939e-01 -1.56741098e-01 2.53420204e-01 6.74118578e-01 3.04772288e-01 -3.06677282e-01 3.76848951e-02 1.99858785e-01 9.74690199e-01 4.85625476e-01 4.42946376e-03 -8.19882378e-02 4.50502276e-01 -3.31590891e-01 2.93658942e-01 1.36773020e-01 -3.58317196e-01 8.52674067e-01 1.95018306e-01 -2.74470627e-01 -9.31078732e-01 -1.14929485e+00 2.09061816e-01 1.21079147e+00 4.99820635e-02 -4.11736131e-01 -1.03201711e+00 -4.57219750e-01 -1.46837696e-01 2.04152152e-01 -6.68852210e-01 -2.88160563e-01 -4.99930292e-01 -7.31017143e-02 7.01601207e-01 1.01485541e-02 6.29903615e-01 -9.06476140e-01 -7.26094842e-01 -1.15923189e-01 -2.22906321e-01 -9.67900097e-01 -1.11969006e+00 -6.21871769e-01 -6.65547729e-01 -1.48670566e+00 -8.35602880e-01 -9.29655015e-01 7.04716802e-01 4.06342417e-01 7.12766826e-01 1.83884144e-01 -2.83001661e-01 1.02979064e+00 -2.70712763e-01 4.41568345e-02 -4.19815063e-01 -5.32158017e-01 4.73454863e-01 6.77871764e-01 3.56187701e-01 -7.01166630e-01 -4.48539674e-01 4.25471842e-01 -9.57197547e-01 -5.31706698e-02 3.10434282e-01 5.79605281e-01 1.61835894e-01 -1.33264840e-01 9.69665274e-02 -4.71412063e-01 2.25886568e-01 -6.16314821e-02 -2.67424405e-01 2.24183813e-01 -2.63662934e-01 -4.04140532e-01 1.13192298e-01 -8.08492601e-01 -1.02300489e+00 4.66525823e-01 9.71964076e-02 -1.00402856e+00 -2.53585912e-02 -3.01966906e-01 -2.49587595e-01 -4.74603593e-01 4.19776350e-01 9.98811945e-02 5.00121176e-01 -3.45493972e-01 4.58556950e-01 7.81083584e-01 1.05573475e+00 -2.82001853e-01 9.44539666e-01 7.27885485e-01 -2.84777403e-01 -9.35704410e-01 -5.45146456e-03 -3.18987846e-01 -1.08245397e+00 -5.95136821e-01 6.16429150e-01 -8.06225717e-01 -9.19861615e-01 6.73507810e-01 -1.19430149e+00 -1.45010194e-02 -1.78908169e-01 4.14980471e-01 -7.27207124e-01 4.08092558e-01 -3.32762510e-01 -4.30571258e-01 -2.63665885e-01 -9.48975980e-01 1.16885900e+00 2.56959230e-01 -3.90110791e-01 -5.68765998e-01 -5.63775375e-03 3.00350368e-01 1.43514290e-01 3.86223704e-01 5.86921215e-01 -4.11524117e-01 -3.96850914e-01 -3.35355252e-01 -5.33325002e-02 2.79269189e-01 4.57988828e-01 4.93790805e-01 -1.05785656e+00 -6.24525607e-01 1.21816710e-01 -2.88810939e-01 3.28940630e-01 1.38616428e-01 6.40495181e-01 -2.76860654e-01 -2.30628833e-01 6.16914988e-01 9.37773108e-01 2.96043128e-01 7.32725084e-01 7.79763311e-02 4.12614048e-01 1.06838953e+00 4.80505764e-01 1.26588762e-01 -2.92491578e-02 8.53934467e-01 2.87729800e-01 -2.00996906e-01 -2.35290423e-01 -2.79661775e-01 8.35401058e-01 3.29659164e-01 -1.05395325e-01 3.02535951e-01 -4.92034256e-01 3.68537575e-01 -1.71152854e+00 -1.19156158e+00 3.65011662e-01 2.47158599e+00 8.16668689e-01 -4.34633762e-01 3.02360773e-01 -3.24753486e-02 9.79805112e-01 -2.08427846e-01 -4.71842766e-01 -1.19744472e-01 -1.51407509e-03 2.83423394e-01 -1.82776921e-03 5.24824560e-01 -8.28320324e-01 1.06229794e+00 6.49002171e+00 4.25562561e-01 -1.35946560e+00 1.98885694e-01 4.11858201e-01 -6.15916312e-01 1.95204511e-01 -1.51638657e-01 -2.74405956e-01 2.84058601e-01 8.83083880e-01 -2.90880084e-01 3.28448117e-01 8.21688831e-01 2.41333798e-01 -9.04388949e-02 -1.49926770e+00 1.47123682e+00 7.66460598e-01 -1.10265315e+00 7.83982053e-02 -2.20082793e-02 2.88401842e-01 -3.53461295e-01 8.36681202e-02 -1.03151433e-01 -6.83839992e-02 -9.98188496e-01 8.99258494e-01 5.13243973e-01 1.17919290e+00 -6.76705480e-01 3.24733198e-01 -1.69930786e-01 -1.02805054e+00 1.14997551e-01 1.89298559e-02 2.17739850e-01 -1.32357478e-02 -2.87831992e-01 -7.07412064e-01 3.36634368e-02 9.45286155e-01 6.09428227e-01 -6.00884557e-01 7.35969365e-01 1.94240451e-01 -1.68755706e-02 -2.61537552e-01 5.16434133e-01 -2.89480209e-01 -2.18665615e-01 5.89374959e-01 9.76736426e-01 4.27388221e-01 2.69928485e-01 -4.18531476e-03 2.24533737e-01 -2.80294210e-01 2.83237666e-01 -7.81879842e-01 3.23952496e-01 2.85539359e-01 1.23289514e+00 -5.07923722e-01 -1.90611705e-01 -3.44524682e-01 1.52154493e+00 -1.06895342e-01 2.00665787e-01 -7.95679092e-01 -3.28178853e-01 7.98695207e-01 3.96866918e-01 1.41474217e-01 4.50402461e-02 6.93034232e-01 -1.23267734e+00 -3.19527574e-02 -1.21832407e+00 1.96735442e-01 -1.17408133e+00 -8.27070475e-01 9.04615998e-01 1.56195447e-01 -1.16827047e+00 -5.21788180e-01 -4.20118213e-01 -4.80821162e-01 4.60401922e-01 -8.20700884e-01 -1.23850906e+00 -4.81145650e-01 1.37511718e+00 4.82227325e-01 -2.12864608e-01 8.03760648e-01 4.48657006e-01 -2.66106218e-01 6.57391250e-01 -2.15329960e-01 2.34437451e-01 1.30748367e+00 -4.89284456e-01 3.77771892e-02 7.74364054e-01 1.77039906e-01 5.87455034e-01 6.09125972e-01 -4.95314240e-01 -1.32605433e+00 -8.53740036e-01 7.89226115e-01 -4.86019313e-01 4.58124220e-01 -3.71665835e-01 -5.75769246e-01 8.52330565e-01 2.89535344e-01 3.11146602e-02 7.44870424e-01 -5.89930594e-01 -3.74344468e-01 -3.80347461e-01 -1.38885224e+00 7.19765842e-01 1.09983027e+00 -6.18352473e-01 -6.48165762e-01 9.64008868e-02 3.29627335e-01 -3.36320817e-01 -6.38906598e-01 9.74392146e-02 1.15656745e+00 -1.13230896e+00 7.99762309e-01 -4.75472420e-01 2.31707245e-02 -4.30786997e-01 -2.20243752e-01 -8.65792871e-01 -1.02493204e-01 -1.02710450e+00 2.25302070e-01 1.37504888e+00 -6.41270950e-02 -3.79418910e-01 7.37119734e-01 1.12775159e+00 5.48035562e-01 -9.88579616e-02 -1.14344895e+00 -6.67824447e-01 -5.89816809e-01 -7.81429484e-02 3.86874944e-01 9.21453416e-01 -4.12556380e-02 1.17942907e-01 -6.76279783e-01 -3.13408375e-02 6.76466048e-01 5.46829663e-02 1.20485270e+00 -1.24319506e+00 -1.14042968e-01 -1.00591086e-01 -5.91502368e-01 -6.31033778e-01 4.20324326e-01 -6.54558241e-01 -2.33921409e-01 -5.55675566e-01 1.10359602e-01 8.23038071e-02 3.76632959e-02 6.49964869e-01 4.97105360e-01 8.42020929e-01 4.79895562e-01 5.73141575e-01 -3.40416491e-01 4.89594221e-01 7.80304670e-01 8.42064768e-02 -2.37238824e-01 -1.43577427e-01 -3.73238146e-01 9.29283738e-01 4.01054144e-01 -5.34425020e-01 -3.84061366e-01 -2.06551328e-01 -3.75738084e-01 1.72133386e-01 4.40549463e-01 -8.80072057e-01 2.39065021e-01 -2.11624075e-02 5.76003432e-01 -6.72375560e-02 6.97801113e-01 -1.19363773e+00 7.80200779e-01 4.98779833e-01 -1.05650045e-01 2.56558865e-01 1.58795327e-01 2.87212819e-01 -1.33169562e-01 -1.70975029e-01 1.25011098e+00 -8.32478479e-02 -7.59805024e-01 3.62175316e-01 -3.49628121e-01 -4.28097636e-01 1.38021874e+00 -5.62760890e-01 1.20961733e-01 -7.16600657e-01 -7.38722861e-01 -3.53299916e-01 1.18279314e+00 6.29241168e-01 7.27649570e-01 -1.48374915e+00 -7.24843323e-01 5.98063588e-01 -1.16753250e-01 -7.27104723e-01 -3.66203934e-02 6.52826369e-01 -6.59567356e-01 -1.66297719e-01 -5.99574149e-01 -6.37421310e-01 -2.08913064e+00 5.03191769e-01 4.36918259e-01 5.47619104e-01 -4.71281022e-01 6.82742953e-01 1.84748217e-01 -4.25992012e-02 1.31819665e-01 2.02600703e-01 1.07502349e-01 1.57503158e-01 8.00105393e-01 2.70212382e-01 -2.69305836e-02 -1.32381546e+00 -4.73183811e-01 8.00650835e-01 1.43888459e-01 -5.99518716e-01 1.07927144e+00 -3.48083526e-01 -3.02974701e-01 2.55535811e-01 1.43993616e+00 2.37620622e-01 -1.28305852e+00 -2.77952373e-01 -3.57723057e-01 -1.08056128e+00 -1.03236541e-01 -4.21876043e-01 -1.31959784e+00 4.33900297e-01 8.09002161e-01 -4.16509420e-01 1.54532886e+00 -4.81866561e-02 8.05498600e-01 2.45089144e-01 6.69906616e-01 -9.04181421e-01 4.33107972e-01 1.58503830e-01 1.05781090e+00 -1.01230705e+00 -1.38340160e-01 -1.76949143e-01 -4.50317025e-01 1.20975924e+00 4.96782362e-01 -9.56772119e-02 7.50758350e-01 7.21105859e-02 2.69843936e-01 -9.36620310e-02 -5.49716055e-01 -1.40288007e-03 1.43174425e-01 7.28612959e-01 1.93807602e-01 -5.32979250e-01 2.07593307e-01 9.78297889e-02 -2.85065055e-01 2.06686556e-01 5.12786984e-01 8.23697031e-01 -1.34695217e-01 -1.15883386e+00 -6.07990384e-01 -1.12153605e-01 -4.82217282e-01 -5.85006829e-03 -7.17063129e-01 8.69804144e-01 1.98754579e-01 8.55602860e-01 1.86473653e-01 -4.31741297e-01 4.04212236e-01 2.39008784e-01 8.43671679e-01 -5.53475082e-01 -5.58900833e-01 3.00516218e-01 -3.60589474e-01 -9.21310186e-01 -8.87276351e-01 -8.23987901e-01 -8.68635774e-01 -6.05476618e-01 -1.08852439e-01 7.03311712e-02 4.50422019e-01 5.35400212e-01 4.94742393e-01 -3.06844831e-01 8.29211533e-01 -1.39487481e+00 2.69039810e-01 -7.17903614e-01 -8.62141848e-01 7.31782317e-01 4.48238283e-01 -5.93320847e-01 -1.91265941e-01 7.33781695e-01]
[12.997709274291992, -0.3532937169075012]
be172ccc-aec9-4f6d-9acf-616b5221c43e
emotion-cause-pair-extraction-a-new-task-to
1906.01267
null
https://arxiv.org/abs/1906.01267v1
https://arxiv.org/pdf/1906.01267v1.pdf
Emotion-Cause Pair Extraction: A New Task to Emotion Analysis in Texts
Emotion cause extraction (ECE), the task aimed at extracting the potential causes behind certain emotions in text, has gained much attention in recent years due to its wide applications. However, it suffers from two shortcomings: 1) the emotion must be annotated before cause extraction in ECE, which greatly limits its applications in real-world scenarios; 2) the way to first annotate emotion and then extract the cause ignores the fact that they are mutually indicative. In this work, we propose a new task: emotion-cause pair extraction (ECPE), which aims to extract the potential pairs of emotions and corresponding causes in a document. We propose a 2-step approach to address this new ECPE task, which first performs individual emotion extraction and cause extraction via multi-task learning, and then conduct emotion-cause pairing and filtering. The experimental results on a benchmark emotion cause corpus prove the feasibility of the ECPE task as well as the effectiveness of our approach.
['Zixiang Ding', 'Rui Xia']
2019-06-04
emotion-cause-pair-extraction-a-new-task-to-1
https://aclanthology.org/P19-1096
https://aclanthology.org/P19-1096.pdf
acl-2019-7
['emotion-cause-pair-extraction', 'emotion-cause-extraction']
['natural-language-processing', 'natural-language-processing']
[ 3.90627444e-01 -2.25209996e-01 3.93886724e-03 -3.79849613e-01 -8.44379365e-01 -4.51745600e-01 5.59428215e-01 3.09496701e-01 -3.16146910e-01 7.35965967e-01 3.42677534e-01 -2.23693512e-02 -2.94471622e-01 -4.13918346e-01 -2.31001109e-01 -6.30423903e-01 -6.99457079e-02 -5.48836440e-02 -9.55264047e-02 -4.20062877e-02 9.30536613e-02 1.61603212e-01 -1.71007168e+00 3.34424168e-01 9.26146269e-01 1.17919910e+00 -1.77563161e-01 2.70580471e-01 -4.13269758e-01 1.00693595e+00 -6.73073351e-01 -4.79324728e-01 -2.87295163e-01 -6.11105442e-01 -9.13592756e-01 -1.08408347e-01 -2.54162759e-01 5.69150783e-02 3.07647020e-01 9.94203687e-01 4.35674816e-01 -6.37402534e-02 6.53133869e-01 -1.79908907e+00 -7.08573163e-02 6.16926372e-01 -6.36301756e-01 -2.02306435e-01 5.84190607e-01 -4.68518019e-01 1.41971087e+00 -7.75376081e-01 3.26764852e-01 1.30259347e+00 5.15212476e-01 5.28684139e-01 -6.68169379e-01 -9.00960028e-01 2.74799585e-01 4.00760025e-01 -1.14535606e+00 -1.25231341e-01 1.01995885e+00 -2.13429898e-01 9.16486621e-01 2.35540763e-01 4.33903754e-01 9.76853192e-01 1.40374452e-01 1.11593747e+00 1.32263637e+00 -3.04566681e-01 3.34007651e-01 -4.75621074e-02 3.37617457e-01 3.10198963e-01 -2.61447102e-01 -4.01342601e-01 -5.02382934e-01 -4.59541619e-01 2.83961564e-01 -3.51102561e-01 -2.73463100e-01 1.25404984e-01 -1.11584997e+00 5.92756391e-01 8.54160935e-02 4.76642847e-01 -6.09808564e-01 -4.25276384e-02 8.41739833e-01 2.52883941e-01 5.96730709e-01 3.45801175e-01 -7.80080855e-01 -2.93574780e-01 -5.24317741e-01 3.63649845e-01 8.45951796e-01 5.62793434e-01 4.87534672e-01 -4.06721711e-01 5.20212911e-02 9.99541521e-01 1.13001257e-01 3.11776876e-01 3.48444700e-01 -2.89466351e-01 2.51168221e-01 8.29192817e-01 1.56467766e-01 -1.23566532e+00 -5.73311925e-01 -9.31255072e-02 -8.51223230e-01 -3.24451566e-01 2.17848077e-01 -5.39102256e-01 -4.47838455e-01 1.82624495e+00 5.38522720e-01 3.24746579e-01 3.11788708e-01 9.79777098e-01 1.13631725e+00 8.54714870e-01 3.75056297e-01 -6.05451107e-01 1.62607002e+00 -5.00834227e-01 -1.23272622e+00 -2.56005496e-01 5.61005592e-01 -9.22115088e-01 9.93494213e-01 6.23957396e-01 -4.69403416e-01 -1.71187535e-01 -7.94255793e-01 9.68375131e-02 -4.12474692e-01 3.45756382e-01 9.73723769e-01 4.09452468e-01 -2.77016997e-01 4.69746888e-02 -3.95460397e-01 -1.82429180e-01 2.31037214e-01 4.73248303e-01 -2.18420729e-01 1.66283965e-01 -1.77452326e+00 6.51166975e-01 5.12776732e-01 1.86470494e-01 -3.85423452e-01 -5.35668135e-01 -8.62586498e-01 8.18187147e-02 7.55433679e-01 -1.83071733e-01 1.12296331e+00 -1.12492263e+00 -1.18864727e+00 5.99590003e-01 -4.67784345e-01 6.26741946e-02 1.51771214e-02 -6.75060332e-01 -8.16744149e-01 -1.17389418e-01 8.80320966e-02 2.74346143e-01 6.67129457e-01 -1.33691061e+00 -9.30631161e-01 -3.31458509e-01 -1.45801753e-01 2.77352661e-01 -6.08680964e-01 6.41109645e-01 -5.98668516e-01 -6.31813347e-01 -3.10437214e-02 -7.09387481e-01 -1.08261809e-01 -5.16093910e-01 -5.38397312e-01 -8.88025403e-01 1.18507004e+00 -4.91944730e-01 1.48802996e+00 -2.20271683e+00 -3.56954262e-02 2.52972841e-01 3.29701245e-01 1.97114140e-01 -2.71998509e-03 2.52340466e-01 -3.39848340e-01 5.64720891e-02 -1.82921991e-01 -2.13634998e-01 4.90465984e-02 2.78866529e-01 -6.18794084e-01 -2.62500905e-02 5.40715575e-01 6.71890438e-01 -1.05598569e+00 -8.27703357e-01 2.74536759e-02 4.50092018e-01 -1.10201210e-01 4.73785371e-01 -2.10342154e-01 1.14206024e-01 -7.29867995e-01 5.23369133e-01 6.19065523e-01 -1.37536814e-02 1.77559406e-01 -3.88467759e-01 -7.56971985e-02 3.57122332e-01 -1.38528478e+00 9.92444396e-01 -4.95418638e-01 2.62798905e-01 8.37348029e-02 -7.11148322e-01 1.11428809e+00 6.79673195e-01 7.88241565e-01 -4.31969792e-01 3.67187262e-01 3.87245923e-01 -5.74133918e-02 -8.35326314e-01 3.75551015e-01 -4.83231246e-01 -6.91528857e-01 4.80895907e-01 -1.10953905e-01 -1.89915478e-01 2.61244059e-01 1.86182961e-01 1.11893451e+00 7.45289922e-02 5.54672420e-01 3.41753066e-02 7.48848736e-01 -2.16119867e-02 1.12710857e+00 -2.58753914e-03 -2.81647146e-01 5.86749129e-02 9.96184051e-01 -3.46821606e-01 -4.31936055e-01 -6.41770840e-01 8.12904239e-02 8.28488588e-01 2.32759625e-01 -6.16626978e-01 -5.17936110e-01 -1.03513050e+00 -2.51887321e-01 6.76570892e-01 -4.86048877e-01 -9.34223086e-02 -5.29659927e-01 -1.05197513e+00 4.80369270e-01 3.21778744e-01 5.67150891e-01 -1.25838864e+00 -6.59093440e-01 2.59331733e-01 -8.23345423e-01 -1.33500576e+00 -1.07886337e-01 3.94834816e-01 -2.72831589e-01 -1.15815818e+00 -3.12892824e-01 -7.83573687e-01 4.02180195e-01 -4.26339768e-02 9.93003070e-01 -1.33386478e-01 -1.45707279e-01 5.43810800e-02 -6.05068266e-01 -6.50381386e-01 -2.36609280e-01 -2.05645640e-03 -9.76106301e-02 3.14575434e-01 7.53798485e-01 -4.32715446e-01 -1.41488165e-01 3.05956095e-01 -8.04261804e-01 5.94587289e-02 7.24552214e-01 6.04912281e-01 6.20910406e-01 6.04354560e-01 1.03508472e+00 -9.38205659e-01 1.02768934e+00 -4.24978614e-01 -2.39032358e-01 4.01397288e-01 -3.98707151e-01 -2.90307730e-01 7.36698091e-01 -2.36010432e-01 -1.32531440e+00 2.67858624e-01 -1.58900395e-01 -1.86284423e-01 -3.95521313e-01 7.30093479e-01 -6.88382685e-01 4.42018151e-01 4.84644845e-02 6.48175254e-02 -3.76736462e-01 -2.71467090e-01 2.51918882e-01 7.62912333e-01 5.51824272e-01 -5.41204631e-01 5.57118416e-01 6.52318373e-02 2.36423071e-02 -4.42925215e-01 -1.01635456e+00 -6.23105288e-01 -4.74012762e-01 -3.96162689e-01 9.11707640e-01 -8.33612084e-01 -9.14514184e-01 4.39980298e-01 -1.41045892e+00 1.07412077e-01 2.20677391e-01 4.86898214e-01 -2.15301558e-01 2.21475020e-01 -5.29812694e-01 -1.03994799e+00 -5.32075465e-01 -8.55087698e-01 1.06010377e+00 1.77840680e-01 -6.33692086e-01 -7.60361314e-01 -1.88420922e-01 1.02237739e-01 -1.04919963e-01 4.48678374e-01 1.22696829e+00 -7.60868132e-01 4.59854901e-02 -2.24148825e-01 -4.44900781e-01 4.03389744e-02 4.43913788e-01 1.62583247e-01 -9.17819440e-01 3.14006239e-01 5.50383031e-02 -5.62979877e-01 5.93769610e-01 4.98091578e-02 1.00954175e+00 -3.38894665e-01 -2.80344427e-01 7.76234269e-02 1.32041812e+00 3.28172952e-01 3.80811960e-01 1.53022781e-01 7.14423060e-01 9.99346614e-01 1.10903788e+00 6.89548075e-01 4.80174869e-01 5.53093791e-01 4.50799137e-01 -4.48726565e-01 2.65603274e-01 -3.36440392e-02 2.91763306e-01 9.93671834e-01 7.69712701e-02 -3.20752114e-01 -7.87191570e-01 6.95265234e-01 -2.00002789e+00 -7.09846556e-01 -6.31615281e-01 1.61515236e+00 1.01436853e+00 -3.72546390e-02 8.38691667e-02 4.96921152e-01 7.16118157e-01 7.45883733e-02 -3.45321745e-01 -4.95839179e-01 -1.30841017e-01 -5.45235071e-03 -1.66003048e-01 4.40464392e-02 -1.41680467e+00 9.16716278e-01 5.15358210e+00 1.01483250e+00 -1.02967668e+00 9.00570825e-02 7.35297859e-01 2.16329336e-01 -2.71770597e-01 -1.04563639e-01 -6.57395482e-01 3.34363669e-01 3.06024909e-01 -8.79570916e-02 4.74231392e-02 7.71491826e-01 3.06717157e-01 -7.54466653e-02 -8.64598036e-01 9.82468128e-01 3.74157317e-02 -4.55361396e-01 -1.17528029e-01 -2.69188821e-01 5.54498255e-01 -6.02077663e-01 -2.49438539e-01 3.30565542e-01 -9.36996099e-03 -6.93376243e-01 5.90796828e-01 1.72294930e-01 5.11125088e-01 -1.20557642e+00 1.08641577e+00 1.23025700e-01 -1.29345024e+00 -4.31366786e-02 -9.95215699e-02 -2.16065228e-01 3.79515499e-01 1.17443323e+00 -6.04918122e-01 6.17231309e-01 6.73236847e-01 6.16637588e-01 -2.64834017e-01 8.44951093e-01 -8.21717799e-01 7.50219166e-01 -2.82230258e-01 -4.30871069e-01 1.50575131e-01 1.33110344e-01 4.62315112e-01 1.37394106e+00 2.27982387e-01 3.47744137e-01 2.23329395e-01 8.60811353e-01 -6.27978668e-02 6.04205787e-01 -1.48750618e-01 -8.45234469e-03 4.52250183e-01 1.73539293e+00 -9.17992890e-01 -2.82003671e-01 -2.60545820e-01 9.63801563e-01 1.89916149e-01 9.03212279e-02 -1.06877589e+00 -8.12491119e-01 4.59737867e-01 -6.25935018e-01 1.50294840e-01 1.67888403e-01 -4.23535138e-01 -9.74151015e-01 2.07233310e-01 -9.29403484e-01 4.33443666e-01 -6.77650869e-01 -1.52205086e+00 7.50971437e-01 -1.18499272e-01 -1.15988505e+00 -2.68658727e-01 -3.45966637e-01 -9.57936823e-01 5.51967800e-01 -1.48710263e+00 -1.18047702e+00 -2.84433812e-01 6.32469475e-01 3.19422930e-01 3.88751984e-01 8.92516494e-01 4.88415390e-01 -8.61609817e-01 5.16027153e-01 -5.65672636e-01 1.71120122e-01 8.65250707e-01 -1.21550965e+00 -1.10265538e-01 8.44519138e-01 -1.72674879e-01 5.28176606e-01 8.33781779e-01 -8.32629323e-01 -1.20894837e+00 -9.95527923e-01 1.61593437e+00 -1.38981417e-01 7.44526446e-01 -4.20895398e-01 -9.13966298e-01 1.56363741e-01 1.88265204e-01 -1.22482836e-01 8.29345882e-01 5.03229558e-01 -4.56419498e-01 -1.77397206e-01 -9.51085210e-01 5.84990621e-01 6.43496037e-01 -2.46868774e-01 -6.98135376e-01 2.08352774e-01 6.60704374e-01 -2.48027537e-02 -8.14946413e-01 6.85804069e-01 5.16689599e-01 -8.60551775e-01 5.12861133e-01 -2.62096167e-01 8.99623036e-01 -2.41745368e-01 1.50557950e-01 -1.29209423e+00 7.47247338e-02 -6.71567500e-01 3.92334089e-02 2.08002830e+00 4.65262890e-01 -4.07101989e-01 2.71077454e-01 8.84252906e-01 4.64205258e-02 -9.93515611e-01 -8.24264765e-01 -5.36082923e-01 -3.38827729e-01 -7.80566692e-01 9.60524261e-01 1.11901581e+00 4.24521297e-01 7.25873530e-01 -5.17988324e-01 5.23280837e-02 3.66226435e-01 5.31385839e-01 5.81269741e-01 -1.08772981e+00 -1.17680006e-01 -4.62910414e-01 4.76248600e-02 -5.93849123e-01 4.07951832e-01 -4.53632057e-01 4.95330900e-01 -1.49326980e+00 2.74510980e-01 -5.27696550e-01 -3.91457975e-01 8.88042510e-01 -8.50464046e-01 -4.36368361e-02 1.62211061e-02 9.92525667e-02 -8.41184258e-01 6.98642910e-01 8.88223290e-01 2.07962275e-01 -2.86379606e-01 -1.50499828e-02 -6.64520025e-01 9.76682723e-01 7.79514551e-01 -5.77059627e-01 -3.05834472e-01 9.02105570e-02 5.94245374e-01 -1.16118453e-01 3.03778619e-01 -5.92014253e-01 -3.50191109e-02 -2.54935145e-01 1.01175502e-01 -6.78125501e-01 2.35597696e-02 -9.38993752e-01 -3.32062662e-01 1.80007383e-01 -2.07195163e-01 -1.44861033e-02 2.21469507e-01 2.05981046e-01 -5.30338466e-01 -1.65800154e-01 3.12991619e-01 2.23916188e-01 -9.74687278e-01 -1.03208438e-01 -4.56648231e-01 1.73418708e-02 1.01406360e+00 3.08239281e-01 -1.08294994e-01 -3.86014968e-01 -2.68076748e-01 3.58559310e-01 -1.42265588e-01 5.32140374e-01 5.68095565e-01 -1.34581506e+00 -6.40332580e-01 -1.64099008e-01 3.18887472e-01 -1.96470946e-01 1.79396734e-01 8.56446028e-01 3.89547765e-01 8.21217149e-02 2.08136499e-01 -1.07579522e-01 -1.66436481e+00 6.23578548e-01 -1.21887423e-01 -5.99250615e-01 -3.13214540e-01 7.86912203e-01 1.36891991e-01 -3.04684222e-01 2.46979967e-01 -7.92322457e-02 -7.51442432e-01 3.60989630e-01 4.05945987e-01 1.62284568e-01 -8.05041641e-02 -6.81384504e-01 -6.41757369e-01 5.80238163e-01 -9.21457354e-03 2.18552686e-02 1.29982436e+00 -1.11448899e-01 -6.86192513e-01 5.27850807e-01 1.15460467e+00 -3.71196941e-02 -6.13398433e-01 -1.06052212e-01 2.69285321e-01 -2.19580278e-01 1.49021506e-01 -9.85875010e-01 -9.77460027e-01 7.06245124e-01 1.57380372e-01 2.24678531e-01 1.60809779e+00 1.20281413e-01 1.29879260e+00 1.40503839e-01 9.41626579e-02 -1.12212062e+00 -7.42913634e-02 6.00134194e-01 8.62577796e-01 -1.07190716e+00 -7.87656382e-02 -9.12740111e-01 -6.80369139e-01 1.05590785e+00 6.90947235e-01 2.12853387e-01 5.36962450e-01 4.49119568e-01 3.61428916e-01 -3.95381153e-01 -7.97551632e-01 -3.94763917e-01 3.77763629e-01 1.14929900e-01 6.56267762e-01 4.86261845e-02 -7.77041376e-01 1.17965794e+00 -4.68066707e-02 6.87846765e-02 6.53542578e-02 7.63826728e-01 -1.19268551e-01 -1.12132215e+00 -4.34136212e-01 3.57145190e-01 -7.89384484e-01 -8.02504644e-02 -9.91825283e-01 5.72454035e-01 6.09085023e-01 1.35600877e+00 -3.14576000e-01 -6.03492320e-01 3.09943289e-01 1.11165807e-01 -3.33654359e-02 -2.99825191e-01 -8.63125145e-01 4.76992130e-01 3.50457847e-01 -5.20075798e-01 -7.23680317e-01 -5.77958643e-01 -1.73465776e+00 2.42693096e-01 -5.08587658e-01 4.75473672e-01 5.58771014e-01 1.31106400e+00 3.02346945e-01 8.28136683e-01 8.66651416e-01 -3.42192948e-01 -1.08432554e-01 -7.32521832e-01 -4.86060321e-01 7.41587877e-01 6.30910620e-02 -6.66578174e-01 -3.80898029e-01 2.86304932e-02]
[12.628816604614258, 6.207989692687988]
957efe2e-d97d-4976-8eef-4fdb862d62b3
a-comprehensive-and-large-scale-dataset-for
null
null
https://openreview.net/forum?id=TnX3iwX_6Iu
https://openreview.net/pdf?id=TnX3iwX_6Iu
A Comprehensive and Large-Scale Dataset for Integrated Argument Mining Tasks
Traditionally, a debate usually requires a manual preparation process, including reading plenty of articles, selecting the claims, identifying the stances of the claims, seeking the evidences for the claims, etc. As the AI debate attracts more attention these years, it is worth exploring the methods to automate the tedious process involved in the debating system. In this work, we introduce a comprehensive and large dataset, which can be applied to a series of argument mining tasks, including claim extraction, stance classification, evidence extraction, etc. Our dataset is collected from over 1k articles related to 123 topics. Near 70k sentences in the dataset are fully annotated based on their argument properties (e.g., claims, stances, evidences, etc.). We further propose two new integrated argument mining tasks associated with the debate preparation process: (1) claim extraction with stance classification (CESC) and (2) claim-evidence pair extraction (CEPE). We adopt a pipeline approach and an end-to-end method for each integrated task separately. Promising experimental results are reported to show the values and challenges of our proposed tasks, and motivate future research on argument mining.
['Anonymous']
2021-11-16
null
null
null
acl-arr-november-2021-11
['claim-extraction-with-stance-classification', 'claim-evidence-pair-extraction-cepe']
['natural-language-processing', 'natural-language-processing']
[ 5.05443633e-01 3.65969151e-01 -5.91060638e-01 -3.66934776e-01 -1.36442947e+00 -7.14356720e-01 9.38210428e-01 6.86215162e-01 -5.13874650e-01 8.77410114e-01 5.42267561e-01 -7.90878832e-01 -6.67010024e-02 -5.95794320e-01 -6.98320985e-01 -3.37498218e-01 4.75943387e-01 7.11325049e-01 4.17576343e-01 -1.77778482e-01 9.37008321e-01 -7.20435455e-02 -1.34682238e+00 7.18482971e-01 1.02651596e+00 1.23787284e+00 -2.12868243e-01 4.34921592e-01 -4.53314513e-01 1.27333510e+00 -8.24121654e-01 -1.15463817e+00 -2.11987957e-01 -4.24153656e-01 -1.38270223e+00 -1.42022297e-01 8.97168666e-02 -1.00227594e-01 3.06890875e-01 1.11687672e+00 4.08694565e-01 -5.62277496e-01 5.08472621e-01 -1.08943439e+00 -4.02090013e-01 1.37121809e+00 -7.56229281e-01 3.31899345e-01 3.16501111e-01 -4.55385238e-01 1.45300555e+00 -8.99874687e-01 8.08881164e-01 1.03691685e+00 4.53528136e-01 -8.84263311e-03 -5.20775259e-01 -5.12014329e-01 5.49270064e-02 5.28810859e-01 -5.27983248e-01 -4.60255980e-01 1.28928781e+00 -5.05973279e-01 4.22875583e-01 2.42493451e-01 6.95892274e-01 1.05703974e+00 -3.74923833e-02 1.36554277e+00 1.31428182e+00 -5.56112349e-01 9.74684879e-02 4.78280783e-02 8.51068795e-01 3.07366401e-01 5.53254962e-01 -6.58830345e-01 -5.48122108e-01 -6.97153091e-01 -5.25493734e-02 -3.68291348e-01 -4.41311002e-02 2.39872068e-01 -1.34858918e+00 1.12517166e+00 -1.94744635e-02 2.47630417e-01 -6.56725585e-01 -2.19350576e-01 9.78560984e-01 2.82154471e-01 8.09726119e-01 2.27791980e-01 -7.40092099e-01 -1.92956924e-01 -1.06585097e+00 6.76721215e-01 1.03577232e+00 7.17837036e-01 2.79523939e-01 -8.43746185e-01 -3.62803191e-01 8.73322785e-01 3.28823596e-01 3.40328604e-01 1.42275319e-01 -7.92935967e-01 1.13046134e+00 9.34843302e-01 2.13385776e-01 -1.02113247e+00 -3.27615798e-01 -1.74866498e-01 -4.58575815e-01 -2.94570088e-01 3.11932981e-01 -3.70380461e-01 -3.22358221e-01 1.19583905e+00 7.61572301e-01 -2.97278047e-01 2.37117931e-01 6.52249336e-01 1.11486316e+00 3.89971614e-01 7.42814541e-02 -5.69607675e-01 2.17049909e+00 -9.59182084e-01 -1.11321151e+00 -1.87242087e-02 5.77003062e-01 -1.34237850e+00 8.88776481e-01 2.33019158e-01 -1.27518082e+00 5.66119216e-02 -9.82093811e-01 -2.87516654e-01 -1.44527853e-01 3.34388226e-01 6.32010996e-01 2.41551980e-01 7.03643188e-02 3.65355492e-01 -3.03926021e-01 2.80464858e-01 8.84765089e-01 -9.08137634e-02 9.56854410e-03 2.40718141e-01 -1.54977286e+00 7.92600453e-01 4.82621402e-01 1.52927935e-01 -1.35733321e-01 -3.60156059e-01 -7.35198617e-01 -1.69724911e-01 1.01614583e+00 -4.62238431e-01 1.32633281e+00 -4.62445408e-01 -1.16117990e+00 1.17986798e+00 -3.93732600e-02 -7.12997794e-01 4.95487720e-01 -6.18624389e-01 -3.91422004e-01 3.00230294e-01 7.13490546e-01 -1.00208491e-01 6.43599808e-01 -9.23961878e-01 -7.60174751e-01 -2.98707008e-01 2.48757824e-01 8.31195936e-02 1.37785673e-01 9.78096843e-01 -1.97312459e-01 -8.73246312e-01 1.70132980e-01 -9.50099289e-01 1.70251742e-01 -3.82110476e-01 -9.30673182e-01 -8.97376895e-01 9.30521250e-01 -9.29076612e-01 1.32719624e+00 -1.79920805e+00 -9.71779674e-02 1.85792167e-02 2.41986349e-01 1.10913165e-01 3.92473370e-01 3.99339408e-01 9.31854621e-02 2.22714245e-01 -4.55672592e-01 2.55915727e-02 1.39361829e-01 1.70908291e-02 -5.17355204e-01 3.12395751e-01 2.69612879e-01 1.12295783e+00 -7.47467935e-01 -1.16063952e+00 -5.16386092e-01 3.01609542e-02 -1.74147680e-01 -3.55757549e-02 -3.79963428e-01 1.81767747e-01 -9.26059723e-01 8.39130282e-01 7.05337942e-01 -5.53783774e-01 3.02112937e-01 -5.29602766e-01 -8.17959607e-02 1.21881056e+00 -1.02723324e+00 1.28309739e+00 -2.07948126e-02 5.07486761e-01 2.96712756e-01 -1.10299861e+00 6.78589880e-01 4.42969680e-01 4.27837819e-01 -4.33731467e-01 5.43050110e-01 7.03840375e-01 1.05490014e-01 -6.06461823e-01 4.79112953e-01 5.47649525e-02 -4.71023530e-01 1.15386403e+00 -5.07584929e-01 1.22158332e-02 6.60643399e-01 5.50780892e-01 8.68335605e-01 -1.73153564e-01 4.71896142e-01 -2.05344006e-01 8.73825490e-01 5.72895944e-01 6.26567721e-01 2.79787630e-01 -4.82406393e-02 1.57017261e-01 9.77860510e-01 -5.74738741e-01 -9.93646860e-01 -2.41260439e-01 -2.20597684e-01 8.07935596e-01 1.15300700e-01 -5.36489069e-01 -6.90934777e-01 -1.31326568e+00 -1.30499646e-01 4.42797750e-01 -7.80864537e-01 4.12852347e-01 -1.11684477e+00 -9.19928253e-01 6.13938868e-01 4.06820476e-01 8.33834946e-01 -1.29838204e+00 -8.71172011e-01 3.79553169e-01 -9.28906381e-01 -1.22798467e+00 -2.71943480e-01 9.94653180e-02 -4.47998106e-01 -1.74127805e+00 -3.91803414e-01 -5.73569000e-01 2.92073727e-01 1.88875452e-01 1.16344786e+00 4.87548143e-01 1.74437985e-01 -3.40197951e-01 -5.69593370e-01 -8.18050921e-01 -4.21515793e-01 1.14415102e-01 -3.88523936e-01 -4.04577926e-02 2.58263767e-01 -1.02195153e-02 -4.73191619e-01 8.63634273e-02 -8.48462403e-01 2.22278565e-01 6.95158362e-01 9.45080996e-01 6.78233266e-01 -1.32798746e-01 5.90379179e-01 -1.51095533e+00 1.27298307e+00 -5.53411126e-01 -2.64485747e-01 5.08188426e-01 -5.30989110e-01 -7.54464120e-02 1.35667890e-01 -3.96968991e-01 -1.18030119e+00 -5.87506354e-01 -3.27603072e-01 4.70819801e-01 2.78582960e-01 9.85511959e-01 -1.83520123e-01 6.50387466e-01 3.34958643e-01 -3.16846251e-01 -5.16380593e-02 -5.63536584e-01 4.27855015e-01 1.01697338e+00 3.45442981e-01 -8.56937170e-01 6.73255920e-01 4.86761898e-01 -3.24121505e-01 -2.37813681e-01 -1.81272757e+00 -3.35788101e-01 -5.18041253e-01 1.37003576e-02 6.73259735e-01 -6.93054557e-01 -6.28419518e-01 3.26124310e-01 -1.51385343e+00 1.59375221e-01 -1.95135877e-01 5.03520429e-01 -3.00658137e-01 5.44995248e-01 -9.28381205e-01 -7.44108379e-01 -9.17323947e-01 -1.08700514e+00 8.69452715e-01 9.71973166e-02 -4.96092737e-01 -5.82898557e-01 1.26139939e-01 1.18967927e+00 -5.86732142e-02 2.48634875e-01 1.07485449e+00 -1.02364659e+00 -5.77521883e-02 -2.43188575e-01 -5.10449767e-01 1.62739828e-01 -1.23075210e-01 -2.40623001e-02 -5.89321256e-01 2.84171551e-01 2.75112838e-01 -6.12745225e-01 8.47067177e-01 3.40766698e-01 8.86240363e-01 -6.97662473e-01 -3.95644069e-01 -2.76221454e-01 6.62494719e-01 -1.90185942e-02 5.85607111e-01 9.30320919e-01 3.96507591e-01 9.45787191e-01 1.16665816e+00 2.56215394e-01 6.15684390e-01 6.65841699e-01 2.12772772e-01 -3.18255983e-02 -1.28140226e-01 9.06242058e-02 2.49737471e-01 9.81938958e-01 -2.68127918e-01 -1.43513769e-01 -8.03024411e-01 6.24727309e-01 -2.07273507e+00 -1.13036609e+00 -7.47878432e-01 1.40371597e+00 1.20110571e+00 7.23418236e-01 1.95696458e-01 7.08125174e-01 8.09639096e-01 2.29951173e-01 -3.96778375e-01 -3.17565948e-01 -4.28344339e-01 2.10587814e-01 7.44906515e-02 1.87673181e-01 -1.31689990e+00 8.01227212e-01 5.37289429e+00 8.83144915e-01 -6.26807511e-01 2.01157704e-01 7.61849701e-01 2.23208681e-01 -3.71421039e-01 4.84433025e-01 -7.37917721e-01 6.29770041e-01 6.32704973e-01 -3.10877934e-02 -3.71742576e-01 7.97634125e-01 -2.67658965e-03 -2.19891027e-01 -4.45471942e-01 4.88588303e-01 1.28357962e-01 -1.83055937e+00 -8.89979377e-02 5.21525852e-02 5.21692514e-01 -1.13774434e-01 -3.52732778e-01 1.33798122e-01 2.36654431e-01 -4.38898146e-01 1.16806841e+00 -4.62024007e-03 4.15822715e-01 -5.29657245e-01 1.17369699e+00 4.22802955e-01 -8.85688126e-01 -7.50170741e-03 -7.95287415e-02 -3.31498347e-02 6.56176507e-01 1.19725084e+00 -4.80507851e-01 7.66813457e-01 5.15508533e-01 6.42188013e-01 -2.42688209e-01 6.69317067e-01 -8.83627474e-01 1.01778364e+00 -1.56275123e-01 -3.76790315e-01 3.02975088e-01 5.38363063e-04 4.10193712e-01 9.59466159e-01 -6.25116900e-02 4.21613246e-01 9.05874446e-02 6.35847867e-01 -6.11441493e-01 5.20031631e-01 9.21001732e-02 -1.99107140e-01 5.51362753e-01 1.51227462e+00 -1.11635816e+00 -8.20840180e-01 -1.74782217e-01 3.60224694e-01 2.35304222e-01 -2.42337778e-01 -9.63745654e-01 -3.80095899e-01 7.84358159e-02 1.13900691e-01 2.22434178e-01 1.04606085e-01 -3.90597016e-01 -1.21290541e+00 4.58794981e-01 -1.18060791e+00 6.54295683e-01 -4.74859416e-01 -1.40689290e+00 5.40074170e-01 -3.23316120e-02 -1.12394714e+00 -2.01869495e-02 -3.27814609e-01 -7.72319436e-01 6.48023248e-01 -1.77659607e+00 -1.28760493e+00 9.54101458e-02 2.67615706e-01 5.90097845e-01 1.66617900e-01 2.94791907e-01 3.51884961e-01 -5.22249222e-01 8.56935680e-02 -5.22188008e-01 4.94119763e-01 6.61797345e-01 -9.07184362e-01 3.82267952e-01 6.17852867e-01 2.79936373e-01 5.33629358e-01 5.92527151e-01 -9.22498405e-01 -1.03836489e+00 -5.29236615e-01 1.55992448e+00 -5.02377212e-01 1.04413664e+00 -5.65485172e-02 -9.05047596e-01 2.57110804e-01 4.63505745e-01 -4.33063686e-01 8.28436971e-01 3.71652186e-01 -4.14740890e-01 3.81407171e-01 -9.13592994e-01 2.89498329e-01 7.64572859e-01 -3.97563636e-01 -1.19782448e+00 5.08934915e-01 6.17959976e-01 -4.60416287e-01 -6.63745821e-01 5.30580819e-01 5.37506342e-01 -4.62554336e-01 7.48029649e-01 -7.47933447e-01 8.85968387e-01 -3.33995640e-01 6.49165586e-02 -5.86577475e-01 5.36707640e-02 -4.48914409e-01 -4.52454835e-01 1.41705298e+00 8.36180329e-01 -3.09407860e-01 5.71998358e-01 2.53193229e-01 -2.03326214e-02 -1.17855656e+00 -7.94866443e-01 -1.02728888e-01 -5.41013069e-02 -5.14021575e-01 7.49980867e-01 1.11557019e+00 3.02357018e-01 1.00838399e+00 -2.67614931e-01 -1.95691898e-01 5.21627247e-01 1.07196093e+00 6.40556335e-01 -1.71487963e+00 -6.03662059e-02 -4.76696670e-01 3.23042810e-01 -9.83663917e-01 1.25705495e-01 -7.18162775e-01 -1.15716368e-01 -1.87480342e+00 6.35992944e-01 -6.47342980e-01 -2.15470437e-02 3.34276021e-01 -5.65939844e-01 1.11522013e-02 -5.59615530e-02 7.53814578e-01 -7.55681872e-01 2.59309083e-01 1.19865072e+00 -4.00369018e-01 7.29132816e-02 2.09005266e-01 -1.20851851e+00 1.11320519e+00 8.39634836e-01 -7.15392530e-01 -3.63117047e-02 -3.31924796e-01 7.74199188e-01 -1.48745552e-01 1.77926481e-01 -2.70515412e-01 3.27719539e-01 -2.08480418e-01 -5.08736894e-02 -1.35063565e+00 6.78618476e-02 -2.87931263e-01 -4.14435446e-01 4.74575132e-01 -3.68430078e-01 1.48902223e-01 -2.18177766e-01 4.80479866e-01 -3.78217280e-01 -4.99784797e-01 3.31372768e-01 -1.02545079e-02 -2.21308786e-02 -1.59417123e-01 -6.91648051e-02 5.22669494e-01 8.90443742e-01 2.37069130e-01 -8.32192004e-01 -3.55820321e-02 -4.83256280e-01 1.83076277e-01 -2.47255452e-02 2.25456625e-01 4.85810220e-01 -1.15697992e+00 -1.32420719e+00 -4.45030063e-01 5.58512434e-02 2.27750931e-03 -8.76935497e-02 1.32285810e+00 -2.63470531e-01 1.36731803e-01 2.21810028e-01 -1.48993775e-01 -1.66351497e+00 3.92286777e-01 -5.98012269e-01 -9.92863476e-01 -6.45535886e-01 7.10192740e-01 -4.51078087e-01 -3.68296430e-02 -1.51984483e-01 -2.66133219e-01 -7.88801014e-01 5.99257290e-01 6.00179315e-01 2.71973789e-01 1.06675275e-01 -6.71933711e-01 -4.39331263e-01 3.55429471e-01 -4.90666747e-01 -1.58440650e-01 1.56203651e+00 9.79167446e-02 -6.31038785e-01 2.50444859e-01 5.77873588e-01 5.88869691e-01 -6.40539110e-01 -3.71291339e-01 5.91483176e-01 -5.40048145e-02 -1.58124059e-01 -8.29300940e-01 -7.47700632e-01 6.81365609e-01 -4.45137501e-01 6.65630996e-01 6.93295181e-01 5.37122369e-01 1.20083880e+00 2.20029294e-01 1.16203323e-01 -1.45445418e+00 -6.61022142e-02 6.00205958e-01 9.88954902e-01 -1.23078609e+00 4.93155509e-01 -8.37795198e-01 -8.06123614e-01 1.01788902e+00 -7.53226429e-02 8.92148986e-02 6.87153637e-01 3.38411719e-01 3.05772182e-02 -9.15758014e-01 -7.75021970e-01 -3.50357704e-02 3.97133470e-01 -3.09651196e-01 5.72455466e-01 6.39767526e-03 -1.26991594e+00 1.03598297e+00 -2.45796487e-01 -3.43530148e-01 3.42351705e-01 1.27078581e+00 -5.51592946e-01 -1.32399786e+00 -4.22778070e-01 5.46828866e-01 -1.00975382e+00 -3.05905640e-01 -1.03491259e+00 6.50768518e-01 2.50100881e-01 1.16947496e+00 -4.24144477e-01 1.03556857e-01 3.53232116e-01 4.46081236e-02 1.91426039e-01 -4.25263971e-01 -8.51216435e-01 1.34559214e-01 9.49512839e-01 -6.86738640e-02 -1.15567446e+00 -7.89157510e-01 -1.36153305e+00 -2.22259805e-01 -9.20444489e-01 7.42967248e-01 7.13169694e-01 1.38246739e+00 2.65401661e-01 3.65032643e-01 3.25093150e-01 -1.73600405e-01 -4.24283653e-01 -1.19178522e+00 -1.05992414e-01 5.57874382e-01 -1.60180628e-01 -9.24068153e-01 -1.32691368e-01 2.14946851e-01]
[9.451274871826172, 9.544551849365234]
a335c7d1-00b9-49e9-a373-66599c762484
genesis-v2-inferring-unordered-object
2104.09958
null
https://arxiv.org/abs/2104.09958v3
https://arxiv.org/pdf/2104.09958v3.pdf
GENESIS-V2: Inferring Unordered Object Representations without Iterative Refinement
Advances in unsupervised learning of object-representations have culminated in the development of a broad range of methods for unsupervised object segmentation and interpretable object-centric scene generation. These methods, however, are limited to simulated and real-world datasets with limited visual complexity. Moreover, object representations are often inferred using RNNs which do not scale well to large images or iterative refinement which avoids imposing an unnatural ordering on objects in an image but requires the a priori initialisation of a fixed number of object representations. In contrast to established paradigms, this work proposes an embedding-based approach in which embeddings of pixels are clustered in a differentiable fashion using a stochastic stick-breaking process. Similar to iterative refinement, this clustering procedure also leads to randomly ordered object representations, but without the need of initialising a fixed number of clusters a priori. This is used to develop a new model, GENESIS-v2, which can infer a variable number of object representations without using RNNs or iterative refinement. We show that GENESIS-v2 performs strongly in comparison to recent baselines in terms of unsupervised image segmentation and object-centric scene generation on established synthetic datasets as well as more complex real-world datasets.
['Ingmar Posner', 'Oiwi Parker Jones', 'Martin Engelcke']
2021-04-20
null
http://proceedings.neurips.cc/paper/2021/hash/43ec517d68b6edd3015b3edc9a11367b-Abstract.html
http://proceedings.neurips.cc/paper/2021/file/43ec517d68b6edd3015b3edc9a11367b-Paper.pdf
neurips-2021-12
['scene-generation', 'unsupervised-object-segmentation']
['computer-vision', 'computer-vision']
[ 6.29321158e-01 4.57573593e-01 2.65978545e-01 -4.20058459e-01 -3.49185616e-01 -6.11346722e-01 1.07790971e+00 3.65599155e-01 -5.18367946e-01 3.74051988e-01 -2.01794673e-02 -3.96465957e-02 -2.18150333e-01 -9.56422806e-01 -7.73744404e-01 -6.49246275e-01 9.68160033e-02 9.22148049e-01 4.37307209e-01 8.76498297e-02 1.90934747e-01 5.72570741e-01 -1.95375109e+00 -1.34615246e-02 7.35723913e-01 4.93159175e-01 3.37205827e-01 7.45422184e-01 -1.41070485e-01 5.06803453e-01 -4.53738272e-01 -1.81372613e-01 3.00489306e-01 -5.50716162e-01 -9.81122553e-01 8.82066250e-01 3.31704617e-01 -5.89515897e-04 1.11946397e-01 1.01405919e+00 2.32548073e-01 3.93013716e-01 1.03538287e+00 -1.10529959e+00 -7.47672200e-01 4.62847620e-01 -3.47001195e-01 -2.49227688e-01 -5.75410686e-02 2.96169877e-01 1.11669493e+00 -8.31615031e-01 8.49741340e-01 1.19214416e+00 4.31756794e-01 5.18276095e-01 -1.94175828e+00 1.85504220e-02 2.32396536e-02 -1.01742394e-01 -1.32514882e+00 -2.80748218e-01 9.28791344e-01 -6.95151806e-01 6.21621907e-01 2.36748338e-01 7.96771049e-01 8.76162410e-01 -4.43499178e-01 7.67143786e-01 1.04322720e+00 -6.16702616e-01 4.90350485e-01 4.08719510e-01 5.12885898e-02 4.48709011e-01 4.17203277e-01 4.77852970e-02 -1.11645851e-02 1.71565324e-01 1.00179696e+00 8.05878267e-02 -6.85046464e-02 -1.04446256e+00 -1.47730231e+00 1.07200384e+00 7.17565894e-01 2.78440803e-01 -4.89655435e-01 2.29779571e-01 2.36100823e-01 -9.25813541e-02 3.37232977e-01 8.53648961e-01 -2.67651320e-01 2.87851661e-01 -1.21840072e+00 2.51073867e-01 5.29738963e-01 9.42015409e-01 1.09260786e+00 2.13216394e-01 -1.19743831e-01 7.48359859e-01 4.55669045e-01 1.05755135e-01 3.83630216e-01 -1.19882357e+00 9.57160518e-02 9.51324821e-01 1.56608850e-01 -1.03577483e+00 -2.21755698e-01 -2.76568681e-01 -8.82990777e-01 6.16242111e-01 4.85232204e-01 4.76267040e-02 -1.37712216e+00 1.53354204e+00 3.82407725e-01 8.02285746e-02 2.04480752e-01 8.63841295e-01 4.63045806e-01 5.87433517e-01 -5.95530979e-02 8.41550529e-02 1.06645370e+00 -8.75808001e-01 -3.12128991e-01 -8.32408741e-02 3.13939899e-01 -4.43559587e-01 1.12029731e+00 3.40068251e-01 -9.60022867e-01 -7.10682273e-01 -8.53083074e-01 -1.13921344e-01 -6.65990949e-01 -8.07536021e-02 6.59792840e-01 7.33426988e-01 -1.13663995e+00 4.83173102e-01 -7.84287274e-01 -4.44303423e-01 7.96297550e-01 3.55614930e-01 -3.67103934e-01 1.83565095e-01 -5.91551721e-01 6.28657460e-01 8.78500640e-01 2.03102306e-01 -9.86231387e-01 -4.80020702e-01 -1.09196889e+00 -1.57742575e-01 2.84569621e-01 -8.97018731e-01 9.49931622e-01 -1.40353131e+00 -1.49838877e+00 8.49087417e-01 1.12774201e-01 -7.09197879e-01 5.90846896e-01 -8.57752115e-02 1.50131315e-01 2.43177801e-01 -2.99363527e-02 1.35799396e+00 1.07252395e+00 -1.79330313e+00 -3.23093683e-01 -5.46469167e-02 5.77760898e-02 2.36958236e-01 -1.48233056e-01 -2.86671489e-01 -4.68293488e-01 -6.36352777e-01 2.02986419e-01 -9.00838673e-01 -8.40993881e-01 6.70355856e-02 -5.70999265e-01 -1.31548747e-01 8.97488713e-01 2.37532258e-02 6.61998093e-01 -1.94411504e+00 5.29349566e-01 2.32506812e-01 4.29143727e-01 1.99061736e-01 -2.11454242e-01 3.49747837e-01 -8.86028484e-02 2.97177643e-01 -8.57068121e-01 -4.90867406e-01 1.45522624e-01 5.22485077e-01 -1.88363373e-01 4.07664388e-01 6.31639302e-01 9.72480297e-01 -1.08283901e+00 -6.28605843e-01 7.23087013e-01 5.10765135e-01 -8.52593541e-01 1.15161203e-01 -6.04008198e-01 4.99258846e-01 -2.26565287e-01 1.31711930e-01 4.97143000e-01 -4.36071068e-01 3.71805876e-02 9.31451246e-02 -9.03614387e-02 -2.07829140e-02 -1.46042216e+00 1.66593409e+00 -1.85905144e-01 7.64734924e-01 -3.77686113e-01 -1.24419606e+00 7.57838070e-01 1.58016309e-01 4.77193266e-01 -1.20816208e-01 1.71427190e-01 -4.98177633e-02 -3.12513262e-02 -3.03031385e-01 5.60597897e-01 -2.46634692e-01 2.56702211e-02 6.06661797e-01 2.43373394e-01 -6.90259993e-01 4.11087781e-01 3.47396135e-01 8.21835220e-01 2.42453799e-01 3.54129039e-02 -2.09712580e-01 3.72924268e-01 1.07032686e-01 3.57192278e-01 8.02656353e-01 1.44510359e-01 1.19879580e+00 2.92617112e-01 -3.44562948e-01 -1.21770501e+00 -1.27278054e+00 -2.08798617e-01 6.80794954e-01 3.18174183e-01 -2.36045033e-01 -9.29291904e-01 -6.76067173e-01 -1.65209562e-01 8.50662887e-01 -8.41279328e-01 6.67798221e-02 -3.60974818e-01 -6.20081723e-01 2.34639943e-01 3.97550344e-01 2.75403231e-01 -1.69402802e+00 -9.89397585e-01 2.60458648e-01 1.35896742e-01 -1.09870028e+00 -5.81845529e-02 3.24139744e-01 -9.85986590e-01 -1.09194350e+00 -7.02274740e-01 -8.68834436e-01 1.33716500e+00 1.21581405e-02 1.26425254e+00 5.67393266e-02 -6.00367248e-01 4.36640054e-01 -3.22154671e-01 -3.18888366e-01 -3.60974252e-01 8.76251981e-02 -6.48425519e-02 2.67854363e-01 9.16699916e-02 -4.79022652e-01 -6.05100513e-01 2.47792993e-02 -1.35328496e+00 2.78779805e-01 7.11963594e-01 6.76931798e-01 6.40525222e-01 9.24536809e-02 3.86346936e-01 -1.10953629e+00 3.86218935e-01 -3.11101586e-01 -7.59049296e-01 -5.84999099e-02 -4.54300493e-01 4.68089104e-01 6.54324114e-01 -5.15597224e-01 -8.41602147e-01 3.38726759e-01 1.11881778e-01 -4.36944723e-01 -6.16346657e-01 3.36370140e-01 -1.12385685e-02 2.65103519e-01 7.75347292e-01 4.21047240e-01 6.82850778e-02 -3.58800471e-01 9.89021540e-01 4.11544591e-01 5.64272821e-01 -4.16061640e-01 1.13608551e+00 5.68922102e-01 -1.21347092e-01 -1.04365134e+00 -5.05148649e-01 -4.50788438e-01 -1.15513158e+00 -3.79964299e-02 1.02606547e+00 -5.06893754e-01 -2.81170130e-01 3.79363626e-01 -1.04670453e+00 -4.94068772e-01 -8.87229204e-01 3.01881284e-01 -8.69960070e-01 4.64836687e-01 -1.69897243e-01 -8.07863832e-01 5.95347025e-02 -1.17600489e+00 1.14420533e+00 9.81446132e-02 -4.53111976e-01 -1.00171292e+00 -6.46526068e-02 2.44412944e-01 9.15436968e-02 5.20946503e-01 8.03093076e-01 -4.03883427e-01 -8.81427050e-01 -5.57335652e-02 -3.58283132e-01 4.82868671e-01 2.54045129e-01 2.34762356e-01 -7.53504992e-01 -1.11169219e-01 -1.51400551e-01 -3.82707775e-01 7.68565297e-01 3.24943006e-01 1.16586363e+00 -2.92747140e-01 -1.54145017e-01 5.15644073e-01 1.49690878e+00 -1.23788072e-02 7.42631614e-01 1.94133058e-01 7.92298257e-01 8.27082634e-01 1.63922653e-01 1.23148136e-01 2.10581362e-01 4.56932604e-01 5.27089238e-01 -2.93671548e-01 -1.51659504e-01 -2.92774439e-01 -4.58115265e-02 6.15723550e-01 -5.98999076e-02 -2.70913720e-01 -9.23624098e-01 1.02821088e+00 -1.73319030e+00 -8.05784464e-01 -2.23364055e-01 2.22137165e+00 7.86890566e-01 2.00775713e-01 2.88879067e-01 3.96864295e-01 5.70791781e-01 1.69437051e-01 -4.48711216e-01 -2.52124220e-01 2.60610562e-02 2.43956223e-01 3.48073125e-01 4.58964497e-01 -1.02780759e+00 1.18107104e+00 6.35501194e+00 4.56128508e-01 -9.40231264e-01 -1.54667273e-01 5.37625670e-01 2.96074510e-01 -4.13485020e-01 2.37116858e-01 -4.73520905e-01 3.08460504e-01 7.05372512e-01 6.82525933e-02 2.29679376e-01 6.97053254e-01 1.11489244e-01 -1.91074803e-01 -1.24609780e+00 8.22765768e-01 -4.67340723e-02 -1.53752971e+00 4.58157808e-01 1.30239561e-01 8.71115565e-01 -1.22934170e-01 3.91589515e-02 3.51051502e-02 6.76939607e-01 -1.10521042e+00 7.15949118e-01 4.74953115e-01 4.80528444e-01 -5.46535254e-01 2.80577779e-01 4.38886464e-01 -9.17486370e-01 1.21768899e-01 -3.06405783e-01 4.13802639e-02 1.06458537e-01 3.72182488e-01 -9.40143108e-01 3.45519125e-01 4.89155740e-01 6.44201636e-01 -6.43761337e-01 1.04074311e+00 -3.61774743e-01 6.97202682e-01 -4.64108318e-01 9.90652759e-03 4.21787441e-01 -2.91157603e-01 4.29679155e-01 1.13547766e+00 -1.64624274e-01 -1.68536291e-01 1.43117920e-01 1.23142231e+00 4.30268124e-02 -4.64052781e-02 -6.14053667e-01 -2.11827651e-01 1.79660574e-01 1.32189333e+00 -1.32955897e+00 -4.20217931e-01 -6.28955290e-02 1.13086283e+00 3.04746777e-01 4.64066476e-01 -5.74607611e-01 -4.11493242e-01 2.92176038e-01 3.77017051e-01 6.10040069e-01 -4.81797218e-01 -2.70221978e-01 -9.16154981e-01 -2.23149687e-01 -7.16796100e-01 -4.10943814e-02 -6.15315318e-01 -1.12553155e+00 6.33865952e-01 2.21150473e-01 -1.18996418e+00 -4.16170478e-01 -5.23754001e-01 -4.31222379e-01 5.13733625e-01 -1.35711443e+00 -1.06001461e+00 -2.27799162e-01 3.63014519e-01 7.81074822e-01 3.58609408e-02 7.82923818e-01 -1.93317771e-01 -3.26157629e-01 5.05637079e-02 7.74361044e-02 2.44568184e-01 2.37482414e-01 -1.63368773e+00 5.35253525e-01 8.47836196e-01 7.09489405e-01 5.87028027e-01 8.23766708e-01 -3.70138913e-01 -9.61674452e-01 -1.29704189e+00 6.13499701e-01 -7.03829944e-01 4.54277068e-01 -8.13981354e-01 -8.44938695e-01 5.81252992e-01 3.26128751e-01 2.56331444e-01 3.60220850e-01 -3.66029367e-02 -8.00861940e-02 1.46721587e-01 -9.22302902e-01 8.24042022e-01 8.40055346e-01 -3.82865220e-01 -6.72994375e-01 4.26006854e-01 6.59256697e-01 8.07752684e-02 -6.11651838e-01 3.27705175e-01 1.37306526e-01 -8.75640750e-01 1.15487683e+00 -4.97807980e-01 4.42708522e-01 -6.81758523e-01 9.57095027e-02 -1.12601376e+00 -3.16614747e-01 -6.36996388e-01 3.69047113e-02 1.07662618e+00 4.76611376e-01 -4.31601226e-01 9.31091309e-01 4.86981601e-01 -7.25701600e-02 -5.84106565e-01 -4.71300423e-01 -6.95915759e-01 -4.23749387e-02 -4.59448099e-01 3.18467051e-01 8.81644785e-01 -6.23995185e-01 3.55491042e-01 -5.76133728e-02 1.22155242e-01 8.19167912e-01 1.78104937e-01 1.08632636e+00 -1.52670431e+00 -1.74753308e-01 -4.96194243e-01 -7.40811765e-01 -9.98375595e-01 8.05875659e-02 -8.40138435e-01 3.77081156e-01 -1.73409474e+00 7.23822042e-02 -5.64720333e-01 -2.26130560e-01 3.79601896e-01 -9.65801552e-02 5.60905337e-01 1.56500667e-01 1.59223661e-01 -7.22497344e-01 7.08561182e-01 1.13652587e+00 -2.02599108e-01 -3.98012668e-01 -1.92909643e-01 -6.58652902e-01 9.93240058e-01 5.56716204e-01 -5.56190968e-01 -7.22707033e-01 -3.40384364e-01 2.60769576e-02 -4.24832791e-01 6.85416937e-01 -9.60870147e-01 1.49145484e-01 -1.35010347e-01 4.34053332e-01 -5.05118549e-01 3.22477400e-01 -9.28492129e-01 2.80995667e-01 2.22056255e-01 -3.43412697e-01 -2.43654743e-01 1.26149999e-02 6.92712784e-01 -1.29239574e-01 -4.46694225e-01 8.27995837e-01 -3.42238694e-01 -8.08376133e-01 1.27058715e-01 -4.94569957e-01 -3.55210640e-02 1.24490273e+00 -9.43639874e-01 2.54950374e-01 -1.30460396e-01 -8.74217749e-01 1.83208045e-02 7.22076058e-01 4.49240714e-01 6.90065622e-01 -1.14450204e+00 -7.06907809e-01 3.43125761e-01 1.43233567e-01 6.18718147e-01 -1.36957213e-01 3.99818897e-01 -7.34973788e-01 1.26989648e-01 -1.07474819e-01 -1.01777649e+00 -9.74455595e-01 6.10787034e-01 1.29711419e-01 -9.45280418e-02 -7.87590325e-01 7.91417420e-01 3.84176850e-01 -6.42297924e-01 -4.52938974e-02 -4.52975333e-01 -2.30207995e-01 1.33327082e-01 -2.81130020e-02 -2.70752218e-02 -3.71794671e-01 -8.40681672e-01 -5.48304319e-02 5.44333458e-01 -9.49967131e-02 -2.25524321e-01 1.49007785e+00 -9.93744656e-03 -5.44831753e-02 6.92144871e-01 1.11305094e+00 -2.41095588e-01 -1.52588165e+00 -1.24975555e-01 3.57044786e-01 -4.04072106e-01 -1.03898600e-01 -3.58695358e-01 -1.03151715e+00 8.75231624e-01 4.25656468e-01 3.04386467e-01 9.64153707e-01 1.47194907e-01 3.90486300e-01 3.67638737e-01 3.24433565e-01 -1.09398353e+00 4.53938186e-01 2.37690240e-01 8.04415643e-01 -1.29052734e+00 6.38339370e-02 -2.89630473e-01 -6.50847733e-01 9.53680038e-01 4.19853806e-01 -4.69743550e-01 4.35328811e-01 5.72804213e-02 8.72503147e-02 -3.58695865e-01 -3.56055707e-01 -5.58897316e-01 4.76634085e-01 7.57062912e-01 2.03468546e-01 -6.32842332e-02 5.80067709e-02 -3.24875750e-02 -1.95369318e-01 -1.82924390e-01 5.02370834e-01 9.23289299e-01 -4.04571772e-01 -1.14349270e+00 -1.67609081e-01 3.59855443e-01 -1.07974716e-01 -3.02062929e-02 -4.50537503e-01 9.63107049e-01 2.30714932e-01 6.96937025e-01 2.59386778e-01 -3.82141396e-02 1.01151347e-01 -5.16979657e-02 5.18076956e-01 -1.16778994e+00 -4.47364390e-01 5.32836057e-02 -3.90559852e-01 -2.37158388e-01 -6.95627689e-01 -7.05107391e-01 -1.27602100e+00 4.36485320e-01 -4.96074617e-01 1.06371284e-01 4.97935176e-01 8.16425979e-01 3.15822333e-01 5.16977191e-01 5.28322816e-01 -1.37740874e+00 -8.90654139e-03 -8.28750789e-01 -4.34639394e-01 6.71021402e-01 2.95876503e-01 -6.21070743e-01 -3.36869031e-01 6.42770708e-01]
[9.652274131774902, 0.6852902770042419]
c270aa83-3b90-4743-b564-7449376ad340
image-question-answering-using-convolutional
1511.05756
null
http://arxiv.org/abs/1511.05756v1
http://arxiv.org/pdf/1511.05756v1.pdf
Image Question Answering using Convolutional Neural Network with Dynamic Parameter Prediction
We tackle image question answering (ImageQA) problem by learning a convolutional neural network (CNN) with a dynamic parameter layer whose weights are determined adaptively based on questions. For the adaptive parameter prediction, we employ a separate parameter prediction network, which consists of gated recurrent unit (GRU) taking a question as its input and a fully-connected layer generating a set of candidate weights as its output. However, it is challenging to construct a parameter prediction network for a large number of parameters in the fully-connected dynamic parameter layer of the CNN. We reduce the complexity of this problem by incorporating a hashing technique, where the candidate weights given by the parameter prediction network are selected using a predefined hash function to determine individual weights in the dynamic parameter layer. The proposed network---joint network with the CNN for ImageQA and the parameter prediction network---is trained end-to-end through back-propagation, where its weights are initialized using a pre-trained CNN and GRU. The proposed algorithm illustrates the state-of-the-art performance on all available public ImageQA benchmarks.
['Hyeonwoo Noh', 'Bohyung Han', 'Paul Hongsuck Seo']
2015-11-18
image-question-answering-using-convolutional-1
http://openaccess.thecvf.com/content_cvpr_2016/html/Noh_Image_Question_Answering_CVPR_2016_paper.html
http://openaccess.thecvf.com/content_cvpr_2016/papers/Noh_Image_Question_Answering_CVPR_2016_paper.pdf
cvpr-2016-6
['multi-modal', 'parameter-prediction']
['miscellaneous', 'miscellaneous']
[ 3.12868923e-01 9.87155586e-02 1.76135883e-01 -6.85266376e-01 -9.95936513e-01 -3.22765350e-01 -1.05209872e-02 3.45120952e-02 -8.20232749e-01 1.04565904e-01 -3.40897925e-02 -3.67859900e-01 1.58008888e-01 -1.09528041e+00 -9.62098002e-01 -8.30892622e-01 9.85145345e-02 6.58512414e-01 6.97362602e-01 -2.13193357e-01 3.00140798e-01 9.56764445e-02 -1.49313128e+00 4.65304315e-01 5.35540640e-01 1.49072909e+00 3.90830785e-01 1.08386815e+00 -2.11618811e-01 9.90474820e-01 -5.44543326e-01 -4.72173691e-01 1.62102550e-01 -5.84343731e-01 -1.05635262e+00 -1.23484045e-01 4.04955417e-01 -6.65940344e-01 -2.82042235e-01 9.13956821e-01 6.09095573e-01 1.54186636e-01 3.47399622e-01 -1.08247256e+00 -8.05075347e-01 6.64612412e-01 -5.68684749e-02 3.31989855e-01 9.02976319e-02 3.62665415e-01 1.41803396e+00 -7.49864459e-01 3.41054827e-01 1.22747469e+00 5.11771739e-01 6.72575474e-01 -8.79051328e-01 -6.86987996e-01 4.29710038e-02 5.61467648e-01 -1.32707155e+00 -1.06815910e-02 9.07072842e-01 -1.09003745e-01 1.10184765e+00 -2.66128629e-01 6.05742812e-01 5.07150054e-01 5.51636331e-02 8.46153378e-01 3.04030657e-01 -1.08487323e-01 4.26289648e-01 -1.68526947e-01 2.53269970e-01 1.23601604e+00 -4.67197418e-01 -3.09175879e-01 2.66327430e-03 -4.78803426e-01 8.42646241e-01 7.45658651e-02 -1.74686432e-01 -3.07017714e-01 -9.52755868e-01 1.35313582e+00 9.95111763e-01 -4.25782390e-02 -4.81365561e-01 6.01785123e-01 3.75158131e-01 5.94589651e-01 -1.23281747e-01 2.69533843e-01 -6.80809021e-01 2.89054394e-01 -6.73235118e-01 2.20530733e-01 6.31682456e-01 6.68894529e-01 1.23015141e+00 -1.45641237e-01 -4.99633998e-01 8.67179751e-01 2.89046735e-01 1.83183774e-01 8.05480957e-01 -8.60119402e-01 5.81394792e-01 9.66315925e-01 -4.08804379e-02 -7.67435730e-01 -3.88271898e-01 -1.98980927e-01 -8.88008654e-01 -9.50563923e-02 3.67806464e-01 -1.67595252e-01 -1.14197910e+00 1.80245888e+00 4.63423491e-01 3.69281292e-01 2.64416963e-01 1.16041911e+00 1.02954388e+00 1.25449336e+00 1.96485654e-01 2.53322691e-01 1.71363854e+00 -1.51419055e+00 -4.11180437e-01 -2.31271926e-02 6.48349285e-01 -4.79648173e-01 1.41096103e+00 1.24069527e-01 -1.23234725e+00 -8.46997499e-01 -1.05956185e+00 -5.97550809e-01 -4.76666629e-01 2.47426391e-01 9.56176594e-02 2.38624811e-01 -1.28953958e+00 2.16118142e-01 -4.37763631e-01 -1.88958570e-01 3.36806566e-01 8.84641349e-01 1.85900740e-02 -5.24500050e-02 -1.45867944e+00 5.16435504e-01 4.64801013e-01 2.96179533e-01 -1.06273448e+00 -4.52154756e-01 -8.77941728e-01 4.04489428e-01 1.93997428e-01 -1.02636397e+00 1.38296771e+00 -1.10544562e+00 -1.84660280e+00 6.46864414e-01 4.37149741e-02 -9.16964710e-01 1.49281537e-02 4.41810898e-02 -1.28458142e-01 7.07094848e-01 -3.74406248e-01 1.14683342e+00 1.28843832e+00 -8.66475761e-01 -6.21885955e-01 -3.01882178e-01 2.94110209e-01 2.29081333e-01 -3.75205964e-01 -8.00720677e-02 -8.29828143e-01 -4.02375877e-01 5.04501089e-02 -8.50124240e-01 -4.66801673e-01 1.05024330e-01 -1.36198163e-01 -5.18714547e-01 8.83376539e-01 -6.25614166e-01 1.33392894e+00 -1.94121265e+00 5.96015155e-02 3.29873025e-01 -9.35571920e-03 4.41555053e-01 -4.57469672e-01 6.13500774e-02 1.02898562e-02 -2.98068821e-01 -5.53641498e-01 -1.21110879e-01 -1.02592401e-01 2.30958909e-01 -2.41848588e-01 2.51092285e-01 6.06714785e-01 1.16233087e+00 -7.49991894e-01 -4.77704197e-01 9.25151855e-02 5.11390150e-01 -8.21468830e-01 8.49575400e-01 -7.99390733e-01 -1.14558131e-01 -7.08366513e-01 2.95359701e-01 5.84450603e-01 -6.44449830e-01 -2.97339827e-01 -3.40500236e-01 2.41763309e-01 1.65048912e-01 -1.11054814e+00 1.68865979e+00 -3.64719659e-01 1.72249794e-01 -1.86584026e-01 -9.03161764e-01 1.13589466e+00 2.67099351e-01 -1.14132687e-02 -6.62818611e-01 4.17870253e-01 1.09785147e-01 7.13606030e-02 -6.65731728e-01 5.13805330e-01 1.98031619e-01 -1.98580980e-01 6.03624821e-01 3.90980989e-01 -8.17774832e-02 -1.43596619e-01 7.41273686e-02 1.22444642e+00 -3.59954774e-01 2.41470560e-02 2.94721499e-02 1.18655682e+00 -1.83223918e-01 2.01789930e-01 6.75695717e-01 -1.87130257e-01 8.87167752e-01 6.26463592e-01 -8.48760307e-01 -1.27193105e+00 -6.64633572e-01 9.45038572e-02 1.39741492e+00 -1.25089124e-01 -2.52619609e-02 -1.11558473e+00 -7.29373157e-01 -2.12933734e-01 2.95177639e-01 -9.52418268e-01 -3.81382227e-01 -8.11355174e-01 -5.78381062e-01 5.50464630e-01 3.44154358e-01 1.08445477e+00 -1.58912134e+00 -9.12337780e-01 4.26788479e-01 -1.07170418e-01 -9.95315850e-01 -7.60752201e-01 2.30687544e-01 -7.59599924e-01 -1.05455875e+00 -8.47720981e-01 -1.32590306e+00 7.78801382e-01 -1.18599311e-01 1.16820133e+00 4.78319347e-01 2.53271796e-02 2.80220509e-01 -2.70338893e-01 3.94389890e-02 -3.43297452e-01 6.02662861e-01 -7.99826860e-01 4.57622468e-01 2.27959678e-01 -2.12841153e-01 -1.03780377e+00 4.75510061e-01 -1.29850125e+00 -1.96193933e-01 5.75065374e-01 1.01451230e+00 8.45650077e-01 -3.19017202e-01 8.37487519e-01 -1.05586195e+00 6.22936845e-01 -4.05495495e-01 -8.54644060e-01 4.64826286e-01 -3.23068857e-01 4.45307285e-01 9.01331127e-01 -4.81850892e-01 -7.67805219e-01 2.83235103e-01 -4.87602651e-01 -4.04111147e-01 1.72847569e-01 2.66844928e-01 -2.23549202e-01 -1.08669907e-01 6.35494053e-01 1.80971697e-01 -1.38544384e-03 -1.43248916e-01 5.66220462e-01 6.20199978e-01 7.07905293e-01 -2.53413618e-01 7.79116571e-01 1.63119569e-01 -3.64299566e-01 -2.97882676e-01 -9.87141848e-01 -3.73017132e-01 -4.48804885e-01 7.63863400e-02 1.20360208e+00 -1.00989091e+00 -8.57289970e-01 6.50885761e-01 -1.25036764e+00 -4.56332982e-01 -1.89460441e-01 6.39992729e-02 -6.05194211e-01 7.10029826e-02 -8.62989306e-01 -4.94379312e-01 -9.44861472e-01 -1.37908375e+00 9.62190390e-01 5.20509183e-01 3.33464116e-01 -8.68047416e-01 -5.57011031e-02 5.45116305e-01 6.76941872e-01 -1.03895135e-01 1.22451973e+00 -8.03677797e-01 -7.01029122e-01 -2.48282179e-01 -4.79762346e-01 4.93541926e-01 -3.66598845e-01 -4.08799678e-01 -9.48080182e-01 -2.28653058e-01 2.88690086e-02 -8.74945581e-01 1.08364236e+00 2.06174672e-01 1.66705942e+00 -3.82404774e-01 4.09842461e-01 6.60602450e-01 1.60493636e+00 -6.38900474e-02 8.38173449e-01 2.58065790e-01 5.74268520e-01 1.65394574e-01 3.50665063e-01 4.88084882e-01 7.71463096e-01 2.34931767e-01 7.91676462e-01 3.65630165e-02 3.97414118e-02 -2.16945603e-01 1.33775085e-01 6.56223476e-01 4.03482169e-01 -2.96966106e-01 -7.52166450e-01 6.25221312e-01 -1.85228574e+00 -5.06171227e-01 3.73134255e-01 2.14293885e+00 1.00865996e+00 1.74042717e-01 1.32332593e-01 -5.54828309e-02 7.67565310e-01 2.80716985e-01 -8.97840738e-01 -4.68119293e-01 2.85867304e-02 3.29672486e-01 2.89892673e-01 4.45181698e-01 -9.62483466e-01 1.23218298e+00 5.43059397e+00 4.47715610e-01 -1.15472651e+00 -1.42362505e-01 9.39949870e-01 1.95247367e-01 -2.90253758e-01 -1.85697585e-01 -8.75010669e-01 2.05791712e-01 1.29446018e+00 5.35596251e-01 4.42265898e-01 8.39012146e-01 -1.32540420e-01 -3.31216417e-02 -8.50415051e-01 8.73439968e-01 2.95512646e-01 -1.28096926e+00 5.57419837e-01 -6.21772885e-01 4.29976583e-01 1.71680734e-01 2.42815405e-01 3.91022444e-01 1.88736379e-01 -9.15009081e-01 4.80958730e-01 4.95700538e-01 5.43919206e-01 -9.08858001e-01 8.26652527e-01 1.15626961e-01 -1.05874777e+00 -5.65061510e-01 -9.90285695e-01 3.13470930e-01 -1.38325945e-01 2.10502908e-01 -8.31351519e-01 -1.48892283e-01 7.61574328e-01 9.97782201e-02 -7.90538669e-01 1.09624982e+00 -1.45403057e-01 7.51727581e-01 -3.45869929e-01 -2.23764107e-01 6.51638925e-01 -1.30352762e-03 -1.52361438e-01 9.85344887e-01 1.73659638e-01 3.69915813e-01 -1.33410409e-01 6.91026986e-01 -5.65017223e-01 1.19350649e-01 6.69526458e-02 1.61886841e-01 2.84242690e-01 1.30319667e+00 -5.66271544e-01 -4.18420404e-01 -2.72311538e-01 1.02098560e+00 6.99987769e-01 3.21958572e-01 -8.72784197e-01 -6.96473122e-01 2.78276026e-01 -1.81168646e-01 8.98470223e-01 5.80351241e-02 1.81258053e-01 -9.29930925e-01 -1.42058864e-01 -7.36298740e-01 1.00457227e+00 -9.34819043e-01 -1.01151025e+00 9.33292747e-01 -3.48937064e-01 -9.63982105e-01 -3.62947762e-01 -3.84456545e-01 -7.17066407e-01 7.58141339e-01 -1.96073222e+00 -1.14532411e+00 -2.80581087e-01 1.13330090e+00 4.98369813e-01 -1.10019721e-01 9.31915283e-01 1.37311414e-01 -5.60428202e-01 9.26039100e-01 -2.27043256e-01 2.72796035e-01 4.75340486e-01 -1.11506593e+00 4.78449851e-01 3.17746937e-01 -9.34267193e-02 3.49585563e-01 3.53708893e-01 -1.00157857e-01 -1.48060024e+00 -1.34600580e+00 9.99498069e-01 4.08091359e-02 6.01524234e-01 -5.30606747e-01 -1.13535023e+00 6.17645562e-01 3.36914003e-01 6.11890495e-01 5.40813029e-01 -4.71886396e-01 -5.98732531e-01 -1.99068666e-01 -1.32922733e+00 3.12753737e-01 3.62842619e-01 -6.55558705e-01 -5.14519989e-01 3.48234892e-01 1.24184132e+00 -4.90632117e-01 -8.14657032e-01 8.08632467e-03 3.70741904e-01 -5.60715914e-01 9.87303138e-01 -6.32991254e-01 3.65556955e-01 -3.46392751e-01 -1.61815733e-02 -8.39632928e-01 -3.58078688e-01 -4.77470487e-01 -5.56255430e-02 8.27698648e-01 7.49649048e-01 -3.91452402e-01 1.07779849e+00 4.84397948e-01 3.77277471e-03 -1.20385265e+00 -8.38375568e-01 6.59531951e-02 -6.78030849e-02 -2.46068642e-01 8.68005931e-01 3.76152426e-01 -7.18640208e-01 7.28410125e-01 -3.10593128e-01 4.14522946e-01 3.04002464e-01 1.57536909e-01 5.22159278e-01 -8.75673234e-01 -2.87789524e-01 2.44161636e-02 -3.30170274e-01 -1.41207790e+00 9.62818414e-03 -6.10103965e-01 1.51253790e-01 -1.33510923e+00 5.49867749e-02 -2.27015451e-01 -5.76385200e-01 5.95687091e-01 -3.26038599e-01 2.85864621e-01 1.38653994e-01 -3.20857875e-02 -1.07847333e+00 8.35025609e-01 1.35742915e+00 -1.89963117e-01 -2.89448321e-01 5.33245690e-03 -3.79461974e-01 4.70001578e-01 8.41424167e-01 -4.72837657e-01 -6.09829545e-01 -8.03934157e-01 4.40489173e-01 3.17597479e-01 4.84403789e-01 -1.04324687e+00 6.83466017e-01 4.37267035e-01 4.00238812e-01 -8.13987911e-01 3.21098745e-01 -8.14291239e-01 -5.19086957e-01 4.37359959e-01 -7.73108065e-01 5.24213672e-01 -5.19409291e-02 4.69781846e-01 -2.64096886e-01 -5.06495714e-01 9.11826909e-01 -2.75500238e-01 -6.03175342e-01 7.30342090e-01 -1.75926164e-01 4.53382879e-02 4.94836152e-01 9.91393179e-02 -2.88455337e-01 -5.09262621e-01 -5.25039554e-01 6.79539502e-01 -4.96553816e-03 3.02363932e-01 1.01590252e+00 -1.21128654e+00 -5.05940795e-01 3.45661521e-01 1.48145542e-01 3.61714602e-01 4.83191043e-01 1.78065553e-01 -6.20536506e-01 1.17506027e-01 -1.22362241e-01 -5.25408208e-01 -7.42841959e-01 5.35812616e-01 6.50196791e-01 -5.27585268e-01 -4.12277371e-01 1.19846904e+00 -1.95899699e-02 -8.00957084e-01 4.59634066e-01 -4.81857896e-01 -5.88007331e-01 -1.06013454e-01 6.47847414e-01 -2.03598067e-01 -5.91792446e-03 -5.11464655e-01 1.14735723e-01 5.76429129e-01 -2.24414021e-01 1.84338149e-02 1.39153636e+00 -3.86706740e-02 -2.41936907e-01 1.98095605e-01 1.79535246e+00 -9.12554562e-01 -1.37546146e+00 -6.38543129e-01 -2.42675543e-01 1.87932238e-01 1.08579487e-01 -6.52612329e-01 -1.41640747e+00 9.44453001e-01 9.14269030e-01 -5.07091358e-02 1.37995517e+00 -5.94057702e-02 1.59620667e+00 9.57039833e-01 -1.45754218e-01 -9.86326337e-01 4.99223471e-01 8.16518307e-01 8.19784284e-01 -1.10583854e+00 -5.18759310e-01 2.26451561e-01 -4.47016239e-01 1.33501267e+00 7.65252829e-01 -6.00425899e-01 9.28111672e-01 -2.56263345e-01 2.21192330e-01 -3.38623941e-01 -9.57049906e-01 -9.50550288e-02 1.49267241e-01 6.51600212e-02 -5.82479089e-02 -3.34312797e-01 -3.04506645e-02 7.33349681e-01 -4.07744437e-01 6.08200468e-02 2.87441581e-01 6.23151720e-01 -6.43659770e-01 -9.70822155e-01 -1.64256856e-01 5.01319468e-01 -3.71658921e-01 -9.13349465e-02 8.46649632e-02 3.61797571e-01 8.30905214e-02 6.95508480e-01 2.99458057e-01 -4.46996987e-01 2.92475879e-01 3.10149997e-01 1.20163158e-01 -4.51015890e-01 -1.33710635e+00 -3.32681924e-01 -3.39353532e-01 -5.51241279e-01 -1.39451161e-01 -3.17602247e-01 -1.67839074e+00 2.03516021e-01 -4.26953673e-01 3.21694046e-01 5.53400815e-01 8.07134032e-01 3.40586126e-01 2.83961326e-01 8.56227577e-01 -3.15441221e-01 -7.73970604e-01 -9.21550274e-01 -1.26326397e-01 2.47981802e-01 5.89071095e-01 4.51310463e-02 -2.38674477e-01 -8.06258544e-02]
[10.385440826416016, 1.9846525192260742]
6c933f00-3d3b-4257-a078-125693e2927d
temporal-pattern-mining-for-analysis-of
2209.04793
null
https://arxiv.org/abs/2209.04793v1
https://arxiv.org/pdf/2209.04793v1.pdf
Temporal Pattern Mining for Analysis of Longitudinal Clinical Data: Identifying Risk Factors for Alzheimer's Disease
A novel framework is proposed for handling the complex task of modelling and analysis of longitudinal, multivariate, heterogeneous clinical data. This method uses temporal abstraction to convert the data into a more appropriate form for modelling, temporal pattern mining, to discover patterns in the complex, longitudinal data and machine learning models of survival analysis to select the discovered patterns. The method is applied to a real-world study of Alzheimer's disease (AD), a progressive neurodegenerative disease that has no cure. The patterns discovered were predictive of AD in survival analysis models with a Concordance index of up to 0.8. This is the first work that performs survival analysis of AD data using temporal data collections for AD. A visualisation module also provides a clear picture of the discovered patterns for ease of interpretability.
['Arcot Sowmya', 'Henry Brodaty', 'Perminder S. Sachdev', 'Gelareh Mohammadi', 'Annette Spooner']
2022-09-11
null
null
null
null
['survival-analysis']
['miscellaneous']
[ 6.76537603e-02 -7.30621442e-02 -2.49514878e-01 -4.94027793e-01 -4.17463407e-02 -2.09031254e-01 6.42430365e-01 5.37556887e-01 -3.63934636e-01 1.01976025e+00 3.65211874e-01 -7.15005994e-01 -8.05318356e-01 -5.34627497e-01 2.10281953e-01 -6.05455279e-01 -1.25609648e+00 7.97079444e-01 4.31169271e-01 1.53438831e-02 -5.16040027e-02 6.31750047e-01 -1.51755285e+00 7.60639310e-01 5.26084363e-01 7.56441474e-01 1.86864585e-01 5.68716705e-01 1.09610066e-01 6.21980011e-01 -3.86425018e-01 7.00074732e-02 -1.99530363e-01 -2.22558409e-01 -7.81346262e-01 -2.75682360e-01 -3.37569982e-01 -1.31005496e-02 1.58280268e-01 3.31290722e-01 3.16564679e-01 -3.59246194e-01 7.08474398e-01 -1.39070022e+00 2.14332238e-01 1.73957020e-01 -5.89847676e-02 6.65600002e-01 5.41049361e-01 -4.94343974e-03 5.71347833e-01 -4.75878894e-01 1.19273114e+00 1.39913154e+00 7.33986795e-01 4.23648357e-01 -1.60418987e+00 -4.84036714e-01 -1.86877668e-01 5.96766770e-01 -9.63395059e-01 -9.68563631e-02 1.77807733e-01 -1.01663995e+00 1.06508887e+00 7.13606596e-01 1.36230278e+00 1.08454812e+00 7.45303333e-01 8.77982676e-02 1.48465908e+00 -3.88592243e-01 5.88250160e-01 5.56536689e-02 7.54802585e-01 3.49057138e-01 -1.68887079e-02 6.74274266e-01 -4.07929957e-01 -9.08524215e-01 4.12051171e-01 4.42462593e-01 7.11823329e-02 -1.19879000e-01 -1.19627857e+00 8.23860168e-01 4.06411253e-02 7.21650720e-01 -5.60626209e-01 -4.70080048e-01 6.73584402e-01 8.62658381e-01 4.90820378e-01 -1.57115564e-01 -9.06044185e-01 -7.12018162e-02 -8.91534686e-01 5.39623141e-01 7.40062952e-01 5.72320342e-01 -1.16388224e-01 -4.64751899e-01 -7.23047405e-02 5.37344337e-01 6.64374530e-01 -1.40203852e-02 6.45904362e-01 -4.93761122e-01 -1.30393475e-01 1.13411403e+00 -1.32456347e-02 -3.57232809e-01 -9.77194369e-01 -3.52732778e-01 -7.25879550e-01 7.18980670e-01 5.34109831e-01 2.19963923e-01 -9.36252773e-01 1.37275708e+00 2.59935796e-01 -3.19432735e-01 1.47250503e-01 3.78464103e-01 2.99203545e-01 2.01674029e-01 5.56132197e-01 -9.03521419e-01 1.82809639e+00 -2.09500398e-02 -9.26147878e-01 1.86495513e-01 1.06503093e+00 -2.68159956e-01 5.70173502e-01 5.33809483e-01 -6.58203244e-01 1.38690457e-01 -8.50264728e-01 3.64378184e-01 -6.24705017e-01 9.23778713e-02 6.67349875e-01 3.66728038e-01 -1.02681851e+00 7.58553028e-01 -1.20201194e+00 -1.12735128e+00 5.65779090e-01 4.08906072e-01 -5.21217227e-01 3.16337943e-02 -1.14040840e+00 1.08966959e+00 6.30245090e-01 -3.60959142e-01 -6.22937083e-01 -1.07914448e+00 -2.01738968e-01 -3.86682302e-01 -3.18148971e-01 -1.01922095e+00 8.33580971e-01 -7.82969594e-01 -5.16428351e-01 8.59238267e-01 -3.89545381e-01 -5.66700101e-01 6.95514679e-01 1.38229325e-01 -9.56680715e-01 -9.95316058e-02 2.17738878e-02 -3.75070684e-02 1.62846789e-01 -7.41395712e-01 -7.29254007e-01 -1.14620841e+00 -6.59002364e-01 -4.96826649e-01 -1.64449383e-02 6.03762925e-01 5.16633153e-01 -7.26189733e-01 -6.29382655e-02 -7.88546443e-01 -7.26835907e-01 -1.51029959e-01 1.53783318e-02 -4.61348325e-01 9.42758799e-01 -1.01302850e+00 1.66313171e+00 -1.84676611e+00 1.23576716e-01 3.08592290e-01 2.75262147e-01 -3.10180724e-01 5.16050994e-01 7.46352255e-01 -5.33549070e-01 6.55640364e-02 -4.26863551e-01 -1.50113463e-01 -4.97974455e-01 4.02037412e-01 -5.88959903e-02 2.93432832e-01 1.45740556e-02 4.34330195e-01 -6.79799497e-01 -3.93607318e-01 -8.10323507e-02 2.41898090e-01 -1.10496178e-01 8.76721814e-02 -2.60790318e-01 4.54499960e-01 -4.33466405e-01 6.18705273e-01 3.68163437e-01 -2.09857255e-01 6.28015101e-01 1.19076312e-01 -4.53906417e-01 -3.05040972e-03 -7.75075853e-01 1.18825448e+00 1.47332430e-01 3.77211273e-01 -3.72519910e-01 -6.41186714e-01 1.05781400e+00 7.09456742e-01 7.48329043e-01 -5.94417870e-01 -1.45681486e-01 2.47885138e-01 1.23284593e-01 -8.64167631e-01 -3.63362044e-01 -3.39565068e-01 2.82230765e-01 5.73979139e-01 -2.39580736e-01 9.17631507e-01 4.38814163e-01 5.36920726e-02 1.61433160e+00 -2.12213069e-01 7.71097064e-01 -5.21863163e-01 4.04324353e-01 5.91077447e-01 7.09226251e-01 6.42487109e-02 -7.56363273e-02 8.93783867e-02 7.26993442e-01 -1.08309972e+00 -1.21079385e+00 -1.02718043e+00 -5.89280963e-01 4.62154180e-01 -9.58199799e-01 -8.36798668e-01 -1.46935388e-01 -6.10394776e-01 -1.26482636e-01 5.66371858e-01 -7.74173260e-01 2.45191772e-02 -3.97569180e-01 -1.42162669e+00 7.29871318e-02 3.33404064e-01 -1.23936959e-01 -1.01148582e+00 -1.00494540e+00 4.97317433e-01 2.11247474e-01 -2.96377391e-01 4.21148926e-01 2.47833401e-01 -1.85394597e+00 -1.56963134e+00 -4.12214339e-01 -6.92247748e-01 5.14848053e-01 -5.80289960e-01 1.01558685e+00 1.00702327e-02 -5.47103047e-01 1.41689792e-01 -4.00193870e-01 -5.15335202e-01 -6.12186491e-01 -2.99721152e-01 8.06411430e-02 -3.50195885e-01 6.45412862e-01 -1.08955324e+00 -4.64751303e-01 2.90349692e-01 -8.95381033e-01 -2.00998202e-01 5.52142143e-01 6.47354722e-01 5.28744698e-01 4.00852352e-01 7.16367245e-01 -4.92764056e-01 6.34835005e-01 -1.06879950e+00 -3.61913741e-01 2.72682428e-01 -1.19601381e+00 -3.10647786e-02 2.97610909e-01 -5.35242200e-01 -6.24887586e-01 3.66848707e-01 4.06414181e-01 7.10911602e-02 -4.45576459e-01 7.74505258e-01 1.00482032e-01 5.36399543e-01 6.76023841e-01 1.03245817e-01 4.72866535e-01 -9.30698812e-01 -2.92743206e-01 4.41266477e-01 1.97649910e-03 1.18467502e-01 7.13534206e-02 5.29326797e-01 4.29903358e-01 -5.27388036e-01 2.06083074e-01 -4.41019297e-01 -1.12611687e+00 -1.95226714e-01 6.70330465e-01 -6.21200144e-01 -5.34300864e-01 3.16807449e-01 -1.08586824e+00 -2.55727291e-01 -2.66239028e-02 7.75149643e-01 -5.52626789e-01 -2.12222666e-01 -2.04426587e-01 -9.32664931e-01 -2.83749998e-01 -5.04884303e-01 4.81155455e-01 -3.29476565e-01 -8.37411523e-01 -1.24833977e+00 5.33109903e-01 -1.66031331e-01 2.51974165e-01 8.09117317e-01 1.57667303e+00 -8.89112711e-01 -9.94627848e-02 -1.73851326e-01 1.06581531e-01 -2.89255202e-01 1.59198612e-01 6.65060133e-02 -4.37208116e-01 -3.35854739e-01 6.58313408e-02 3.91824514e-01 5.85290074e-01 4.34707165e-01 4.60298687e-01 -4.75930482e-01 -9.56830323e-01 -8.46037790e-02 1.22842312e+00 1.05136681e+00 7.98110008e-01 9.25971210e-01 -3.23655754e-02 1.08973289e+00 6.42919123e-01 4.99731213e-01 2.69920975e-01 1.04822373e+00 2.44444549e-01 6.51054382e-02 2.55558342e-01 4.02416855e-01 3.99524271e-01 8.52246061e-02 -3.59173536e-01 3.23091567e-01 -1.26313996e+00 4.90161061e-01 -2.02944160e+00 -9.83799160e-01 -8.74302804e-01 2.25076199e+00 6.29948556e-01 6.29131272e-02 7.37590611e-01 5.37262917e-01 4.27277297e-01 -4.44301695e-01 7.43954617e-04 -4.11202252e-01 3.53313647e-02 1.24832958e-01 1.79873139e-01 2.81371444e-01 -8.81163001e-01 8.33873004e-02 7.15782690e+00 1.30497813e-01 -7.25182772e-01 2.11139724e-01 4.04177517e-01 -2.33960375e-01 3.46676148e-02 2.83885360e-01 -3.19927037e-01 5.86221278e-01 1.51906633e+00 -2.52901554e-01 2.76716910e-02 3.22482586e-01 9.52863634e-01 -1.57447964e-01 -1.13876939e+00 5.17973661e-01 -6.22288048e-01 -1.19485569e+00 -1.22686706e-01 4.26739633e-01 -3.37957107e-02 -2.53829271e-01 -5.09237528e-01 -1.99749947e-01 2.66610920e-01 -1.03287280e+00 5.67073047e-01 1.28550494e+00 5.34894705e-01 -5.68325996e-01 8.39775205e-01 1.43426970e-01 -8.67715120e-01 -5.28166294e-01 2.81286716e-01 -2.03759789e-01 6.24314845e-01 8.58044744e-01 -1.24875128e+00 6.35128915e-01 1.09037066e+00 1.03992546e+00 -7.52349198e-01 1.21321297e+00 3.42200041e-01 6.93112552e-01 -9.05467793e-02 2.65291065e-01 -6.63260221e-02 1.82345118e-02 8.54438484e-01 1.17062068e+00 6.79102957e-01 8.67219344e-02 -2.45255053e-01 7.21187472e-01 1.09979844e+00 1.60444230e-01 -7.68243909e-01 2.90851351e-02 2.19370380e-01 8.33637357e-01 -5.89451492e-01 -2.43801951e-01 -1.75072178e-01 3.96359593e-01 -3.46215278e-01 1.41640440e-01 -1.64956659e-01 2.64881313e-01 4.31585401e-01 8.15452278e-01 -1.27821475e-01 -3.62790674e-01 -3.72743487e-01 -5.40019333e-01 1.31031796e-01 -7.90781140e-01 1.26780033e+00 -6.37332082e-01 -1.42555559e+00 8.07688177e-01 4.76202935e-01 -1.28344202e+00 -4.67493683e-01 -4.55890208e-01 -8.20982933e-01 9.98191178e-01 -5.76293468e-01 -1.19233453e+00 7.44510219e-02 7.15358853e-01 6.51159808e-02 -5.10890186e-01 1.22044206e+00 3.18296522e-01 -4.81176615e-01 -3.44714612e-01 2.13256687e-01 -6.23197973e-01 4.66682345e-01 -1.36321616e+00 -4.34764847e-02 3.44083726e-01 -7.91399062e-01 5.77687621e-01 8.75606835e-01 -1.19929862e+00 -8.04050505e-01 -1.09373343e+00 1.49320793e+00 -4.67085868e-01 8.30691516e-01 -1.30379707e-01 -1.02019548e+00 5.36522508e-01 -1.10013276e-01 -3.03716451e-01 9.45863008e-01 1.39249593e-01 3.39377746e-02 -1.15675673e-01 -1.40566385e+00 2.88446277e-01 8.37667584e-01 8.94377157e-02 -7.48065472e-01 2.96163857e-01 3.61950845e-01 5.43351352e-01 -1.54483867e+00 3.30355465e-01 9.56930697e-01 -9.83759582e-01 9.31159019e-01 -1.12767172e+00 8.07854757e-02 -3.04434568e-01 3.36684436e-02 -1.04001045e+00 -3.55986625e-01 -3.45483392e-01 -2.70774364e-01 1.02999294e+00 5.39783001e-01 -6.82277679e-01 3.88852030e-01 5.76556742e-01 1.27560675e-01 -6.78264320e-01 -1.35025430e+00 -8.61538529e-01 -2.34956443e-01 -2.69933134e-01 4.62704778e-01 8.98755968e-01 2.30112419e-01 1.03572749e-01 -1.11744195e-01 6.30127788e-02 5.79140663e-01 -2.26235807e-01 1.01464711e-01 -2.09126186e+00 2.21348032e-01 -2.66249806e-01 -1.11390114e+00 5.49266160e-01 -3.49902630e-01 -8.13694656e-01 -9.19661582e-01 -1.54172313e+00 1.57949373e-01 -6.49265170e-01 -3.18654776e-01 4.95852143e-01 4.99463409e-01 -3.60618830e-01 -1.66821882e-01 5.90997338e-01 1.14634000e-01 3.98733206e-02 7.54868448e-01 -7.84709528e-02 -6.15273297e-01 3.05588990e-01 -1.92056626e-01 3.18382889e-01 1.01956332e+00 -9.49814498e-01 -6.04035616e-01 4.88075733e-01 8.28689709e-02 1.89572304e-01 7.76908934e-01 -8.61921966e-01 6.18761918e-03 -1.76397100e-01 5.15688598e-01 -8.40520918e-01 -1.36953697e-01 -1.21692085e+00 1.31027317e+00 1.13214266e+00 -1.31416470e-01 4.62949067e-01 1.84584439e-01 6.28289998e-01 -1.28369913e-01 8.59641135e-02 5.52131891e-01 -7.81333167e-03 -6.93359137e-01 1.20214731e-01 -8.28169227e-01 -7.23524988e-01 1.39406395e+00 -2.95340925e-01 -2.84559041e-01 2.41575435e-01 -1.78387845e+00 1.03664890e-01 3.16715240e-01 4.24244255e-01 4.61740702e-01 -1.48627281e+00 -7.28518248e-01 1.79146856e-01 1.80444479e-01 -4.52914029e-01 1.63760543e-01 1.48507321e+00 -2.22928897e-01 4.71280843e-01 -7.25466192e-01 -7.09141374e-01 -2.00759959e+00 7.27583110e-01 3.08191150e-01 -6.16273522e-01 -1.10395384e+00 -2.54229128e-01 -4.09788668e-01 2.05230027e-01 3.19342539e-02 -2.81054139e-01 -8.34266841e-01 6.34136081e-01 9.40657794e-01 7.61057615e-01 2.43852764e-01 -4.58246559e-01 -7.98453033e-01 1.05284251e-01 -1.59329981e-01 -2.43282706e-01 1.92562592e+00 -1.68715566e-01 -6.97674870e-01 8.97250533e-01 9.54539835e-01 -6.16216242e-01 -6.70324504e-01 1.30910844e-01 9.47347105e-01 -6.28192052e-02 -1.26581877e-01 -1.25238109e+00 -6.26213193e-01 2.96616435e-01 1.35441935e+00 5.05241334e-01 1.16332603e+00 2.91967958e-01 1.42333418e-01 7.11400658e-02 2.29038462e-01 -6.09665692e-01 -6.19933486e-01 2.00793087e-01 1.27826083e+00 -8.40545177e-01 8.61342251e-02 -1.17647782e-01 -3.07176679e-01 1.64355385e+00 -6.62857443e-02 1.85182735e-01 1.03361452e+00 2.49505654e-01 -1.71082050e-01 -6.13056481e-01 -1.31307888e+00 2.40887910e-01 1.23369925e-01 9.93235648e-01 3.77332956e-01 3.34479749e-01 -1.08411407e+00 1.03576684e+00 1.28332660e-01 7.11408496e-01 2.26004094e-01 1.15693879e+00 -2.98254162e-01 -1.45449948e+00 -4.62351471e-01 7.76941061e-01 -4.36064303e-01 7.64236674e-02 -6.28957927e-01 1.05254579e+00 1.74293384e-01 7.23709285e-01 1.41969681e-01 -2.63500988e-01 6.04492068e-01 5.87955892e-01 2.19701394e-01 -9.78534818e-02 -5.14989018e-01 -8.28382149e-02 4.93109912e-01 -5.45478880e-01 -6.98651552e-01 -1.17717016e+00 -1.22233415e+00 -1.55062795e-01 3.06263298e-01 3.42577919e-02 5.87101340e-01 8.11621070e-01 4.71958935e-01 6.31375253e-01 5.46111941e-01 -1.70037106e-01 7.81445056e-02 -1.06573641e+00 -8.06989431e-01 1.47985026e-01 3.45460057e-01 -7.83972800e-01 -3.09004754e-01 5.46341836e-01]
[7.979310989379883, 5.497529983520508]
11f3bdfe-bea2-4ae1-853b-65f25cdab2bf
learning-compatibility-across-categories-for
1603.09473
null
http://arxiv.org/abs/1603.09473v3
http://arxiv.org/pdf/1603.09473v3.pdf
Learning Compatibility Across Categories for Heterogeneous Item Recommendation
Identifying relationships between items is a key task of an online recommender system, in order to help users discover items that are functionally complementary or visually compatible. In domains like clothing recommendation, this task is particularly challenging since a successful system should be capable of handling a large corpus of items, a huge amount of relationships among them, as well as the high-dimensional and semantically complicated features involved. Furthermore, the human notion of "compatibility" to capture goes beyond mere similarity: For two items to be compatible---whether jeans and a t-shirt, or a laptop and a charger---they should be similar in some ways, but systematically different in others. In this paper we propose a novel method, Monomer, to learn complicated and heterogeneous relationships between items in product recommendation settings. Recently, scalable methods have been developed that address this task by learning similarity metrics on top of the content of the products involved. Here our method relaxes the metricity assumption inherent in previous work and models multiple localized notions of 'relatedness,' so as to uncover ways in which related items should be systematically similar, and systematically different. Quantitatively, we show that our system achieves state-of-the-art performance on large-scale compatibility prediction tasks, especially in cases where there is substantial heterogeneity between related items. Qualitatively, we demonstrate that richer notions of compatibility can be learned that go beyond similarity, and that our model can make effective recommendations of heterogeneous content.
['Ruining He', 'Julian McAuley', 'Charles Packer']
2016-03-31
null
null
null
null
['product-recommendation']
['miscellaneous']
[ 5.56925870e-02 -4.13977802e-01 -2.49560416e-01 -4.93484855e-01 -2.64156044e-01 -1.02094483e+00 9.19851512e-02 3.83403391e-01 -6.26674201e-03 1.10467307e-01 4.19202894e-01 -2.19782934e-01 -6.36913717e-01 -5.56039095e-01 -6.05041325e-01 -3.42545748e-01 -3.17342371e-01 4.47708845e-01 9.50428694e-02 -6.17348492e-01 2.97701925e-01 2.11260661e-01 -1.77467263e+00 4.45335358e-01 6.95560515e-01 1.10755634e+00 5.60544789e-01 3.86283219e-01 1.20634608e-01 3.39369208e-01 -1.24285936e-01 -7.59259820e-01 5.15146613e-01 -3.51025462e-01 -7.46469676e-01 3.22608888e-01 8.01793754e-01 -3.42263505e-02 1.70510635e-01 1.03666925e+00 1.34207949e-01 4.45236921e-01 8.59771609e-01 -1.01096773e+00 -1.16893256e+00 6.37708664e-01 -5.43464899e-01 -1.20665722e-01 7.29559541e-01 -2.02535257e-01 1.74972498e+00 -6.73171282e-01 5.06565452e-01 9.84446466e-01 8.44312727e-01 2.67343879e-01 -1.47574520e+00 -2.69179791e-01 5.97313762e-01 7.96351805e-02 -1.15972197e+00 -1.06401831e-01 7.01092482e-01 -5.34102499e-01 5.98875940e-01 6.75690591e-01 7.13255882e-01 8.39692593e-01 -1.51809687e-02 5.28637350e-01 8.45011353e-01 -3.02267700e-01 1.80797517e-01 3.62268776e-01 2.40246192e-01 4.89608616e-01 4.23190355e-01 -2.53441155e-01 -3.08436304e-01 -5.42079136e-02 4.43891406e-01 2.14361340e-01 -1.52865127e-01 -6.15920007e-01 -1.10677135e+00 8.11697662e-01 3.40241730e-01 4.64348018e-01 -2.48933867e-01 -2.78207809e-01 1.68015614e-01 5.67250252e-01 1.86478734e-01 1.02822459e+00 -5.56257606e-01 2.82953292e-01 -4.26106751e-01 3.80562484e-01 9.47101593e-01 1.29847467e+00 6.26624703e-01 -6.45827591e-01 4.30220179e-02 8.60079050e-01 1.78048924e-01 7.24433810e-02 2.48304248e-01 -7.98736811e-01 3.84043306e-01 6.05130374e-01 4.00285512e-01 -1.36161304e+00 -5.90667784e-01 -4.43080246e-01 -6.83410883e-01 -1.10220797e-01 5.11646330e-01 2.07158893e-01 -3.00224215e-01 1.98599780e+00 3.47218722e-01 -7.51201808e-02 -2.09062353e-01 1.13158083e+00 6.96023762e-01 2.13429824e-01 -3.55863050e-02 -1.36930630e-01 1.64326227e+00 -8.86027336e-01 -4.23134357e-01 -1.05088241e-01 8.24123859e-01 -1.08888137e+00 1.55567753e+00 3.63000691e-01 -1.22610438e+00 -7.69671321e-01 -1.00376809e+00 -1.61750495e-01 -4.57371742e-01 6.36168616e-03 1.00229537e+00 6.80364788e-01 -8.25403214e-01 8.95743847e-01 6.53625056e-02 -5.67811728e-01 1.69444364e-02 4.76012349e-01 -3.59707624e-01 -2.94375986e-01 -1.04223514e+00 9.40236151e-01 4.56237532e-02 -1.36330426e-01 -1.05487138e-01 -7.97768235e-01 -8.10761034e-01 2.19196364e-01 6.12688780e-01 -8.10370147e-01 1.00127971e+00 -1.17137229e+00 -9.73746419e-01 6.64570451e-01 2.46222228e-01 1.40279979e-01 1.24702431e-01 -1.70672655e-01 -7.31187880e-01 -3.30113113e-01 9.25047770e-02 3.58140111e-01 6.38302028e-01 -1.21184862e+00 -8.01537871e-01 -4.57331002e-01 6.81878209e-01 4.79412675e-01 -4.77424502e-01 2.68323839e-01 -6.17744267e-01 -9.09940422e-01 2.42800992e-02 -1.22770929e+00 -1.68922678e-01 2.90339649e-01 -2.60427147e-01 -2.31942207e-01 2.08465293e-01 -4.01504129e-01 1.13065875e+00 -2.23215747e+00 2.90755302e-01 4.57024485e-01 2.75336325e-01 1.32151069e-02 -4.81353253e-01 3.86192501e-01 -2.65440308e-02 -5.81729859e-02 3.04136872e-01 -8.03050250e-02 3.73042285e-01 3.48261036e-02 5.67668602e-02 2.68462837e-01 -2.03428715e-02 7.66830802e-01 -9.23268497e-01 -5.77504300e-02 -6.41948283e-02 3.19360197e-01 -6.39849901e-01 1.58383459e-01 -3.18196744e-01 2.27398112e-01 -3.73771369e-01 5.56361139e-01 6.20306373e-01 -6.11672461e-01 5.20314991e-01 -6.30049050e-01 1.77147329e-01 3.31566960e-01 -1.57943475e+00 1.65587437e+00 -5.45872271e-01 1.35216221e-01 -1.19673952e-01 -8.63418281e-01 7.60281026e-01 -4.78606176e-04 5.85049391e-01 -8.10418367e-01 1.86043620e-01 -2.78431154e-03 2.11106032e-01 -5.43622553e-01 6.93328023e-01 -1.95506930e-01 -9.69463885e-02 6.57121062e-01 -2.47968599e-01 2.01844692e-01 3.03844243e-01 1.88052699e-01 9.44549441e-01 -1.95934381e-02 2.14870468e-01 -5.10999322e-01 2.81139702e-01 -1.97218597e-01 3.93207222e-01 6.45592272e-01 2.24895984e-01 4.20442969e-01 2.18204945e-01 -3.47163290e-01 -1.03768623e+00 -9.81779873e-01 5.57972305e-02 1.46396971e+00 7.11398602e-01 -6.49083018e-01 -3.12811315e-01 -8.21753621e-01 3.62181664e-01 3.56826842e-01 -7.05222905e-01 -1.70379609e-01 -2.91090459e-01 -3.88078898e-01 -6.69702962e-02 6.02704346e-01 -1.37643933e-01 -6.31641626e-01 -1.48375213e-01 2.02744260e-01 -3.17547880e-02 -9.89204586e-01 -9.48403716e-01 5.06829210e-02 -5.98323286e-01 -1.11499310e+00 -4.75892454e-01 -1.02189493e+00 6.56393647e-01 9.00456250e-01 1.59410286e+00 2.90249258e-01 -2.10321367e-01 2.80546010e-01 -5.53391039e-01 1.05115166e-02 3.50535400e-02 -3.28453392e-01 3.89962465e-01 1.08891562e-01 3.32724363e-01 -6.44710720e-01 -7.26889551e-01 9.24890101e-01 -8.60760689e-01 1.22079931e-01 5.01759112e-01 6.56691015e-01 6.91724777e-01 3.99710745e-01 5.02535343e-01 -1.01240504e+00 7.35793293e-01 -6.99399173e-01 -2.11405441e-01 6.57688916e-01 -6.22058451e-01 5.61478771e-02 9.33388054e-01 -8.96459460e-01 -7.10775197e-01 -1.35497957e-01 -2.49436516e-02 -8.74900296e-02 -7.08827749e-02 4.73923415e-01 -4.34180051e-01 -9.60363597e-02 5.07910550e-01 -3.11041653e-01 -2.84561396e-01 -8.88567924e-01 6.86212718e-01 4.56479281e-01 2.89584965e-01 -8.25834095e-01 6.46459997e-01 3.41637254e-01 6.17473200e-02 -3.70282650e-01 -1.08640158e+00 -6.62383854e-01 -6.57005847e-01 5.25479428e-02 4.95774925e-01 -8.17389071e-01 -7.96393335e-01 -4.80890600e-03 -6.17860019e-01 -4.92327288e-02 -3.11296850e-01 4.57760453e-01 -3.54864597e-01 4.25960183e-01 -5.17028272e-01 -2.98279494e-01 -1.47210620e-02 -9.09157217e-01 7.10457146e-01 2.35614404e-01 -4.41217721e-01 -9.93465722e-01 5.55798262e-02 3.83896381e-01 4.25892949e-01 -2.32037440e-01 1.22626019e+00 -7.92802632e-01 -3.00520092e-01 -1.06496871e-01 -2.99962491e-01 9.63077918e-02 5.65798223e-01 -2.62322158e-01 -2.72559613e-01 -4.66818839e-01 -1.77702501e-01 -1.45936817e-01 4.57429945e-01 1.99658662e-01 1.06717622e+00 -4.27013516e-01 -2.28756875e-01 4.37191278e-01 1.23833573e+00 7.29623064e-02 3.91315073e-01 -1.23714032e-02 9.55212474e-01 1.07166362e+00 7.57487297e-01 3.88927788e-01 6.45065486e-01 1.21703112e+00 1.00813448e-01 -8.97028744e-02 3.44863161e-02 -1.52900979e-01 6.69912845e-02 8.72191072e-01 -9.72837582e-02 -2.52291620e-01 -1.24614961e-01 3.20149034e-01 -1.92329443e+00 -9.57554519e-01 -1.96644679e-01 2.45280862e+00 6.78494275e-01 -2.33330391e-02 6.24676585e-01 -2.32353583e-02 6.92234516e-01 -3.17018837e-01 -5.74756324e-01 -4.27824378e-01 -6.13266937e-02 1.13877440e-02 7.81419650e-02 1.76286712e-01 -1.05777431e+00 5.49999952e-01 5.90694332e+00 6.58993959e-01 -8.25491548e-01 -1.61190704e-01 3.88126880e-01 -1.87043279e-01 -6.87393963e-01 -1.03811592e-01 -7.31015265e-01 6.29447758e-01 4.84676987e-01 2.00864300e-01 5.43367267e-01 7.08914042e-01 -7.60956258e-02 3.18719521e-02 -1.58671236e+00 9.27349627e-01 1.54305443e-01 -1.15283513e+00 -3.01156752e-02 1.00081854e-01 7.56101251e-01 -5.82798064e-01 4.01181489e-01 8.04409385e-02 2.97309548e-01 -8.86345029e-01 6.62927747e-01 3.59760731e-01 4.65012729e-01 -6.86820567e-01 3.63160878e-01 1.54510602e-01 -1.53238118e+00 -1.43327564e-01 -6.12256229e-01 -1.64789602e-01 -2.64628530e-02 5.31112611e-01 -1.18881904e-01 5.77720225e-01 7.59954035e-01 8.48166943e-01 -5.81154585e-01 1.02103376e+00 1.91148505e-01 -1.04564335e-02 -1.20341875e-01 -1.17825739e-01 -7.08755758e-03 -3.79155725e-01 1.31030157e-01 9.61274326e-01 4.12076294e-01 1.45649642e-01 2.43143395e-01 7.11633027e-01 -1.37566105e-01 5.56300163e-01 -4.73249584e-01 6.54150695e-02 3.01893771e-01 1.38937914e+00 -5.93424797e-01 -1.15944013e-01 -8.24763656e-01 9.60884988e-01 4.04528439e-01 -1.16543490e-02 -7.57437766e-01 4.50258935e-03 9.69501793e-01 2.54049003e-01 5.64888656e-01 -7.53348693e-02 -1.29273966e-01 -1.27961981e+00 3.79810870e-01 -9.82427478e-01 6.01099789e-01 -6.66834533e-01 -2.03789997e+00 4.50230360e-01 -2.38383785e-01 -1.54449487e+00 2.55341858e-01 -7.65610576e-01 -5.22080183e-01 6.26852989e-01 -1.24276996e+00 -1.25753438e+00 -1.82262152e-01 7.67253280e-01 3.62447888e-01 -1.16115939e-02 7.19292462e-01 4.54002082e-01 -3.78476560e-01 7.90009856e-01 3.60649467e-01 -4.09050792e-01 8.22060049e-01 -1.17900038e+00 3.53391528e-01 3.99935424e-01 6.67019844e-01 9.44479465e-01 6.98338628e-01 -5.68258882e-01 -1.78100419e+00 -7.55023420e-01 8.45972478e-01 -4.19865072e-01 7.59367526e-01 -5.22032320e-01 -9.43621039e-01 4.34167236e-01 -1.87853828e-01 -2.36800201e-02 1.19148517e+00 9.16454136e-01 -9.16382134e-01 -5.06641231e-02 -9.68162894e-01 7.02675641e-01 1.36429942e+00 -4.39581871e-01 -4.94569421e-01 5.38022637e-01 5.81897140e-01 -5.13583384e-02 -1.15903640e+00 2.44267508e-01 1.05087960e+00 -9.47059631e-01 1.23909163e+00 -1.04555655e+00 3.67019743e-01 -3.01849425e-01 -5.21427989e-01 -1.37286568e+00 -9.12588716e-01 -4.99935001e-01 -1.45229399e-01 1.12737048e+00 4.30120736e-01 -2.27489173e-01 6.62746072e-01 9.33786035e-01 1.77385807e-02 -9.22806382e-01 -2.76119351e-01 -1.06990612e+00 -2.21814200e-01 -2.47024536e-01 8.07117581e-01 1.11656606e+00 3.64350230e-01 6.79563940e-01 -6.60639942e-01 1.06425732e-01 3.83695215e-01 6.24340892e-01 5.37361443e-01 -1.47520030e+00 -9.05311763e-01 -6.45632923e-01 -3.94671768e-01 -1.34240270e+00 -1.83467865e-02 -8.07140112e-01 -1.74192026e-01 -1.29210055e+00 4.08392012e-01 -9.98716593e-01 -6.37045860e-01 1.69692501e-01 -3.38610351e-01 3.55412036e-01 3.48989815e-01 3.17833811e-01 -9.55049694e-01 5.72826527e-02 1.26502085e+00 -9.56246182e-02 -2.71486759e-01 3.27090323e-01 -1.37117255e+00 6.27847612e-01 4.72352266e-01 -2.04749852e-01 -5.72093904e-01 -3.70051503e-01 8.29262137e-01 -1.53542653e-01 1.23980865e-01 -5.87627709e-01 1.87719092e-02 -3.07034969e-01 4.45688963e-01 -1.68338522e-01 3.75122815e-01 -1.14421999e+00 5.04899502e-01 -3.77399940e-03 -5.73210537e-01 2.94945359e-01 -1.35023534e-01 5.76227725e-01 2.94352442e-01 -2.35599607e-01 5.19770801e-01 -7.37614781e-02 -7.23506808e-01 3.17509532e-01 1.58716142e-02 -3.53153446e-04 9.74278688e-01 -1.81876078e-01 -3.33332360e-01 -3.76446724e-01 -7.32236147e-01 3.86505127e-02 8.13400507e-01 7.87268221e-01 4.51714486e-01 -1.58095932e+00 -5.26001275e-01 3.04475594e-02 5.95203996e-01 -6.34859443e-01 3.23916763e-01 8.29443216e-01 1.04783721e-01 9.40015889e-04 -3.94453913e-01 -2.18910187e-01 -1.35200083e+00 1.10151589e+00 -5.93387373e-02 -2.13249668e-01 -5.94727337e-01 8.41531336e-01 6.36049509e-01 -2.22176984e-01 2.80782253e-01 -3.91905606e-01 -3.74933481e-01 1.48066491e-01 6.11193895e-01 3.21852975e-02 1.74551293e-01 -6.77080274e-01 -1.94009617e-01 8.73726606e-01 -4.41709548e-01 3.73802066e-01 1.20508254e+00 -3.41438919e-01 9.78626832e-02 1.61910951e-01 1.13796687e+00 2.87999064e-01 -1.00400603e+00 -5.14412463e-01 -3.92392017e-02 -8.58110070e-01 -3.21320087e-01 -9.41067040e-01 -1.10690022e+00 3.53552699e-01 3.16747606e-01 5.85308194e-01 1.19588733e+00 4.01434153e-01 8.63554060e-01 4.83070552e-01 4.35035378e-01 -1.01852083e+00 2.62018710e-01 8.41775015e-02 6.93950534e-01 -1.37091911e+00 1.54919431e-01 -5.06941438e-01 -9.81758535e-01 7.91919410e-01 4.75724995e-01 -6.51991144e-02 7.30253935e-01 -4.14042510e-02 -3.63025457e-01 -2.97416449e-01 -7.12834239e-01 -4.39309299e-01 9.55526173e-01 5.43599069e-01 5.62772870e-01 2.82144874e-01 -2.20331132e-01 9.92620468e-01 -4.37958241e-02 -5.11747658e-01 9.16878656e-02 7.50222802e-01 -3.55722159e-01 -1.36851680e+00 -4.86835465e-02 7.10901558e-01 -1.88569754e-01 -1.44908160e-01 -4.44653302e-01 4.41155195e-01 3.31139922e-01 1.05101085e+00 1.02821700e-01 -7.54369676e-01 5.62560499e-01 -5.29424429e-01 7.52993107e-01 -6.05126977e-01 -8.34345281e-01 5.43889590e-02 2.79021204e-01 -5.23059964e-01 -4.55221862e-01 -6.61922395e-01 -8.04636896e-01 -5.15230358e-01 -5.55450082e-01 5.34061939e-02 6.31606519e-01 1.02317500e+00 3.97497922e-01 8.82452354e-02 9.39422309e-01 -6.67539299e-01 -3.62220377e-01 -4.68578547e-01 -1.04839647e+00 1.09438848e+00 9.95135680e-03 -8.81589293e-01 -1.38369258e-02 -4.57330011e-02]
[10.151455879211426, 5.530247211456299]
e8f81668-b53a-4139-a6de-cc828aa6c0f8
local-frequency-domain-transformer-networks
2105.04637
null
https://arxiv.org/abs/2105.04637v1
https://arxiv.org/pdf/2105.04637v1.pdf
Local Frequency Domain Transformer Networks for Video Prediction
Video prediction is commonly referred to as forecasting future frames of a video sequence provided several past frames thereof. It remains a challenging domain as visual scenes evolve according to complex underlying dynamics, such as the camera's egocentric motion or the distinct motility per individual object viewed. These are mostly hidden from the observer and manifest as often highly non-linear transformations between consecutive video frames. Therefore, video prediction is of interest not only in anticipating visual changes in the real world but has, above all, emerged as an unsupervised learning rule targeting the formation and dynamics of the observed environment. Many of the deep learning-based state-of-the-art models for video prediction utilize some form of recurrent layers like Long Short-Term Memory (LSTMs) or Gated Recurrent Units (GRUs) at the core of their models. Although these models can predict the future frames, they rely entirely on these recurrent structures to simultaneously perform three distinct tasks: extracting transformations, projecting them into the future, and transforming the current frame. In order to completely interpret the formed internal representations, it is crucial to disentangle these tasks. This paper proposes a fully differentiable building block that can perform all of those tasks separately while maintaining interpretability. We derive the relevant theoretical foundations and showcase results on synthetic as well as real data. We demonstrate that our method is readily extended to perform motion segmentation and account for the scene's composition, and learns to produce reliable predictions in an entirely interpretable manner by only observing unlabeled video data.
['Sven Behnke', 'Jan Nogga', 'Hafez Farazi']
2021-05-10
null
null
null
null
['motion-segmentation']
['computer-vision']
[ 4.17283654e-01 1.57707259e-01 -1.46104246e-01 -2.68312156e-01 -1.55801073e-01 -6.30914032e-01 8.26926887e-01 -4.47519839e-01 -1.43015862e-01 4.63180482e-01 3.36038619e-01 -1.55304030e-01 2.75485009e-01 -4.86575246e-01 -1.08237994e+00 -7.68008590e-01 -1.37788430e-01 1.69204146e-01 2.28573948e-01 6.56689554e-02 5.64489327e-02 5.85914612e-01 -1.65616786e+00 5.86526573e-01 2.44528845e-01 7.64765382e-01 5.02621114e-01 1.18869519e+00 1.87314406e-01 1.47350252e+00 -1.71346754e-01 -1.75940841e-01 7.18611255e-02 -5.83272219e-01 -8.90667200e-01 6.97602332e-01 4.84992303e-02 -4.96038347e-01 -7.56910563e-01 7.58878708e-01 -1.97899505e-01 2.51712650e-01 5.62077165e-01 -8.39415491e-01 -5.80864966e-01 3.18914354e-01 -3.17127883e-01 3.17545354e-01 4.62403297e-01 2.41796687e-01 9.98893321e-01 -7.36572206e-01 9.88449574e-01 9.61427808e-01 2.60823756e-01 6.60168111e-01 -1.13119876e+00 -9.99796540e-02 6.04071736e-01 6.39009058e-01 -7.84387112e-01 -5.56148589e-01 8.44639719e-01 -7.18405366e-01 1.00709450e+00 2.45881841e-01 8.22272360e-01 1.43376803e+00 5.21252394e-01 1.03981328e+00 5.98167181e-01 -3.33098829e-01 3.20738479e-02 -2.02367216e-01 -1.79552808e-01 7.14209497e-01 -2.78736740e-01 2.03437835e-01 -4.13399667e-01 3.93450737e-01 8.66361201e-01 2.86730677e-01 -5.93643725e-01 -3.32548082e-01 -1.51467681e+00 5.22182882e-01 3.26975465e-01 2.35527769e-01 -5.89715421e-01 4.23691839e-01 1.82497725e-01 3.34777743e-01 5.16722262e-01 1.34458035e-01 -4.70902592e-01 -2.01175004e-01 -7.47819364e-01 1.31990910e-01 6.05303586e-01 7.80882657e-01 7.83873320e-01 3.53328556e-01 3.93841416e-02 1.49792001e-01 2.79541165e-01 1.87213480e-01 6.51445746e-01 -1.17490458e+00 2.53353655e-01 3.46155614e-01 3.02843541e-01 -1.04238141e+00 -2.92616963e-01 -3.54411930e-01 -7.31623054e-01 2.05160871e-01 4.83054549e-01 -2.05547094e-01 -1.03248870e+00 1.93536198e+00 2.25466732e-02 6.39920175e-01 7.50496611e-02 8.79402816e-01 4.34199363e-01 1.12551737e+00 1.07177660e-01 -5.91739595e-01 9.23477709e-01 -9.76928115e-01 -6.88042700e-01 -3.42208534e-01 5.08073628e-01 -5.19954443e-01 5.79536498e-01 2.57915825e-01 -1.28624415e+00 -8.23551178e-01 -8.53838801e-01 -1.20581448e-01 -1.70306861e-01 1.05701298e-01 3.78049761e-01 -3.63980159e-02 -1.30681705e+00 7.35565960e-01 -1.34197617e+00 -4.83214289e-01 1.11778639e-01 4.43131447e-01 -4.28693563e-01 2.83866912e-01 -9.05218363e-01 8.40974331e-01 3.37449819e-01 4.11605835e-01 -1.06511807e+00 -1.79144427e-01 -9.46863532e-01 1.60441414e-01 3.01532090e-01 -9.13368762e-01 1.23685348e+00 -1.67946935e+00 -1.58606124e+00 7.36954093e-01 -6.67514384e-01 -7.22905636e-01 4.67972904e-01 -2.74114221e-01 -3.11217159e-01 1.53830126e-01 -2.71823704e-01 6.15762770e-01 1.04986513e+00 -1.17803299e+00 -7.13428736e-01 -3.72601092e-01 1.04626998e-01 3.36364746e-01 -3.82712409e-02 -8.60787928e-02 -4.40365940e-01 -7.75867224e-01 1.55662090e-01 -1.16411197e+00 -4.42461580e-01 -1.46126091e-01 -1.34052441e-01 5.29322475e-02 9.92167532e-01 -8.23657274e-01 1.09814227e+00 -1.94194770e+00 7.54966080e-01 -2.90914416e-01 2.47485831e-01 2.13360116e-01 -1.21206366e-01 2.59236604e-01 -3.21599215e-01 -7.78278410e-02 2.81572752e-02 -3.34185213e-01 -2.86828578e-01 4.42313015e-01 -8.07251632e-01 5.52834034e-01 3.26562196e-01 1.12962675e+00 -8.81434798e-01 -8.00291002e-02 4.86524463e-01 5.73311388e-01 -5.34618616e-01 4.86715853e-01 -5.50005853e-01 8.72039497e-01 -4.87135887e-01 9.87537503e-02 1.02447001e-02 -4.29810882e-01 4.40561622e-01 -1.32312641e-01 -1.73875436e-01 1.15721866e-01 -7.76820004e-01 1.54259121e+00 -2.65237778e-01 1.02793956e+00 -2.94717729e-01 -1.37727225e+00 5.81245601e-01 6.76041722e-01 5.66617191e-01 -4.07438576e-01 9.34603736e-02 -3.15437606e-03 6.92749396e-02 -8.69819283e-01 5.44447303e-01 -9.36447605e-02 1.55252233e-01 5.33837795e-01 1.62491933e-01 4.70260918e-01 1.14159785e-01 -2.39068065e-02 8.74574363e-01 7.53474355e-01 4.53868210e-01 1.26753956e-01 6.65004015e-01 -2.08360508e-01 6.94900393e-01 4.70544785e-01 -1.36220798e-01 7.63623953e-01 5.09092510e-01 -9.74701345e-01 -1.18982697e+00 -9.66117620e-01 4.45356101e-01 9.43445921e-01 2.05185086e-01 -1.89779386e-01 -6.71558321e-01 -3.99985433e-01 -6.08594239e-01 6.65357709e-01 -8.36785734e-01 -2.44442955e-01 -8.38647664e-01 -5.79549551e-01 -2.21824516e-02 5.18507659e-01 1.62658542e-01 -1.35820138e+00 -9.32064533e-01 3.99431437e-01 -2.57089436e-01 -1.45486140e+00 -1.62267402e-01 8.03130940e-02 -8.23242903e-01 -1.02421260e+00 -4.79422182e-01 -6.89295530e-01 5.79835415e-01 2.19197586e-01 1.07676363e+00 -8.91403332e-02 4.83428687e-02 5.96915483e-01 -2.30927646e-01 3.58092748e-02 -5.84782898e-01 -1.88112378e-01 1.10008515e-01 5.38385987e-01 1.59410864e-01 -7.44147301e-01 -5.25623083e-01 1.20577365e-01 -9.72883284e-01 5.63232005e-01 2.36150354e-01 7.41171360e-01 5.50216794e-01 -1.44018322e-01 1.66230947e-01 -8.46656322e-01 -1.22857327e-02 -4.88997102e-01 -4.74062383e-01 4.48220849e-01 7.24373711e-03 1.45795986e-01 9.70978200e-01 -3.87545913e-01 -1.39970398e+00 3.40352714e-01 3.24250348e-02 -7.19738245e-01 -3.51012379e-01 4.09062296e-01 -5.18255979e-02 2.71192938e-01 2.40859956e-01 5.40323973e-01 -1.48434013e-01 -1.85372010e-01 3.38915318e-01 2.04467103e-01 6.46174014e-01 -2.71792680e-01 7.79692590e-01 7.54095018e-01 -7.14702010e-02 -8.51784289e-01 -8.60072970e-01 -1.30645499e-01 -1.08122754e+00 -5.56638718e-01 1.20652485e+00 -9.81842518e-01 -7.13775218e-01 4.71243232e-01 -1.46734202e+00 -4.15912539e-01 -3.60469818e-01 4.27330703e-01 -9.22590911e-01 2.92018086e-01 -7.37942815e-01 -7.57491231e-01 8.59726444e-02 -1.28442287e+00 8.24031293e-01 2.54601479e-01 -2.82814711e-01 -1.36590779e+00 1.19845301e-01 2.16548175e-01 -6.61366805e-02 4.09619182e-01 7.87703395e-01 -3.00253123e-01 -9.76108670e-01 -6.84014857e-02 1.65379152e-01 1.53270930e-01 2.34173506e-01 4.22141403e-01 -9.52672958e-01 -1.13515794e-01 3.93473864e-01 1.05388507e-01 9.85654294e-01 5.60909152e-01 1.23232508e+00 -3.65028799e-01 -2.58001357e-01 8.19705963e-01 1.19078159e+00 4.59240198e-01 6.68802679e-01 1.97320238e-01 9.28560078e-01 7.48500228e-01 3.11902881e-01 4.00196224e-01 3.64185929e-01 6.33454502e-01 5.30913889e-01 1.08123101e-01 8.16461816e-02 -3.16248238e-01 8.65259528e-01 1.02318227e+00 -5.64715266e-01 -4.25212085e-01 -7.55963683e-01 3.67597014e-01 -2.21523762e+00 -1.45875490e+00 1.02408901e-02 2.00526714e+00 3.02414387e-01 2.27615610e-03 -7.00052604e-02 -7.84498304e-02 7.05094039e-01 5.41581690e-01 -8.16112638e-01 -4.03338134e-01 -1.69974685e-01 -1.71152025e-01 1.60564169e-01 5.59111297e-01 -1.16720617e+00 1.04335630e+00 6.30067110e+00 2.00374991e-01 -1.38370025e+00 -1.59194440e-01 1.11061454e+00 -3.88848484e-02 -2.66399235e-01 1.47020116e-01 -4.51044470e-01 3.01043928e-01 1.06583703e+00 -9.77994651e-02 5.60916603e-01 6.55366778e-01 6.25454664e-01 2.47912426e-02 -1.33119464e+00 9.00041759e-01 6.19795397e-02 -1.55478907e+00 3.77825111e-01 1.19912475e-02 7.40669966e-01 9.06666517e-02 1.89554274e-01 1.06569240e-02 1.53498501e-01 -1.06766176e+00 9.63216960e-01 7.89863825e-01 4.84616131e-01 -3.45566899e-01 3.50305855e-01 5.17584741e-01 -1.22182739e+00 -2.10976660e-01 -2.38571122e-01 -5.48302591e-01 4.80004042e-01 7.19424561e-02 -4.79224384e-01 3.51732492e-01 4.03377384e-01 1.39125025e+00 -4.53395844e-01 4.91516322e-01 -2.62812078e-01 5.01653969e-01 5.11233583e-02 3.33819002e-01 3.96226823e-01 -4.34965521e-01 7.23824859e-01 9.59641755e-01 4.62256968e-01 2.47383952e-01 3.95747423e-02 7.68333733e-01 8.29952955e-02 -3.82739693e-01 -7.15867400e-01 -1.26889050e-01 -2.74630547e-01 1.07155323e+00 -7.41968572e-01 -4.01852697e-01 -6.24474883e-01 1.15714574e+00 3.58798444e-01 8.78652155e-01 -1.02218151e+00 5.05976439e-01 8.76445293e-01 1.04414396e-01 4.59793955e-01 -3.70477825e-01 -8.45722482e-03 -1.65560365e+00 4.94805500e-02 -6.77005768e-01 1.71594068e-01 -8.99877548e-01 -8.08991134e-01 8.54669034e-01 -3.14608544e-01 -1.44968653e+00 -8.78680527e-01 -6.85422421e-01 -6.22063398e-01 4.49043185e-01 -1.41473746e+00 -1.16891146e+00 -1.55346513e-01 6.60204589e-01 9.20426190e-01 -6.72432706e-02 6.40081882e-01 -7.56108016e-02 -6.97024405e-01 -1.18929043e-01 1.19378284e-01 1.48684531e-01 1.06437571e-01 -1.07184315e+00 4.43734407e-01 1.19883728e+00 6.41816318e-01 3.07980120e-01 8.72624636e-01 -4.17624146e-01 -1.39682162e+00 -1.12642395e+00 8.17250192e-01 -5.19492447e-01 8.32706630e-01 -3.59523654e-01 -9.57193196e-01 1.37860787e+00 1.55040368e-01 2.19141334e-01 4.33355898e-01 -4.28496897e-01 -2.36657672e-02 -8.42311699e-03 -3.16622734e-01 9.63086605e-01 1.13570833e+00 -6.52464449e-01 -5.21510839e-01 2.66970575e-01 6.08591616e-01 -4.21580553e-01 -4.50208664e-01 3.47206444e-01 6.09421015e-01 -1.44044518e+00 9.17929649e-01 -9.91866708e-01 7.42572427e-01 -2.96912402e-01 -9.78044719e-02 -9.92915690e-01 -4.34896797e-01 -8.52473795e-01 -4.81698304e-01 7.55315423e-01 2.42782071e-01 -3.77655625e-01 9.02496457e-01 8.38065147e-01 -7.47710094e-02 -7.24946976e-01 -6.90559804e-01 -1.99570596e-01 -2.84463018e-01 -5.31403244e-01 1.76391199e-01 7.48781085e-01 -1.08014740e-01 3.55003625e-01 -7.80659258e-01 2.19479799e-01 2.70386815e-01 3.98140401e-01 7.01083302e-01 -8.27679932e-01 -4.33464825e-01 -2.26974934e-01 -6.69233739e-01 -1.38028288e+00 5.51799715e-01 -4.84181464e-01 4.40088958e-02 -1.35115516e+00 1.30741790e-01 2.42432192e-01 -2.68842489e-01 7.45345131e-02 -9.57788527e-02 8.22373033e-02 3.72800529e-01 4.05534655e-01 -7.05673873e-01 6.97102487e-01 1.24127924e+00 2.21558958e-02 -1.73970804e-01 1.84214875e-01 -1.78627044e-01 1.19532549e+00 5.33214629e-01 -5.93607239e-02 -5.87570667e-01 -7.91926920e-01 2.09907711e-01 4.54709798e-01 4.92291838e-01 -9.65480268e-01 2.28639558e-01 -3.50344777e-01 6.33975089e-01 -3.59780401e-01 5.33249080e-01 -8.50263715e-01 5.13649106e-01 4.85842347e-01 -4.53767985e-01 3.15742314e-01 -2.37596095e-01 7.68087626e-01 -4.78507251e-01 2.01663196e-01 5.56658626e-01 -4.39864069e-01 -1.18136609e+00 5.01565814e-01 -8.17976058e-01 -2.74594456e-01 1.26959670e+00 -5.06627023e-01 -4.63111475e-02 -7.33629942e-01 -1.20634878e+00 -3.27400938e-02 5.18881023e-01 5.17500997e-01 6.79651618e-01 -1.02455723e+00 -4.47346956e-01 2.44237393e-01 -2.87809908e-01 -1.01836585e-01 4.03360397e-01 7.28627145e-01 -5.99878609e-01 5.73554695e-01 -2.95645475e-01 -7.60660231e-01 -1.10909998e+00 6.76910460e-01 4.48202103e-01 -3.52672219e-01 -6.98571563e-01 6.20448232e-01 8.73511374e-01 2.22117454e-02 -2.53157225e-02 -4.10890549e-01 -5.01739860e-01 1.45058427e-02 5.43473244e-01 3.00652510e-03 -4.68321234e-01 -1.19068646e+00 7.02869520e-02 5.68754196e-01 4.37538400e-02 6.36120737e-02 1.48618972e+00 -5.27030647e-01 -9.00550783e-02 8.03461075e-01 1.18567979e+00 -5.58571696e-01 -1.87062716e+00 -1.34146869e-01 8.21713880e-02 -2.69221604e-01 -3.42724353e-01 -1.20454691e-01 -1.23268914e+00 1.02787900e+00 1.42668664e-01 1.68381751e-01 1.28157473e+00 -2.89789718e-02 7.88163483e-01 2.88649976e-01 2.70151049e-01 -7.76224792e-01 1.15601152e-01 5.12055874e-01 6.92016780e-01 -1.10637915e+00 -2.04999357e-01 -2.29737058e-01 -6.78388059e-01 1.43603754e+00 5.09713113e-01 -2.93656260e-01 5.71211398e-01 5.49466722e-02 -1.02943145e-01 7.37054050e-02 -1.29571879e+00 -6.77564368e-02 3.28177601e-01 3.64789009e-01 4.97900218e-01 -8.64439234e-02 2.88695395e-01 2.80727535e-01 6.58305511e-02 -1.64624956e-03 7.51377642e-01 5.86266398e-01 -1.93664998e-01 -9.20529127e-01 -1.21593527e-01 2.84410775e-01 -4.02256638e-01 1.97854459e-01 -1.09246813e-01 5.78101456e-01 1.73164979e-02 5.84711850e-01 3.03340763e-01 -4.64568287e-01 -1.02325581e-01 1.78693533e-02 4.75409329e-01 -5.03962278e-01 -2.26559833e-01 2.02354878e-01 -2.19102457e-01 -8.52227211e-01 -8.93764198e-01 -9.02984500e-01 -1.19379771e+00 -1.83534652e-01 1.39363408e-01 -1.80827782e-01 1.95000455e-01 1.37334025e+00 2.67003089e-01 6.15258217e-01 6.40065074e-01 -1.23457122e+00 -9.10408944e-02 -5.34593225e-01 -3.24558586e-01 6.64390266e-01 6.93293273e-01 -4.40872341e-01 -1.71216503e-01 8.95622909e-01]
[8.600727081298828, 0.28991633653640747]
b7c93619-8996-49ee-a51e-6b49ac095a8f
person-search-challenges-and-solutions-a
2105.01605
null
https://arxiv.org/abs/2105.01605v1
https://arxiv.org/pdf/2105.01605v1.pdf
Person Search Challenges and Solutions: A Survey
Person search has drawn increasing attention due to its real-world applications and research significance. Person search aims to find a probe person in a gallery of scene images with a wide range of applications, such as criminals search, multicamera tracking, missing person search, etc. Early person search works focused on image-based person search, which uses person image as the search query. Text-based person search is another major person search category that uses free-form natural language as the search query. Person search is challenging, and corresponding solutions are diverse and complex. Therefore, systematic surveys on this topic are essential. This paper surveyed the recent works on image-based and text-based person search from the perspective of challenges and solutions. Specifically, we provide a brief analysis of highly influential person search methods considering the three significant challenges: the discriminative person features, the query-person gap, and the detection-identification inconsistency. We summarise and compare evaluation results. Finally, we discuss open issues and some promising future research directions.
['Alex Hauptmann', 'Xiaojun Chang', 'Yun Xiao', 'Pengzhen Ren', 'Xiangtan Lin']
2021-05-01
null
null
null
null
['person-search']
['computer-vision']
[ 4.46449667e-02 -8.04924130e-01 -2.38936961e-01 -2.83753812e-01 -7.16647208e-01 -6.97391748e-01 8.03593874e-01 -2.03842465e-02 -8.23790193e-01 5.14656186e-01 2.21591845e-01 1.95547119e-01 -1.97352022e-01 -5.57924032e-01 1.43541709e-01 -6.11333609e-01 3.94892216e-01 7.04391181e-01 1.69367954e-01 1.05281798e-02 3.96669209e-01 4.75973785e-01 -1.61934781e+00 -2.25242138e-01 6.17215872e-01 6.49329245e-01 7.35968873e-02 4.87034172e-01 1.24945045e-01 7.30943233e-02 -1.00513697e+00 -7.76297033e-01 2.04343989e-01 -3.76892745e-01 -8.87071848e-01 -7.53538171e-03 7.12805569e-01 -3.92563462e-01 -6.83419287e-01 1.18508732e+00 1.30786276e+00 3.40245336e-01 4.74959284e-01 -1.45365071e+00 -7.81274199e-01 -7.72141516e-02 -8.06728482e-01 6.89745188e-01 9.92478669e-01 1.96938202e-01 5.73508739e-01 -9.16945159e-01 3.62887889e-01 1.67038226e+00 5.56984842e-01 7.05196142e-01 -6.92368388e-01 -8.16490114e-01 3.38860299e-03 5.59469998e-01 -1.89731324e+00 -3.62979680e-01 3.70880574e-01 -4.69069332e-01 6.46156073e-01 6.40480459e-01 8.07988107e-01 9.73518550e-01 -1.92345083e-01 1.10090482e+00 9.49797690e-01 -5.33975482e-01 -3.71732175e-01 3.92999470e-01 4.95760053e-01 7.05815554e-01 4.99121726e-01 2.08517581e-01 -6.12528861e-01 -6.48855448e-01 6.22328818e-01 1.16595358e-01 -1.93616748e-01 -1.35810465e-01 -1.16783595e+00 7.58598149e-01 3.19011897e-01 3.03415537e-01 -7.30130449e-02 -2.45050728e-01 4.92706567e-01 -5.97843304e-02 6.73207641e-02 3.17661196e-01 3.36842090e-01 -8.93259887e-03 -1.14199495e+00 7.59412825e-01 5.70603251e-01 8.43513012e-01 5.34040511e-01 -4.18089062e-01 -7.03289390e-01 1.02981436e+00 2.35340983e-01 1.03092122e+00 6.08462214e-01 -2.97673166e-01 4.11310971e-01 6.96457863e-01 2.53406435e-01 -1.50218928e+00 -2.87148923e-01 -4.96241719e-01 -9.22732890e-01 -4.56027776e-01 3.10396940e-01 1.11055143e-01 -7.19950378e-01 1.21909249e+00 5.18089354e-01 -2.09860653e-01 -2.82917589e-01 1.30764449e+00 1.53580010e+00 3.36662561e-01 1.51257873e-01 -2.30132211e-02 2.08719778e+00 -1.01602495e+00 -6.29335880e-01 -5.71251333e-01 -5.99518791e-02 -8.72151315e-01 7.19436705e-01 -2.29954839e-01 -9.58317101e-01 -7.14929581e-01 -4.90305513e-01 -1.06899731e-01 -6.39704287e-01 4.12097514e-01 4.33230400e-01 9.91113842e-01 -8.85460794e-01 -2.30236366e-01 -2.34286021e-02 -1.01552212e+00 2.80014664e-01 4.40043867e-01 -2.76824951e-01 -2.46549740e-01 -1.31718171e+00 6.82538390e-01 2.11633697e-01 2.09356412e-01 -3.00047755e-01 -1.51264802e-01 -7.82827020e-01 -1.26480520e-01 4.42478627e-01 -1.01315033e+00 9.13277328e-01 -2.47296870e-01 -6.47998154e-01 1.67446172e+00 -6.28628373e-01 -1.04359232e-01 7.04169214e-01 -6.33738711e-02 -7.44579077e-01 2.53467679e-01 8.22114587e-01 4.73938435e-01 7.92641997e-01 -8.41824055e-01 -8.66154373e-01 -7.28698134e-01 -3.01722437e-01 4.07317936e-01 -1.79229066e-01 8.77968311e-01 -1.21665752e+00 -7.21356213e-01 -1.92531925e-02 -9.43592727e-01 -8.96302685e-02 -2.07808375e-01 -4.83706057e-01 -7.20923364e-01 7.27633417e-01 -6.66825354e-01 1.45787692e+00 -1.92052722e+00 -2.09621415e-01 3.34183186e-01 1.64475158e-01 3.77548814e-01 1.28334597e-01 5.58420241e-01 2.12885097e-01 9.92115438e-02 2.28350908e-01 -2.99524665e-01 6.61062077e-02 -4.11241651e-01 -1.94154784e-01 6.10732734e-01 -5.82789004e-01 1.24645114e+00 -7.59248912e-01 -1.14610457e+00 2.59020567e-01 2.53718287e-01 2.18578726e-01 6.63652718e-02 6.69961214e-01 3.65583867e-01 -7.99438834e-01 1.19200945e+00 6.05712831e-01 -1.98359922e-01 -4.74082440e-01 -2.28968039e-01 -1.64748415e-01 -5.72047591e-01 -1.14967334e+00 1.24448800e+00 3.01923662e-01 5.27279437e-01 1.45097211e-01 -5.89722872e-01 8.40528607e-01 3.28597166e-02 4.85067278e-01 -8.08508813e-01 -9.38647166e-02 2.46034890e-01 -2.79170305e-01 -6.30154073e-01 9.37801659e-01 3.79673749e-01 -4.70881760e-01 6.28928065e-01 -4.91838604e-01 3.00890118e-01 4.47261274e-01 2.26711646e-01 5.87738752e-01 -3.39662850e-01 6.63850963e-01 -2.67362922e-01 1.05256021e+00 1.06367823e-02 1.72413275e-01 1.16854930e+00 -7.55614102e-01 5.04460156e-01 -4.01845753e-01 -7.45328009e-01 -5.52175760e-01 -6.29875124e-01 -2.08669789e-02 1.10366416e+00 8.17808151e-01 -3.91436100e-01 -7.39306450e-01 -4.60339814e-01 9.68026966e-02 -2.11332858e-01 -5.29026747e-01 4.15979140e-02 -7.76579320e-01 -9.04546916e-01 7.76410103e-01 7.57194087e-02 1.18587208e+00 -1.26831686e+00 -5.85708916e-01 -1.41396582e-01 -6.44269407e-01 -1.06978536e+00 -1.11232758e+00 -8.40016901e-01 -1.77158996e-01 -1.28319931e+00 -1.68533909e+00 -1.21465480e+00 7.60941207e-01 7.63320386e-01 9.49829102e-01 4.27246153e-01 -8.47301781e-01 9.57413912e-01 -6.91508576e-02 -4.33929235e-01 3.27118546e-01 2.63551086e-01 3.42431337e-01 9.49670970e-02 1.05911589e+00 2.35586226e-01 -1.06623542e+00 8.42103362e-01 -4.51113194e-01 -4.74310905e-01 4.10673589e-01 7.86227167e-01 4.89969015e-01 2.42748797e-01 1.46707758e-01 -2.50500858e-01 1.10719872e+00 -1.76581159e-01 -3.21831524e-01 6.84392571e-01 -5.34465313e-01 -4.83286977e-01 -1.74334541e-01 -4.01399314e-01 -8.14606428e-01 -1.59534454e-01 5.35878353e-02 5.25158979e-02 -2.46097699e-01 5.86917885e-02 -1.39467359e-01 -4.31755394e-01 6.64806664e-01 7.40095913e-01 -3.06859642e-01 -4.32389855e-01 -2.21971244e-01 8.95158172e-01 7.52902567e-01 -5.21280050e-01 8.67804646e-01 6.12486124e-01 -4.10979599e-01 -9.76114035e-01 -6.00484312e-01 -1.45551121e+00 -5.49589753e-01 -5.54042637e-01 9.47051764e-01 -9.39672887e-01 -1.06390309e+00 7.24026144e-01 -1.21362484e+00 3.64173263e-01 1.44838169e-01 3.26838911e-01 -6.90625980e-02 7.94740558e-01 -3.63360673e-01 -1.06691682e+00 -8.64892066e-01 -1.03003347e+00 1.48345709e+00 8.28781188e-01 -3.04453760e-01 -6.40038669e-01 -1.25665506e-02 7.88585126e-01 1.66002318e-01 -3.07732783e-02 1.78778931e-01 -5.99537015e-01 -5.94825089e-01 -7.68679559e-01 -5.75755656e-01 -5.91090620e-01 1.78863540e-01 -8.06559980e-01 -7.92961657e-01 -6.95312858e-01 -2.89582044e-01 -4.07702960e-02 9.14030731e-01 3.19210112e-01 8.71663868e-01 -9.09607038e-02 -1.10258162e+00 3.67488533e-01 1.13848865e+00 2.52306402e-01 5.22122860e-01 5.55637300e-01 6.84194088e-01 6.72196031e-01 6.36670113e-01 2.90636003e-01 3.66908669e-01 9.60807681e-01 -3.58625412e-01 -4.10442166e-02 -8.73685926e-02 -3.33550125e-01 -2.38442019e-01 -4.83021662e-02 -3.42367321e-01 -4.67250228e-01 -9.96510983e-01 5.87810636e-01 -1.96097553e+00 -1.54243004e+00 3.99160711e-03 2.03386331e+00 4.56021160e-01 -3.87197435e-01 6.97791517e-01 4.88300668e-03 1.30789101e+00 1.51640311e-01 -5.48579991e-01 4.13399488e-01 -2.59399235e-01 -2.73587048e-01 3.05834293e-01 1.93207875e-01 -1.36896813e+00 9.64681149e-01 6.86321354e+00 1.12247777e+00 -5.73815763e-01 6.12596832e-02 3.70633841e-01 1.31760210e-01 3.71570170e-01 -3.55195343e-01 -1.62074566e+00 6.99266016e-01 -9.08738822e-02 -4.58916068e-01 7.15831444e-02 7.27581739e-01 1.00572594e-01 -4.51769829e-01 -9.20707643e-01 2.10476661e+00 6.31557465e-01 -9.16991293e-01 1.48023471e-01 3.23138654e-01 5.30273020e-01 -6.22124493e-01 1.82860315e-01 2.34080046e-01 -2.72752553e-01 -1.05927062e+00 5.39333701e-01 6.09118521e-01 8.38003993e-01 -6.21952355e-01 8.38010192e-01 4.67604697e-01 -1.60414660e+00 -3.23131010e-02 -3.10761601e-01 2.00149462e-01 4.97695446e-01 1.58899710e-01 -3.52774203e-01 3.61063391e-01 1.12839878e+00 4.17735308e-01 -8.54568779e-01 1.55599308e+00 2.33095765e-01 -7.04898906e-04 -4.59541380e-01 -3.38476002e-01 -1.14570996e-02 -1.29322499e-01 7.47154355e-01 1.42035890e+00 1.89310610e-01 2.92972803e-01 7.26803899e-01 7.15536356e-01 3.36374998e-01 1.75692976e-01 -4.22001779e-01 2.22598225e-01 5.18523753e-01 1.08247042e+00 -8.34531903e-01 -2.98622578e-01 -2.19994843e-01 1.29861641e+00 -1.50314316e-01 5.12744963e-01 -4.89920735e-01 -3.16870362e-01 4.38006222e-01 2.34193712e-01 -8.34972709e-02 -6.86612278e-02 1.85073137e-01 -1.04173994e+00 1.69826373e-01 -9.62160587e-01 1.01144493e+00 -6.32969081e-01 -1.37186503e+00 3.78740013e-01 2.91792601e-01 -1.12407303e+00 -3.07677090e-01 -2.30147481e-01 -5.27446926e-01 1.23281014e+00 -1.16060102e+00 -1.35034204e+00 -5.82403660e-01 8.60005975e-01 6.68909788e-01 -7.21839249e-01 5.77086091e-01 5.25416672e-01 -6.45901620e-01 1.09606266e+00 -4.36170250e-01 6.59432113e-01 7.83634484e-01 -5.71781278e-01 5.12022495e-01 8.37467313e-01 5.33576123e-02 1.07975388e+00 5.64132631e-01 -9.43705559e-01 -1.25285602e+00 -6.25251234e-01 1.20126104e+00 -6.42751694e-01 4.57580760e-02 -1.65216416e-01 -3.96996289e-01 2.51329809e-01 3.75061482e-02 -2.17618257e-01 6.12310171e-01 7.56509695e-03 -3.13616917e-02 2.81202346e-02 -1.22233653e+00 7.33168006e-01 1.31279850e+00 -6.38247848e-01 -2.53867418e-01 3.77712309e-01 -3.07608899e-02 -3.28582317e-01 -2.51016289e-01 5.52832372e-02 7.78945267e-01 -8.74936104e-01 1.53124881e+00 -3.07708327e-03 -6.14146054e-01 -3.21981609e-01 2.23668784e-01 -6.91309333e-01 -5.71037233e-01 -5.82180977e-01 6.22246750e-02 1.22092509e+00 -1.88530251e-01 -6.14321649e-01 9.64633167e-01 8.90198946e-01 6.40112400e-01 -3.58281225e-01 -7.66425610e-01 -8.15973520e-01 -4.95495349e-01 9.24142823e-02 7.34332085e-01 5.11600614e-01 -1.26629204e-01 4.69715953e-01 -6.96219027e-01 4.77210917e-02 9.89964664e-01 3.47392023e-01 9.91382837e-01 -1.22130358e+00 -4.98033389e-02 -4.94867146e-01 -5.72829604e-01 -1.39418983e+00 -6.03705496e-02 -5.94311655e-01 -1.60219148e-01 -1.66794837e+00 9.50351596e-01 -3.20191383e-01 2.24539489e-01 1.01257212e-01 -5.25824726e-01 3.80955875e-01 2.65019596e-01 7.74153948e-01 -7.80584216e-01 1.95792273e-01 1.06707931e+00 -5.36057115e-01 -1.35842994e-01 4.27372545e-01 -7.44186997e-01 4.83872920e-01 6.92580521e-01 -7.81834126e-02 -9.66251269e-02 -4.96573478e-01 1.11444190e-01 -3.04675490e-01 8.51093829e-01 -8.20803881e-01 9.67978776e-01 -5.13225868e-02 6.14685714e-01 -1.08465338e+00 6.25704646e-01 -6.06204987e-01 2.98716892e-02 4.92713749e-01 -5.80112077e-02 2.64023870e-01 -1.69179723e-01 6.48731589e-01 -3.59677434e-01 -3.12069297e-01 6.05104923e-01 -5.90810537e-01 -1.27364504e+00 7.60046244e-01 -7.78814182e-02 1.19786352e-01 1.04888201e+00 -7.92749465e-01 -1.93432689e-01 -5.23928165e-01 -4.54549342e-01 5.92061222e-01 1.90138459e-01 5.00385225e-01 7.93365359e-01 -1.41002941e+00 -9.28215265e-01 2.32910335e-01 3.74078393e-01 -5.15830457e-01 3.79815131e-01 4.00679320e-01 -1.73722029e-01 7.82565355e-01 1.82985678e-01 -6.74463153e-01 -1.98988986e+00 5.97571194e-01 4.01963532e-01 -1.31122217e-01 -4.04238790e-01 9.06240284e-01 4.04860348e-01 -2.87994146e-01 4.13710445e-01 7.23789155e-01 -5.64912856e-01 8.67361352e-02 1.05768478e+00 6.12407386e-01 -4.77273852e-01 -1.23221362e+00 -8.79522324e-01 1.14597237e+00 -6.19605533e-04 -1.63771123e-01 5.22742033e-01 -5.15949845e-01 -1.09784998e-01 -2.67342299e-01 9.23445284e-01 7.24959699e-03 -3.19984466e-01 -5.63236475e-01 2.98896357e-02 -7.80010998e-01 -5.40167570e-01 -6.88414216e-01 -7.40221620e-01 5.75986087e-01 9.04251456e-01 2.51415342e-01 8.19274664e-01 4.08379257e-01 8.17923427e-01 3.77078354e-01 8.09732676e-01 -1.34168661e+00 1.13368087e-01 3.11424047e-01 1.00657237e+00 -1.61314046e+00 5.06739736e-01 -5.95949829e-01 -2.58961827e-01 6.65271878e-01 6.53751373e-01 3.32512528e-01 5.89256406e-01 -3.56448472e-01 -1.03200123e-01 -3.68394554e-01 2.89056957e-01 -5.77650785e-01 7.25657940e-01 8.22464645e-01 2.94580148e-03 -1.01930998e-01 -3.90275449e-01 5.79219699e-01 -5.04942238e-01 -1.80081502e-01 -4.32394415e-01 7.27419198e-01 -3.58038932e-01 -9.25559878e-01 -1.06531715e+00 4.16039199e-01 -3.76246333e-01 -8.15375224e-02 -6.97093010e-01 6.49985909e-01 2.38651827e-01 1.31886268e+00 -1.90966547e-01 -1.12116881e-01 4.82765853e-01 -2.52152890e-01 4.32943881e-01 -3.94246936e-01 -5.44157743e-01 -1.73641667e-01 -1.43891186e-01 -2.88505822e-01 -7.63620913e-01 -7.86581874e-01 -5.34247339e-01 -5.44578791e-01 -3.09798896e-01 3.34298760e-01 2.76762664e-01 8.62747729e-01 1.08726300e-01 -3.81258845e-01 1.77508220e-01 -5.75588822e-01 -2.00340942e-01 -7.12536454e-01 -4.61299032e-01 3.48899424e-01 2.17601493e-01 -5.82783282e-01 -3.19520831e-02 -4.70581464e-02]
[14.762056350708008, 0.8365119695663452]
751ae346-93b9-42db-8350-7c4c453888cd
twitter-spam-detection-a-systematic-review
2011.14754
null
https://arxiv.org/abs/2011.14754v2
https://arxiv.org/pdf/2011.14754v2.pdf
Twitter Spam Detection: A Systematic Review
Nowadays, with the rise of Internet access and mobile devices around the globe, more people are using social networks for collaboration and receiving real-time information. Twitter, the microblogging that is becoming a critical source of communication and news propagation, has grabbed the attention of spammers to distract users. So far, researchers have introduced various defense techniques to detect spams and combat spammer activities on Twitter. To overcome this problem, in recent years, many novel techniques have been offered by researchers, which have greatly enhanced the spam detection performance. Therefore, it raises a motivation to conduct a systematic review about different approaches of spam detection on Twitter. This review focuses on comparing the existing research techniques on Twitter spam detection systematically. Literature review analysis reveals that most of the existing methods rely on Machine Learning-based algorithms. Among these Machine Learning algorithms, the major differences are related to various feature selection methods. Hence, we propose a taxonomy based on different feature selection methods and analyses, namely content analysis, user analysis, tweet analysis, network analysis, and hybrid analysis. Then, we present numerical analyses and comparative studies on current approaches, coming up with open challenges that help researchers develop solutions in this topic.
['Ebrahim Mahdipour', 'Mohammad Akbari', 'Mostafa Haghi Kashani', 'Sepideh Bazzaz Abkenar']
2020-11-30
null
null
null
null
['spam-detection']
['natural-language-processing']
[-1.48079082e-01 -5.59889734e-01 -5.27167499e-01 1.74505889e-01 5.66668138e-02 -3.36478710e-01 6.86414301e-01 5.16784847e-01 -3.32372516e-01 6.03018641e-01 1.75936341e-01 -3.29896122e-01 -7.72095844e-02 -1.10229850e+00 3.49110216e-01 -6.03217363e-01 8.29844847e-02 1.65052891e-01 6.71779037e-01 -5.48328042e-01 8.65743935e-01 5.06693244e-01 -1.72559893e+00 -1.69786550e-02 1.15783787e+00 7.79849291e-01 -6.65866071e-03 1.12432509e-03 -5.87407708e-01 6.03454113e-01 -7.03877449e-01 -4.64355052e-01 -1.26920581e-01 -5.62424123e-01 -5.30162930e-01 -3.21602225e-02 -3.43992800e-01 -8.85869861e-02 -4.62980986e-01 1.30434752e+00 3.53071988e-01 1.53540015e-01 4.51919109e-01 -1.49438202e+00 -3.40683967e-01 5.89971125e-01 -8.00096691e-01 4.32365000e-01 5.19164741e-01 -2.31434435e-01 6.99478149e-01 -5.79243064e-01 2.93851346e-01 1.35167956e+00 5.59124827e-01 3.03346246e-01 -5.76405168e-01 -1.02559578e+00 2.20450476e-01 5.37346303e-01 -1.10284555e+00 -9.13427328e-04 7.63809502e-01 -2.29624331e-01 3.83160055e-01 5.63085556e-01 7.71759748e-01 7.27589190e-01 2.72812426e-01 9.01667953e-01 1.10280907e+00 -2.91504532e-01 -3.10017951e-02 5.38502812e-01 5.30250609e-01 4.29918200e-01 6.04079902e-01 -2.45676026e-01 -3.58489245e-01 -7.80637264e-01 1.49930924e-01 2.64555216e-01 -1.44039258e-01 2.07642779e-01 -5.80872059e-01 1.14292276e+00 2.74609894e-01 5.50074518e-01 -6.27958477e-02 -4.90134835e-01 7.98820555e-01 3.93718302e-01 8.31306577e-01 2.00210512e-01 -6.35622218e-02 -1.48211822e-01 -8.00229669e-01 2.99378514e-01 9.73713398e-01 4.77745712e-01 7.02542782e-01 2.20237058e-02 2.86316901e-01 6.97926700e-01 4.91940498e-01 5.99944055e-01 1.03649342e+00 -2.10247397e-01 3.20606709e-01 1.07101679e+00 -1.93348490e-02 -1.87356043e+00 -4.57181752e-01 -3.17726791e-01 -8.72131228e-01 -2.74097651e-01 1.88470706e-01 -1.05652317e-01 -1.97973043e-01 1.01360559e+00 4.26007658e-01 4.03540909e-01 -5.38202167e-01 6.83657825e-01 8.94084513e-01 5.53212285e-01 5.01145385e-02 -6.00210369e-01 1.29890358e+00 -8.04523945e-01 -9.36787724e-01 1.73816741e-01 4.59322095e-01 -9.95647430e-01 7.80905187e-01 4.98236954e-01 -6.58749998e-01 -1.29003823e-01 -7.84771919e-01 5.87623060e-01 -7.77158737e-01 -3.12436640e-01 6.94049478e-01 1.11355352e+00 -4.41228658e-01 6.39709115e-01 -6.68070912e-01 -7.00181484e-01 3.41509968e-01 3.35142344e-01 3.66767496e-02 2.85029411e-01 -1.50311732e+00 8.49327326e-01 1.25916302e-01 -1.64458156e-01 -6.37270138e-02 8.29679798e-03 -2.35825002e-01 -1.27418727e-01 3.74971122e-01 -3.77199739e-01 1.10192442e+00 -1.12432683e+00 -1.31752336e+00 6.71633363e-01 -2.33985081e-01 -5.40018260e-01 4.56177592e-01 -6.34054467e-02 -8.43605101e-01 6.95416555e-02 2.32175127e-01 -4.91765559e-01 1.03935254e+00 -9.30038035e-01 -1.00257671e+00 -4.05242234e-01 -3.60723555e-01 5.87004889e-03 -8.42912734e-01 6.98326647e-01 -1.70964479e-01 -7.23466933e-01 3.23298216e-01 -7.95570076e-01 -2.12216675e-01 -7.28178740e-01 -5.21971941e-01 -4.55323011e-01 1.39664865e+00 -3.05949688e-01 1.94656491e+00 -1.86673295e+00 -2.15175956e-01 5.43936968e-01 3.87995481e-01 5.99036455e-01 4.77975756e-01 8.55485976e-01 2.86983788e-01 3.39351058e-01 -7.88463801e-02 3.73593941e-02 -4.22814459e-01 -1.79926872e-01 -6.30338788e-01 7.40412772e-01 -4.18432325e-01 4.55759883e-01 -9.90939021e-01 -5.25946736e-01 3.45534027e-01 5.82665950e-02 -8.27285647e-02 5.73617928e-02 1.87400341e-01 2.35080868e-01 -1.00916123e+00 7.85702407e-01 8.58890891e-01 -2.58206308e-01 1.50852352e-01 1.49635196e-01 -3.36225539e-01 2.59902656e-01 -1.06860757e+00 4.28595245e-01 -2.31923923e-01 3.93751025e-01 1.91376865e-01 -1.11789620e+00 9.27617848e-01 1.03739172e-01 5.19649029e-01 -2.94796050e-01 7.17725992e-01 6.85955703e-01 -1.12158358e-01 -5.97587347e-01 4.24129099e-01 9.18169096e-02 7.45780841e-02 5.80252767e-01 -6.68652356e-01 1.47601366e-01 4.69631970e-01 3.87608290e-01 9.75221813e-01 -6.21193290e-01 5.46322763e-01 -1.68174803e-01 1.05049825e+00 5.30811548e-02 1.23459533e-01 6.14287198e-01 -5.06472409e-01 7.34514743e-02 3.49358469e-01 -2.92195439e-01 -3.46529990e-01 -7.09931433e-01 -1.12057246e-01 9.64606643e-01 6.19944692e-01 -7.13982046e-01 -7.94503033e-01 -8.22382212e-01 3.16177458e-01 2.82109052e-01 -3.19129586e-01 -1.25302270e-01 -7.79357970e-01 -1.23282182e+00 4.08177912e-01 -1.57799274e-01 7.74680614e-01 -1.03509378e+00 -6.15592301e-02 3.11585069e-01 -1.18011326e-01 -7.42756307e-01 -1.51968271e-01 -4.87262607e-01 -1.20157135e+00 -1.36670053e+00 -4.23022687e-01 -7.04420865e-01 6.85440242e-01 1.09692180e+00 6.09958112e-01 7.14124739e-01 -9.47883725e-02 1.37631014e-01 -7.70329416e-01 -4.18991923e-01 -3.45136702e-01 4.21858788e-01 3.06292027e-01 1.23425879e-01 7.64109015e-01 -6.53996944e-01 -3.77993435e-01 5.94668329e-01 -8.76949012e-01 -3.37608933e-01 4.16377217e-01 5.17697513e-01 -2.13651434e-01 4.56250697e-01 1.11313534e+00 -1.27292097e+00 1.11135483e+00 -1.01401997e+00 -5.47542512e-01 -1.64834082e-01 -8.70748580e-01 -5.33836126e-01 7.76805043e-01 -2.05167487e-01 -9.28681195e-01 -4.18598801e-01 -1.17481992e-01 4.14125144e-01 5.32490835e-02 5.61612189e-01 3.18570957e-02 -3.22995514e-01 5.85086942e-01 3.37399364e-01 4.77964282e-01 -6.41599774e-01 9.56098959e-02 1.40146768e+00 -2.79026508e-01 9.01871733e-03 9.11292732e-01 7.27009356e-01 -3.11059862e-01 -1.14555871e+00 -7.69293487e-01 -9.02534068e-01 -6.57409728e-02 -3.46204013e-01 2.44276062e-01 -3.77849281e-01 -9.70294476e-01 9.29967642e-01 -8.42293978e-01 6.00716710e-01 4.74220902e-01 3.31930399e-01 -2.37704907e-02 8.34989011e-01 -7.66680717e-01 -9.52297926e-01 -4.96590465e-01 -9.42415714e-01 3.32569152e-01 5.73933840e-01 -3.29974651e-01 -1.00272238e+00 -1.76717609e-01 3.30914587e-01 7.76312232e-01 -1.45677283e-01 6.34552002e-01 -9.56053674e-01 -4.99048710e-01 -7.21158981e-01 -3.89361650e-01 1.32242665e-01 4.12455708e-01 -2.34972257e-02 -5.18647850e-01 -2.33165696e-01 3.16914588e-01 2.66802400e-01 8.00963342e-01 9.21965167e-02 1.24230874e+00 -5.48820198e-01 -9.83077407e-01 1.98611543e-01 9.50256944e-01 1.87499031e-01 5.18236995e-01 9.36940253e-01 3.42152089e-01 6.12747788e-01 7.69357979e-01 5.62842965e-01 1.59450457e-01 2.55055547e-01 5.69134593e-01 3.52121979e-01 4.60000724e-01 -1.77490324e-01 1.71401128e-01 1.09999335e+00 4.34936211e-03 -3.30513388e-01 -6.59371495e-01 1.86823323e-01 -2.00553393e+00 -1.04267311e+00 -6.74995244e-01 2.11314917e+00 5.64194143e-01 2.72021919e-01 5.49963474e-01 3.63254786e-01 1.20153177e+00 2.76075572e-01 -2.31121346e-01 -3.46022472e-02 -2.08350182e-01 -6.58973306e-02 6.58827543e-01 3.92261058e-01 -1.30081224e+00 9.67362523e-01 6.28892517e+00 1.10550952e+00 -1.35026050e+00 8.90793204e-02 3.39578211e-01 3.41651946e-01 -1.36265174e-01 6.61032274e-02 -1.02934635e+00 1.00583172e+00 7.71254301e-01 -7.45849848e-01 1.02878205e-01 8.61731946e-01 6.84154928e-01 -3.97134095e-01 5.14404364e-02 9.22087371e-01 2.53854454e-01 -1.24060595e+00 1.59048125e-01 -6.60454333e-02 5.95664501e-01 -1.40592992e-01 9.15108249e-03 1.14981674e-01 -1.65119603e-01 -6.42618179e-01 -1.48427961e-02 3.67316157e-01 -7.25412667e-02 -8.25131238e-01 9.44829762e-01 5.73403358e-01 -1.06032550e+00 -3.03141385e-01 -2.01207802e-01 -4.88218755e-01 4.81068105e-01 1.06596053e+00 -5.43100536e-01 6.00219369e-01 5.71187615e-01 9.36831892e-01 -6.55663729e-01 1.50980580e+00 -1.78342074e-01 9.11338210e-01 -1.86590016e-01 -8.56281579e-01 1.53471291e-01 -5.23678660e-01 7.98383296e-01 1.08157110e+00 1.01785481e-01 -1.83098912e-01 2.27413446e-01 2.98105776e-01 1.71510652e-01 6.96762204e-01 -5.20531654e-01 -8.24329183e-02 5.14362752e-01 1.33314884e+00 -1.17098784e+00 -4.41090107e-01 -5.31655490e-01 5.10242045e-01 -1.79563895e-01 1.05928473e-01 -7.02231646e-01 -7.97719896e-01 4.49430287e-01 8.05076897e-01 -5.45117080e-01 -3.79105121e-01 -3.42716336e-01 -1.03961003e+00 -3.40971947e-01 -1.04870248e+00 4.14453983e-01 -4.03825156e-02 -1.56939602e+00 3.29377055e-01 -2.98390575e-02 -1.44810319e+00 2.16525465e-01 -5.13401926e-01 -8.61145139e-01 6.86658382e-01 -1.28114414e+00 -5.74283063e-01 -2.10441053e-01 2.44878396e-01 5.09052336e-01 -5.49170196e-01 4.21931416e-01 5.14099598e-01 -6.33577704e-01 1.65005669e-01 3.95683706e-01 -3.38639691e-02 6.51874602e-01 -5.90095639e-01 2.22923309e-01 5.03622532e-01 -4.51332390e-01 1.01819122e+00 7.60662675e-01 -9.22640562e-01 -1.11963570e+00 -7.62742221e-01 1.12537074e+00 -2.10206166e-01 1.12467182e+00 -2.20865128e-03 -8.85055780e-01 2.31004223e-01 1.03872418e-01 -5.15464187e-01 6.70727730e-01 -4.65234630e-02 2.90109981e-02 -5.36356568e-02 -1.09063351e+00 7.40709126e-01 6.27269924e-01 -1.79417118e-01 -4.05961573e-01 5.40271699e-01 2.38100588e-01 1.98774901e-03 -1.24874637e-01 1.79911837e-01 4.30742770e-01 -1.18267965e+00 6.17345631e-01 -5.48044026e-01 -1.47856520e-02 -3.81094068e-01 4.68130082e-01 -9.70923126e-01 -8.25564265e-02 -7.21928954e-01 -1.70922652e-01 1.08448529e+00 8.78619775e-02 -1.23503578e+00 1.05375803e+00 6.98215291e-02 2.60337621e-01 -6.84209526e-01 -7.09213197e-01 -6.66466713e-01 -1.92896351e-01 -2.04985246e-01 3.98829609e-01 1.04372811e+00 4.81189609e-01 2.65631825e-01 -4.04521763e-01 -3.80220354e-01 4.21754062e-01 7.79460818e-02 8.14439714e-01 -1.50157404e+00 4.04084861e-01 -8.64242256e-01 -4.58521873e-01 -1.12616205e+00 1.48888677e-01 -7.05979347e-01 -6.99650288e-01 -1.38945723e+00 2.14201033e-01 -3.77120912e-01 1.27256632e-01 -9.53985453e-02 -2.49700814e-01 1.26367033e-01 -2.16967568e-01 6.85786724e-01 -6.16536677e-01 3.09488833e-01 1.13805163e+00 -4.04293612e-02 -2.89388180e-01 7.85954237e-01 -8.40359330e-01 1.19144976e+00 1.11857021e+00 -4.85878855e-01 -2.11668313e-01 2.34688178e-01 5.54896712e-01 -4.91304040e-01 -3.15538011e-02 -7.12198615e-01 3.79454106e-01 -3.25318307e-01 -2.46352404e-01 -6.65201664e-01 -5.30043766e-02 -7.57546723e-01 -3.36004466e-01 8.18331242e-01 1.56850100e-01 1.44558251e-01 -2.08102345e-01 6.98275983e-01 -4.71520901e-01 -5.17846704e-01 8.50690544e-01 -1.66599020e-01 -4.70118165e-01 1.76338762e-01 -1.02999079e+00 -6.69044033e-02 1.10841537e+00 -2.52591908e-01 -5.39061844e-01 -6.31222904e-01 -4.51096475e-01 2.99783915e-01 2.01964006e-01 6.22428298e-01 4.47555155e-01 -1.14040887e+00 -4.01600599e-01 7.94878751e-02 -1.79022700e-01 -6.86158895e-01 1.09043464e-01 1.10261726e+00 -7.13383138e-01 3.49890709e-01 7.57776871e-02 -1.59740463e-01 -1.40419114e+00 4.00172055e-01 5.23851137e-04 -7.14703351e-02 -2.70928562e-01 5.83670616e-01 -3.43634665e-01 -1.34002119e-01 2.22844005e-01 2.96682388e-01 -7.88000524e-01 4.90340859e-01 7.39311993e-01 1.08542538e+00 5.84209077e-02 -8.73025954e-01 -3.76504183e-01 3.21529657e-01 -2.97488749e-01 2.74859339e-01 8.06840003e-01 -3.64309072e-01 -7.50857890e-01 3.08974057e-01 1.05224431e+00 5.48947692e-01 1.65775895e-01 -1.21823214e-01 3.24115992e-01 -8.75819564e-01 -2.48419076e-01 -1.75940767e-01 -8.10596883e-01 4.57430631e-01 2.75171876e-01 1.27602732e+00 8.55129838e-01 -3.18756938e-01 1.18746281e+00 3.42302561e-01 4.37819898e-01 -1.30319762e+00 1.31763369e-01 8.49548757e-01 3.66742671e-01 -1.34495831e+00 2.19873518e-01 -1.00089383e+00 -2.37256527e-01 1.10967922e+00 6.27770424e-01 -3.26651871e-01 1.25294507e+00 -1.87297761e-02 6.07949160e-02 -2.35854387e-02 -1.63825676e-01 -1.72431976e-01 -9.44955721e-02 3.02219599e-01 6.76067054e-01 -1.12769403e-01 -1.25213468e+00 7.45293856e-01 -1.97626382e-01 -3.53194028e-01 4.79114383e-01 9.79048848e-01 -1.22129691e+00 -1.41197765e+00 -6.09970689e-01 1.07540524e+00 -7.05011070e-01 -4.88718823e-02 -5.10689318e-01 6.67527974e-01 -1.78394660e-01 1.37033629e+00 -2.38993675e-01 -4.95516926e-01 2.25197881e-01 -2.66469002e-01 -1.85131073e-01 -5.60838282e-01 -7.09589422e-01 -3.17736834e-01 3.45600247e-02 -4.67156582e-02 -5.13926387e-01 -5.60991824e-01 -1.17160046e+00 -8.22609305e-01 -9.43216145e-01 7.74588823e-01 6.02672040e-01 1.00984251e+00 1.57238811e-01 2.86120921e-02 9.65430140e-01 -5.40478349e-01 -5.65645695e-01 -8.83376598e-01 -7.26314902e-01 4.07073230e-01 -1.04940593e-01 -8.38453829e-01 -9.58971918e-01 -5.38557649e-01]
[7.88826322555542, 10.057147979736328]
0993a9ac-b4bf-4046-9185-4f494d15c5ca
jentab-meets-semtab-2021-s-new-challenges
null
null
https://www.semanticscholar.org/paper/JenTab-Meets-SemTab-2021's-New-Challenges-Abdelmageed-Schindler/4f492fee6a7ae51d3f2527d9036a1beaf6f1e44b
http://ceur-ws.org/Vol-3103/paper4.pdf
JenTab Meets SemTab 2021's New Challenges
While tables are a rich source of structured information, their automated use is oftentimes prevented by the inherent ambiguity contained within. Issues ranging from mere typos over inconsistent naming conventions to homonymy among values pose substantial barriers to exploiting this source of knowledge. Although the Semantic Web can alleviate many of these issues, the actual annotation process remains challenging. To foster new ideas and the improvement of existing approaches, the Semantic Web Challenge on Tabular Data to Knowledge Graph Matching (SemTab) since 2019 hosts yearly competitions allowing systems to present their current capabilities. Datasets of different origins and characteristics highlight the various challenges present in this area. In this paper, we report on the evolution of our system, “JenTab”, during SemTab2021. We re-designed the system architecture, optimized individual modules, and developed various pipelines to target specific challenges posed throughout the challenge. JenTab is among the top 5 systems in the first two rounds of SemTab2021. The results demonstrate JenTab’s flexibility and its ability to quickly address new challenges.
['Sirko Schindler', 'Nora Abdelmageed']
2021-10-01
null
null
null
semtab-iswc-2021-10
['graph-matching', 'table-annotation', 'table-annotation', 'column-type-annotation', 'cell-entity-annotation']
['graphs', 'knowledge-base', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing']
[-1.19772665e-01 2.88502932e-01 -2.95007795e-01 -3.51227820e-01 -7.12199330e-01 -1.10264277e+00 6.60023749e-01 5.65219998e-01 -1.02479421e-01 7.10691512e-01 3.76426518e-01 -3.81603181e-01 -4.84015316e-01 -9.32946086e-01 -4.40034389e-01 2.75422186e-01 6.95962384e-02 7.77260482e-01 5.11916518e-01 -5.56194842e-01 3.74262244e-01 1.99339405e-01 -1.76105654e+00 5.54860413e-01 8.59126210e-01 9.53173876e-01 -2.21109137e-01 3.48739475e-02 -9.92370009e-01 1.04557931e+00 -4.19947356e-01 -6.95348561e-01 3.36807758e-01 -4.93921079e-02 -1.25942028e+00 -4.37431872e-01 7.28839576e-01 3.13225716e-01 -9.21155065e-02 1.09812748e+00 1.12885438e-01 -1.01975419e-01 4.93070781e-02 -1.63310194e+00 -5.33304989e-01 8.64933491e-01 -2.80924171e-01 1.71346758e-02 7.77933240e-01 -2.66920596e-01 1.39057219e+00 -7.74280310e-01 1.22017634e+00 9.01775599e-01 1.04053700e+00 2.72398263e-01 -1.17386413e+00 -6.68796182e-01 8.01565424e-02 3.81966114e-01 -1.43761122e+00 -6.67252362e-01 2.79086977e-01 -5.88953614e-01 1.10633171e+00 6.18970871e-01 5.79552054e-01 6.72382653e-01 -3.77438605e-01 2.35076159e-01 1.02322400e+00 -2.04547048e-01 1.18061654e-01 4.29941744e-01 2.62524307e-01 5.81029236e-01 8.25023949e-01 -6.96813464e-01 -8.09871197e-01 -2.72207856e-01 2.60714710e-01 -3.35000277e-01 -5.27771153e-02 -8.34059596e-01 -1.19188368e+00 3.58249426e-01 2.60862410e-01 4.18410838e-01 5.76424934e-02 -2.16176555e-01 5.87943912e-01 3.91349614e-01 -7.33127445e-02 9.52382863e-01 -5.68008125e-01 -3.61890405e-01 -7.45156586e-01 4.12207484e-01 1.29915762e+00 1.47659171e+00 7.62103915e-01 -6.18544340e-01 3.83896530e-01 8.17722082e-01 1.22067355e-01 6.20342046e-02 1.73849657e-01 -9.64157164e-01 7.40094960e-01 1.24761188e+00 1.77793026e-01 -1.24740791e+00 -4.86468256e-01 -3.69142473e-01 -2.15784580e-01 -3.98409627e-02 7.03371584e-01 3.45398515e-01 -6.10342979e-01 1.46215224e+00 4.15076524e-01 -7.90859520e-01 -6.34182766e-02 7.11326659e-01 9.35390115e-01 1.48943961e-01 1.07697450e-01 1.22724302e-01 1.51531184e+00 -5.95433712e-01 -7.95565844e-01 -2.72567600e-01 7.81725824e-01 -8.25527072e-01 9.35862780e-01 2.09775716e-01 -9.49620903e-01 -5.02030402e-02 -9.99136209e-01 -3.57492328e-01 -1.18118858e+00 -7.60932744e-01 7.30966508e-01 6.85040236e-01 -1.05024624e+00 5.57845831e-01 -4.81321186e-01 -1.08106172e+00 2.44925752e-01 7.36472979e-02 -6.97156310e-01 -1.08070605e-01 -1.40420258e+00 1.10004783e+00 5.12664557e-01 -3.00929338e-01 1.62426725e-01 -1.01161849e+00 -6.69631898e-01 7.19901174e-02 1.03909373e+00 -6.59708738e-01 1.17145085e+00 -2.97006369e-01 -5.58947623e-01 9.44712460e-01 1.38585165e-01 -4.22511071e-01 4.61883545e-01 4.80540469e-02 -8.10160697e-01 -2.06893101e-01 5.60997367e-01 3.61617982e-01 -7.07626194e-02 -9.19130206e-01 -7.69822955e-01 -2.56179988e-01 1.98546842e-01 4.84765917e-02 -5.14894366e-01 2.86043227e-01 -6.54298604e-01 -3.60440254e-01 3.02597374e-01 -5.39947450e-01 1.63689822e-01 9.47852284e-02 -2.79130876e-01 -2.32494444e-01 6.02916598e-01 -5.70628762e-01 1.60489082e+00 -1.92071509e+00 -9.64037701e-02 3.52574527e-01 3.71867388e-01 -8.93228725e-02 2.97980368e-01 1.19829011e+00 9.93959513e-03 6.19973183e-01 -2.59155482e-01 1.32773489e-01 4.29405332e-01 3.06155682e-01 -1.08426064e-01 -9.61462408e-02 -1.01309195e-01 8.25203359e-01 -1.01101232e+00 -7.12534845e-01 -5.22507578e-02 7.11418912e-02 -3.98954660e-01 -3.40395361e-01 -5.57004452e-01 -1.19588338e-01 -2.60263950e-01 1.01845264e+00 5.58683455e-01 -6.01288140e-01 9.00864124e-01 -4.39368874e-01 -5.50498486e-01 7.70123720e-01 -1.49894750e+00 1.83428359e+00 -5.61465882e-03 3.71814638e-01 1.31799713e-01 -5.37589431e-01 9.05712783e-01 3.16436589e-02 8.33419800e-01 -9.53526437e-01 -3.05559963e-01 5.94550788e-01 -2.40611255e-01 -5.63308477e-01 7.38298893e-01 2.46477708e-01 -2.72485077e-01 4.01947349e-01 -1.23486169e-01 -1.14457622e-01 6.80832148e-01 5.57019889e-01 1.26027107e+00 3.03232193e-01 5.70540309e-01 -4.75783736e-01 2.61881739e-01 8.07122469e-01 5.27219355e-01 5.90549469e-01 2.22845618e-02 4.45084751e-01 5.66727519e-01 -6.83135271e-01 -1.04936898e+00 -8.91563952e-01 -2.53612369e-01 7.63706326e-01 7.32930377e-02 -1.21591675e+00 -4.61077124e-01 -8.57267082e-01 4.17572349e-01 6.70175254e-01 -4.77176577e-01 2.47544721e-01 -3.65751326e-01 -3.71445715e-01 5.66816211e-01 4.51815248e-01 4.93928790e-01 -7.02765524e-01 -4.57809806e-01 2.95952231e-01 -4.36362475e-01 -1.33305109e+00 -9.91890430e-02 -3.42733450e-02 -3.80125672e-01 -1.64363194e+00 2.32764810e-01 -4.24817830e-01 2.88908809e-01 1.88199073e-01 1.55288076e+00 7.40345418e-02 -3.37031454e-01 4.13422465e-01 -3.22896898e-01 -3.21803093e-01 -3.03798407e-01 3.86603236e-01 -2.80323833e-01 -4.71983641e-01 7.20502138e-01 -4.58314061e-01 -2.80793458e-01 3.22089911e-01 -9.39727306e-01 1.21224321e-01 1.83223471e-01 3.78060818e-01 3.76733691e-01 3.60677652e-02 4.99311507e-01 -1.42189372e+00 3.78379464e-01 -6.56448066e-01 -8.21909308e-01 5.48538268e-01 -1.24418831e+00 1.66977748e-01 4.51721698e-01 3.52304459e-01 -7.78828204e-01 -1.11495435e-01 7.71528780e-02 2.08560795e-01 9.56853554e-02 9.58503783e-01 -2.33431950e-01 2.07005674e-03 5.87860048e-01 -2.33975723e-01 8.07047635e-02 -9.77838516e-01 3.95098627e-01 4.87427264e-01 7.08423316e-01 -7.85415769e-01 9.80354846e-01 3.64782959e-01 -1.07326142e-01 -1.02255471e-01 -8.72745872e-01 -4.54397053e-01 -5.14410019e-01 -1.12639971e-01 4.71487582e-01 -8.24810624e-01 -6.78695023e-01 1.06292121e-01 -7.34661996e-01 8.90881270e-02 -4.14957404e-01 -2.71659106e-01 -4.96283881e-02 3.34478885e-01 -1.62881121e-01 -4.03695822e-01 -2.89956421e-01 -7.20309913e-01 4.60793555e-01 -2.83752531e-02 -6.85240149e-01 -9.91882265e-01 9.13751796e-02 8.16142261e-01 7.75497913e-01 4.60277498e-01 1.15124917e+00 -9.54564750e-01 -7.73643076e-01 -2.21237212e-01 -5.66663921e-01 -2.26669207e-01 3.80295932e-01 -6.46080868e-03 -6.46579623e-01 -7.28439614e-02 -7.50394881e-01 -2.46056646e-01 1.98389336e-01 -6.15743995e-01 1.03772771e+00 -3.16306382e-01 -2.95650095e-01 4.23693299e-01 1.62426746e+00 1.06778488e-01 5.52926958e-01 1.19522774e+00 6.55407429e-01 8.16232622e-01 4.91064668e-01 2.30469123e-01 1.03958881e+00 8.60420048e-01 4.95355666e-01 2.06651926e-01 -1.74899891e-01 -5.10809839e-01 -1.94154143e-01 8.95099759e-01 2.56304413e-01 -3.23657617e-02 -1.46065092e+00 6.35865092e-01 -1.93947446e+00 -8.94632638e-01 -3.38378549e-01 2.13183284e+00 9.13033128e-01 3.03391606e-01 1.31276920e-01 5.32099698e-03 5.29121280e-01 -1.69072393e-02 -4.72355843e-01 -3.66299093e-01 -3.49077255e-01 1.21871583e-01 5.88771999e-01 2.15862989e-01 -6.91922784e-01 8.17062140e-01 6.98097801e+00 3.47374111e-01 -6.62828445e-01 -1.08598679e-01 -1.09703459e-01 2.31601782e-02 -5.28498173e-01 4.06970590e-01 -8.81833375e-01 4.34421629e-01 9.14273262e-01 -7.86475480e-01 6.81371927e-01 8.01407814e-01 -3.01209211e-01 1.12802893e-01 -1.08081639e+00 7.35056221e-01 5.80019259e-04 -1.76931489e+00 1.78516388e-01 -3.86132896e-02 4.04989392e-01 2.08897516e-01 -2.68273413e-01 3.69044214e-01 6.10919714e-01 -9.89496648e-01 7.72424757e-01 4.11818355e-01 9.38373029e-01 -4.02885407e-01 4.96526957e-01 -2.31368229e-01 -1.32283413e+00 1.62780993e-02 -8.47380012e-02 -9.90353823e-02 -2.31587872e-01 3.66596162e-01 -7.55386710e-01 1.01625168e+00 1.02226758e+00 8.88994813e-01 -9.57720339e-01 1.04076290e+00 1.78375244e-01 1.50025170e-02 -4.48489398e-01 9.57926512e-02 -2.72932798e-02 -2.89116260e-02 4.75092530e-01 1.27168894e+00 1.68999106e-01 -2.78242737e-01 8.15631002e-02 6.57493532e-01 -3.36047530e-01 3.25584799e-01 -6.07742786e-01 -3.09129745e-01 9.98068213e-01 1.25419998e+00 -6.16720438e-01 -3.77288520e-01 -7.37014413e-01 2.82265544e-01 5.24283648e-01 5.01183942e-02 -5.32880723e-01 -5.69386303e-01 9.57701325e-01 5.18523932e-01 9.34325308e-02 -7.18447492e-02 -5.25253892e-01 -1.29138339e+00 3.78122598e-01 -1.30624247e+00 1.07737100e+00 -1.03722298e+00 -1.07466185e+00 4.48587298e-01 8.95712972e-02 -9.29988384e-01 -1.26142785e-01 -5.18040299e-01 2.05087915e-01 6.77976966e-01 -1.32907927e+00 -9.39092338e-01 -4.23833519e-01 3.93187881e-01 2.21498728e-01 6.07780106e-02 9.55759525e-01 6.35873377e-01 -3.27543437e-01 5.17270386e-01 1.87640414e-01 1.54519990e-01 9.75549519e-01 -1.42206097e+00 7.77575433e-01 7.67315030e-01 1.67001024e-01 9.81418252e-01 6.85990512e-01 -7.78698146e-01 -1.75348127e+00 -7.69323885e-01 1.27588189e+00 -9.41731632e-01 1.35507858e+00 -6.79975510e-01 -1.04067373e+00 8.49833190e-01 1.52938649e-01 -2.13976011e-01 5.86869001e-01 3.96502942e-01 -9.83188689e-01 -2.99697369e-01 -1.07200575e+00 5.70267200e-01 1.49006450e+00 -5.78867257e-01 -7.02441156e-01 2.70116478e-01 6.28967047e-01 -4.66031104e-01 -1.33305824e+00 3.40385616e-01 7.68193185e-01 -9.00985241e-01 7.36633599e-01 -7.88702905e-01 2.02201366e-01 -5.97634852e-01 -3.05103868e-01 -9.77343202e-01 -2.42722854e-01 -7.95425177e-01 -4.58029397e-02 1.49907029e+00 5.93618810e-01 -8.01151633e-01 8.37571859e-01 1.06887794e+00 -5.54521978e-02 -2.75991112e-01 -7.73392022e-01 -9.03257847e-01 -1.82591900e-01 -4.35652554e-01 1.08248711e+00 1.39942539e+00 5.90590119e-01 1.15745626e-01 1.33719333e-02 -6.55514523e-02 7.24727869e-01 2.86200553e-01 9.18432236e-01 -1.61247456e+00 3.04171331e-02 -4.49123085e-01 -5.68644345e-01 -3.14014375e-01 -5.32549560e-01 -1.22180343e+00 -6.41422272e-01 -2.07801247e+00 2.52100438e-01 -6.25907004e-01 -3.61117750e-01 8.80952239e-01 1.99204147e-01 1.24079190e-01 2.44870916e-01 3.52628648e-01 -8.24150681e-01 -3.31226408e-01 7.98062801e-01 -1.52085973e-02 1.46941096e-01 -7.63005078e-01 -1.24036539e+00 2.87532151e-01 6.89363599e-01 -5.40371060e-01 -3.60154808e-01 -6.11399531e-01 8.97732735e-01 -3.26322228e-01 1.08519815e-01 -1.15022135e+00 7.12581992e-01 -3.46147418e-01 1.93174742e-02 -5.60138822e-01 1.00868352e-01 -1.01825595e+00 8.22987080e-01 1.69557437e-01 -2.24666297e-01 3.88112873e-01 4.32223469e-01 1.43341988e-01 -1.78858086e-01 -4.02621255e-04 3.65632683e-01 -2.76107013e-01 -9.62312996e-01 -1.80394184e-02 -1.85824007e-01 6.60212278e-01 7.11107433e-01 -2.02614009e-01 -9.82714534e-01 3.88000309e-02 -6.36620045e-01 6.10762417e-01 9.85117316e-01 7.48140812e-01 -4.66594385e-04 -1.08045197e+00 -2.66629428e-01 3.38904634e-02 6.88831627e-01 -2.29778573e-01 -1.74754903e-01 7.17246354e-01 -5.59124112e-01 5.68738222e-01 -5.05950630e-01 -1.02444299e-01 -1.14932704e+00 6.49599910e-01 3.37577254e-01 -3.75657380e-01 -7.44918048e-01 2.35064253e-01 -6.00330293e-01 -7.11174250e-01 1.88717514e-01 -3.17058638e-02 -1.53153077e-01 2.72824913e-01 3.68996501e-01 4.52999264e-01 5.16102374e-01 -3.14426214e-01 -6.77850068e-01 3.05177540e-01 -2.35946357e-01 1.04771607e-01 1.38836825e+00 -2.58190811e-01 -5.29330254e-01 4.45319653e-01 7.17495322e-01 3.21112365e-01 -5.32462478e-01 -3.56535465e-01 8.51134837e-01 -5.88997662e-01 -3.79813105e-01 -1.42559218e+00 -7.17577279e-01 2.18142197e-01 -5.69081008e-02 7.08305657e-01 7.51912594e-01 -4.03887890e-02 7.96737313e-01 5.19771934e-01 7.42245793e-01 -1.18641293e+00 -5.88199258e-01 3.72722000e-01 6.77978039e-01 -1.01626933e+00 1.72114700e-01 -7.70823240e-01 -3.14441293e-01 1.12028420e+00 7.71483958e-01 6.74220741e-01 3.59686971e-01 4.09912974e-01 3.25239807e-01 -5.52546024e-01 -9.68596280e-01 -2.81930983e-01 -4.86908946e-04 5.95434248e-01 5.09013891e-01 -2.39541858e-01 -3.63394231e-01 3.72230649e-01 -4.33870256e-01 4.80371229e-02 4.91167873e-01 1.25332713e+00 -2.45127141e-01 -1.47461164e+00 -3.94376159e-01 5.34551084e-01 -3.70616317e-01 -1.47472978e-01 -8.23548079e-01 1.14889598e+00 -1.31047340e-02 8.28088880e-01 -3.14259119e-02 -4.76376057e-01 7.52677798e-01 4.14439827e-01 2.13907629e-01 -5.58574677e-01 -8.62307429e-01 -5.08805275e-01 7.51189590e-01 -7.38263786e-01 -9.17028859e-02 -8.71085227e-01 -1.19810462e+00 -6.37128234e-01 2.48603821e-01 5.30064285e-01 8.32638860e-01 5.33637762e-01 7.11497128e-01 2.65648991e-01 -3.96258198e-02 3.88846785e-01 -3.77534688e-01 -4.31317478e-01 -2.96886772e-01 6.83182120e-01 -2.32950419e-01 -7.53884196e-01 -1.70513064e-01 5.84201552e-02]
[9.310271263122559, 7.959108352661133]
dc895f3e-6001-4a6e-968c-a0ee421534d5
aligntransformer-hierarchical-alignment-of
2203.10095
null
https://arxiv.org/abs/2203.10095v1
https://arxiv.org/pdf/2203.10095v1.pdf
AlignTransformer: Hierarchical Alignment of Visual Regions and Disease Tags for Medical Report Generation
Recently, medical report generation, which aims to automatically generate a long and coherent descriptive paragraph of a given medical image, has received growing research interests. Different from the general image captioning tasks, medical report generation is more challenging for data-driven neural models. This is mainly due to 1) the serious data bias: the normal visual regions dominate the dataset over the abnormal visual regions, and 2) the very long sequence. To alleviate above two problems, we propose an AlignTransformer framework, which includes the Align Hierarchical Attention (AHA) and the Multi-Grained Transformer (MGT) modules: 1) AHA module first predicts the disease tags from the input image and then learns the multi-grained visual features by hierarchically aligning the visual regions and disease tags. The acquired disease-grounded visual features can better represent the abnormal regions of the input image, which could alleviate data bias problem; 2) MGT module effectively uses the multi-grained features and Transformer framework to generate the long medical report. The experiments on the public IU-Xray and MIMIC-CXR datasets show that the AlignTransformer can achieve results competitive with state-of-the-art methods on the two datasets. Moreover, the human evaluation conducted by professional radiologists further proves the effectiveness of our approach.
['Xian Wu', 'Jing Zhang', 'Xiaoxia Xie', 'Shen Ge', 'Fenglin Liu', 'Di You']
2022-03-18
null
null
null
null
['medical-report-generation']
['medical']
[ 4.02859569e-01 3.12392086e-01 -1.41482100e-01 -3.35356414e-01 -1.13044035e+00 -8.49511772e-02 6.14224195e-01 -1.31689347e-02 -1.01635568e-01 6.91154122e-01 8.34367752e-01 -1.17509671e-01 -7.31867105e-02 -6.15087748e-01 -4.86360759e-01 -9.74972725e-01 2.39882201e-01 4.00453389e-01 7.65518844e-03 1.95284843e-01 1.73652083e-01 4.76410165e-02 -1.45190799e+00 7.61294007e-01 9.11311924e-01 1.04037976e+00 5.40012240e-01 6.44558609e-01 -1.49241716e-01 1.11984527e+00 -5.11773288e-01 -1.41965881e-01 -2.51121461e-01 -7.94459045e-01 -8.12290430e-01 4.55816001e-01 1.88591406e-01 -4.24360752e-01 -4.66043234e-01 9.78265643e-01 6.97417676e-01 -1.90420747e-01 8.24358165e-01 -1.12781060e+00 -1.07242346e+00 5.69977880e-01 -8.12016308e-01 2.74484068e-01 2.05497175e-01 2.35896602e-01 9.78546381e-01 -8.16379547e-01 7.34809458e-01 1.15603960e+00 1.42151415e-01 7.39278376e-01 -7.57064521e-01 -5.41712821e-01 2.50504673e-01 2.37046584e-01 -1.33324242e+00 1.80260353e-02 7.13521600e-01 -6.41145289e-01 5.25009334e-01 5.09223163e-01 6.55512512e-01 1.19523060e+00 3.43733698e-01 1.13463163e+00 1.03580463e+00 -7.83601478e-02 -1.00094609e-01 -6.86792482e-04 -1.17468853e-02 8.19004238e-01 7.80380741e-02 1.43777087e-01 -9.25559029e-02 -7.79299289e-02 7.96009600e-01 4.51532632e-01 -5.75226724e-01 -1.38789102e-01 -1.74562156e+00 9.00538027e-01 7.56426990e-01 5.78318655e-01 -7.07616985e-01 -6.59288093e-02 4.23012227e-01 -2.85777718e-01 5.43136835e-01 3.48117173e-01 -2.30369326e-02 3.74547184e-01 -9.85099256e-01 3.51113051e-01 1.53942421e-01 9.29840147e-01 3.29989344e-01 -9.75277051e-02 -1.02462566e+00 7.81366289e-01 3.43004912e-01 3.75130326e-01 9.19121504e-01 -4.26902533e-01 6.42869174e-01 7.26184249e-01 -2.28095409e-02 -1.00932610e+00 -3.55373472e-01 -6.38949871e-01 -1.33563530e+00 -2.22544208e-01 6.07379060e-03 -2.01499164e-02 -1.31576562e+00 1.54849195e+00 1.24833778e-01 -5.95046692e-02 2.16841549e-01 1.10186553e+00 1.35537815e+00 9.07122612e-01 2.55875885e-01 -4.36219066e-01 1.73763585e+00 -1.14722860e+00 -9.94297087e-01 -1.37114182e-01 6.55081928e-01 -6.04153156e-01 1.05513477e+00 -1.14126485e-02 -1.05219531e+00 -6.39764905e-01 -9.77001429e-01 -8.15667212e-02 -8.62792879e-02 4.85161215e-01 5.00121474e-01 -7.32198581e-02 -8.78819346e-01 -8.69724974e-02 -5.05344450e-01 -1.62855104e-01 6.82765961e-01 -1.70904137e-02 -3.28340530e-01 -4.22872990e-01 -1.13244915e+00 6.60612762e-01 6.20225191e-01 1.63265526e-01 -9.53045607e-01 -6.58140957e-01 -9.17001963e-01 6.10367768e-02 2.23541722e-01 -9.53064859e-01 1.18251228e+00 -8.26976001e-01 -7.02858925e-01 1.13096464e+00 -2.27442179e-02 -2.93350637e-01 5.89325190e-01 7.50973895e-02 -2.70264506e-01 3.22586924e-01 3.28927189e-01 1.00037384e+00 7.22621500e-01 -1.28713441e+00 -7.78512716e-01 -3.40078533e-01 -2.84359634e-01 4.60780650e-01 -1.04584962e-01 -2.56452113e-01 -5.17793059e-01 -1.08527029e+00 -4.00987342e-02 -7.24807382e-01 -4.05203640e-01 -1.68657139e-01 -8.69790375e-01 -3.73076141e-01 6.71600282e-01 -6.85629964e-01 1.32394385e+00 -2.22234178e+00 1.84103400e-01 -1.06962621e-01 5.39986491e-01 5.05967960e-02 -5.74302971e-02 4.95564453e-02 -3.98724645e-01 3.14623676e-02 -2.88297027e-01 -6.21167943e-02 -2.57063866e-01 4.18494567e-02 -3.53008717e-01 1.83533803e-01 2.45967582e-01 1.13494623e+00 -9.14189339e-01 -9.57304239e-01 1.41164482e-01 2.33699366e-01 -3.22248876e-01 6.99978471e-01 -1.37380555e-01 5.25086701e-01 -6.41569555e-01 6.32671297e-01 4.47194099e-01 -6.77653491e-01 -2.53277600e-01 -5.97973466e-01 1.03984669e-01 -5.09267077e-02 -4.61512506e-01 1.67950583e+00 -2.52306253e-01 3.04928899e-01 -3.82946044e-01 -7.61185169e-01 7.56264031e-01 6.08958364e-01 6.22204661e-01 -8.88024688e-01 8.82062986e-02 7.35094622e-02 -6.71757683e-02 -9.26195085e-01 2.63981849e-01 -3.30251545e-01 -2.25310788e-01 5.40685236e-01 -2.01710582e-01 2.98729371e-02 6.33685989e-03 4.92102832e-01 1.05024409e+00 -1.21743254e-01 3.46843332e-01 -3.86919230e-02 5.62836468e-01 1.28052652e-01 5.83153188e-01 5.45365572e-01 -4.44798619e-02 1.04824352e+00 4.87724990e-01 -5.53548694e-01 -9.77912605e-01 -1.12971568e+00 3.25614624e-02 6.68424547e-01 2.37488061e-01 -1.23885207e-01 -7.59548604e-01 -1.16195405e+00 -2.87873805e-01 6.69087231e-01 -1.09131956e+00 -2.44744956e-01 -4.48671103e-01 -9.18124855e-01 2.20219120e-01 7.61882484e-01 4.37616706e-01 -1.58394587e+00 -5.38728654e-01 8.19257051e-02 -7.03037262e-01 -9.44060981e-01 -1.04067349e+00 -1.42792016e-01 -5.13984621e-01 -9.75003362e-01 -1.32284343e+00 -9.24788177e-01 1.11697435e+00 2.19476745e-01 1.13568211e+00 2.56243676e-01 -6.63688302e-01 8.44758153e-02 -5.30628264e-01 -4.25185531e-01 -4.19390976e-01 -5.49093410e-02 -4.82476264e-01 1.53609410e-01 9.67478082e-02 -3.91567573e-02 -9.59268749e-01 5.84313534e-02 -1.24816644e+00 9.05465364e-01 1.22549784e+00 1.08377075e+00 9.24517095e-01 -4.77699414e-02 6.91428304e-01 -1.08886886e+00 6.63568377e-01 -6.72618687e-01 -9.14437175e-02 3.04823279e-01 -5.65978408e-01 5.76954074e-02 3.50051790e-01 -3.86693001e-01 -1.10162723e+00 1.53298266e-02 -2.37277448e-01 -4.57304120e-01 -1.21043101e-01 3.40377927e-01 -1.29018307e-01 6.77514672e-01 2.51425773e-01 7.31048405e-01 -4.39275727e-02 -8.45497847e-02 2.94266135e-01 7.59418964e-01 7.37440109e-01 -1.46015704e-01 5.83292007e-01 3.55175555e-01 -2.16767728e-01 -1.99476108e-01 -1.25207639e+00 -2.75455594e-01 -2.36762032e-01 -2.30964243e-01 1.53777349e+00 -8.92151952e-01 -4.30329114e-01 3.27163339e-01 -1.30072808e+00 2.35310405e-01 -2.80655086e-01 3.83949071e-01 -7.10669696e-01 1.68621033e-01 -6.80329621e-01 -5.25349975e-01 -8.33468974e-01 -1.54064775e+00 1.34525907e+00 4.11145836e-01 -9.70767718e-03 -7.53913820e-01 -1.63058400e-01 4.34753120e-01 1.08833611e-01 5.36171019e-01 1.29412615e+00 -5.09638190e-01 -7.15532243e-01 6.07007518e-02 -6.96104467e-01 1.48858488e-01 2.85677701e-01 -3.27475846e-01 -7.92330205e-01 -2.30605662e-01 2.63524950e-02 -3.13883185e-01 9.36075687e-01 5.30312657e-01 1.47053528e+00 -5.89891076e-01 -4.56334144e-01 4.68279034e-01 1.18483782e+00 3.16081256e-01 6.36953831e-01 9.79175419e-02 9.13213968e-01 6.06607854e-01 9.60922420e-01 2.99118310e-01 6.41992509e-01 5.13710320e-01 6.69132292e-01 -7.73911893e-01 -2.82257915e-01 -4.82351899e-01 -1.38749769e-02 9.22222674e-01 1.86626256e-01 -2.10927457e-01 -7.81645238e-01 8.57188761e-01 -1.91365612e+00 -8.52949142e-01 -9.29253772e-02 1.66881561e+00 9.34852183e-01 -1.30744219e-01 2.93000117e-02 -6.33000284e-02 7.65569150e-01 2.89726138e-01 -4.64803070e-01 -2.92303599e-02 -3.17631289e-02 -3.25043350e-01 1.66807741e-01 2.02594884e-02 -1.00813448e+00 3.81248534e-01 5.33466768e+00 8.74620855e-01 -1.01372969e+00 1.55974284e-01 1.06003106e+00 -1.36807086e-02 -4.76368576e-01 -5.42220294e-01 -6.81769550e-01 7.57927597e-01 4.24610764e-01 -2.40657255e-01 -3.27136904e-01 7.33249366e-01 1.08395629e-01 2.52362788e-01 -1.16336107e+00 1.16458762e+00 3.36801112e-01 -1.54909182e+00 7.23650455e-01 3.22704315e-01 7.65393853e-01 -2.16186225e-01 2.43920341e-01 2.87738860e-01 1.60091937e-01 -1.15396786e+00 6.67850375e-01 8.69694412e-01 1.08446097e+00 -7.77045131e-01 1.06295371e+00 2.11094186e-01 -1.03227103e+00 -8.49397406e-02 -3.26302201e-01 6.06092513e-01 2.10305095e-01 6.04845226e-01 -1.00055075e+00 8.23071122e-01 5.19782603e-01 7.88226247e-01 -6.34401619e-01 1.00846040e+00 -1.02819838e-01 3.64099473e-01 3.73015612e-01 7.85704553e-02 5.40814161e-01 7.20448941e-02 3.57059896e-01 1.30109751e+00 4.45813179e-01 2.19724223e-01 2.47799978e-01 9.64158118e-01 -7.99519569e-02 1.71348080e-01 -4.95312154e-01 -1.58405378e-02 6.62300736e-02 1.42941833e+00 -5.99729478e-01 -5.20844281e-01 -3.61600757e-01 9.07149076e-01 1.18011959e-01 2.73360819e-01 -8.89887750e-01 -3.35938990e-01 1.39199778e-01 2.54036456e-01 3.18362653e-01 5.00205100e-01 -1.80196285e-01 -1.14321613e+00 6.92820400e-02 -9.10655975e-01 7.11608946e-01 -1.13649106e+00 -1.31332541e+00 1.09524012e+00 -1.80722892e-01 -1.59660089e+00 -4.16835129e-01 -2.29958117e-01 -5.37679613e-01 7.81140268e-01 -1.55950177e+00 -1.27431655e+00 -6.62515402e-01 5.62461793e-01 7.81091690e-01 -1.73831031e-01 8.42621446e-01 2.71710664e-01 -4.48223710e-01 4.03272122e-01 -3.72343361e-01 3.16584140e-01 5.06013453e-01 -1.40269911e+00 1.37436435e-01 6.20437264e-01 -2.16926746e-02 2.78068602e-01 4.66728956e-01 -6.05585754e-01 -8.35117459e-01 -1.46878946e+00 7.39432752e-01 -3.32234472e-01 2.87980318e-01 -1.27545744e-01 -9.83513057e-01 5.55692017e-01 3.10422719e-01 1.24308981e-01 6.64107800e-01 -5.38655818e-01 -1.82058677e-01 -1.21254683e-01 -1.07884550e+00 4.98377681e-01 7.67536104e-01 -1.70196727e-01 -6.95113122e-01 5.20498633e-01 9.84931767e-01 -4.53238070e-01 -9.39946294e-01 5.60825229e-01 3.21453571e-01 -7.60308862e-01 8.02383900e-01 -4.88900483e-01 8.91374946e-01 -4.70475197e-01 2.29270831e-02 -1.31063104e+00 -4.83163834e-01 -8.18054825e-02 4.59536426e-02 1.33068705e+00 3.23634773e-01 -2.27578744e-01 4.73587811e-01 1.04975305e-01 -2.07164109e-01 -1.30106628e+00 -6.74514413e-01 -6.06319755e-02 -1.65744796e-01 4.64298651e-02 6.58245921e-01 8.71396303e-01 -2.86149114e-01 4.99610513e-01 -4.88995761e-01 -3.72006819e-02 5.48294306e-01 4.07639474e-01 2.90526211e-01 -9.14253235e-01 -1.37381658e-01 -3.51379156e-01 -3.09272051e-01 -9.34158444e-01 -2.49006808e-01 -1.02045047e+00 2.87163466e-01 -1.98923719e+00 8.53746116e-01 -3.90159100e-01 -4.91388649e-01 5.46630085e-01 -6.64228499e-01 3.63308817e-01 7.76424631e-02 3.58040810e-01 -7.78241277e-01 6.06430531e-01 1.84504318e+00 -4.08441186e-01 1.76606163e-01 -7.14350119e-02 -1.04800105e+00 5.67709684e-01 3.45697492e-01 -4.93977547e-01 -5.42192698e-01 -4.13564980e-01 -4.23187912e-02 4.39132214e-01 2.54646450e-01 -8.03994000e-01 -3.70022319e-02 -1.14527538e-01 6.77245378e-01 -1.10372865e+00 2.97664385e-02 -6.80419445e-01 -1.55695125e-01 6.08214021e-01 -5.34488797e-01 1.77246496e-01 -1.42220736e-01 5.96531928e-01 -4.47997332e-01 1.35957077e-01 8.17033529e-01 -4.04149294e-01 -4.12690520e-01 6.51354253e-01 -1.05216578e-01 1.67983443e-01 1.21676314e+00 -5.49609028e-02 -4.55127180e-01 -2.45946720e-01 -7.13089228e-01 4.74192739e-01 7.40442500e-02 7.66585052e-01 9.89384115e-01 -1.55840218e+00 -1.20596683e+00 2.27022633e-01 5.93772471e-01 3.42550933e-01 7.17468858e-01 9.31864321e-01 -4.10596281e-01 5.56593895e-01 -9.15370807e-02 -8.73940587e-01 -1.22881711e+00 7.34983861e-01 2.58767277e-01 -7.89318144e-01 -8.95982504e-01 6.72274590e-01 1.13581789e+00 -5.73648848e-02 1.07505985e-01 -2.85301745e-01 -3.91430527e-01 6.37003127e-03 9.14930046e-01 -2.31135458e-01 -7.69513696e-02 -7.19627261e-01 -2.52231389e-01 5.01651287e-01 -7.59305894e-01 2.11698145e-01 1.23506594e+00 -1.73787117e-01 9.12007689e-02 2.36366585e-01 1.14223278e+00 -3.51743132e-01 -1.01834178e+00 -2.76258320e-01 -2.08535701e-01 -2.52897978e-01 7.03003556e-02 -9.47887242e-01 -1.28408933e+00 9.33501124e-01 6.45338714e-01 1.17792457e-01 1.31576335e+00 3.77734393e-01 9.27511513e-01 -3.33020598e-01 2.68886536e-02 -5.90358377e-01 4.04110461e-01 -1.97129592e-01 1.20026278e+00 -1.06010842e+00 -1.29573941e-01 -3.01709324e-01 -1.19348836e+00 6.62392735e-01 7.80178308e-01 9.78084058e-02 1.89657941e-01 1.38944283e-01 3.44939739e-01 -3.67311060e-01 -8.90615106e-01 -1.38268441e-01 5.40746331e-01 5.61091959e-01 4.21302855e-01 1.37222618e-01 -3.88273805e-01 8.74544263e-01 -6.05849922e-02 -7.84336682e-03 4.22586322e-01 5.94500721e-01 -3.15884054e-01 -6.50664389e-01 -3.55133861e-01 8.32105517e-01 -7.11419821e-01 -1.40306249e-01 -1.04314357e-01 5.61699688e-01 2.25960031e-01 7.46590018e-01 1.99776180e-02 -3.01166773e-01 3.33319545e-01 -2.12626532e-01 2.17034116e-01 -7.40375042e-01 -4.46020037e-01 1.40050039e-01 -2.63764590e-01 -4.08330023e-01 -3.00906748e-01 -4.93885428e-01 -1.43360388e+00 3.76239359e-01 -1.46769196e-01 3.46042216e-01 3.21120799e-01 9.68364418e-01 3.73145252e-01 1.20909226e+00 6.61782503e-01 -4.60011303e-01 -2.97046155e-01 -1.13278294e+00 -5.95620632e-01 7.86890388e-01 5.63575685e-01 -5.50829887e-01 -5.60883470e-02 2.81630963e-01]
[15.034757614135742, -1.412042498588562]
54123187-8a80-492c-85a2-2a1992ce789a
do-multi-hop-question-answering-systems-know
2002.09919
null
https://arxiv.org/abs/2002.09919v2
https://arxiv.org/pdf/2002.09919v2.pdf
Do Multi-Hop Question Answering Systems Know How to Answer the Single-Hop Sub-Questions?
Multi-hop question answering (QA) requires a model to retrieve and integrate information from different parts of a long text to answer a question. Humans answer this kind of complex questions via a divide-and-conquer approach. In this paper, we investigate whether top-performing models for multi-hop questions understand the underlying sub-questions like humans. We adopt a neural decomposition model to generate sub-questions for a multi-hop complex question, followed by extracting the corresponding sub-answers. We show that multiple state-of-the-art multi-hop QA models fail to correctly answer a large portion of sub-questions, although their corresponding multi-hop questions are correctly answered. This indicates that these models manage to answer the multi-hop questions using some partial clues, instead of truly understanding the reasoning paths. We also propose a new model which significantly improves the performance on answering the sub-questions. Our work takes a step forward towards building a more explainable multi-hop QA system.
['Hwee Tou Ng', 'Yixuan Tang', 'Anthony K. H. Tung']
2020-02-23
null
https://aclanthology.org/2021.eacl-main.283
https://aclanthology.org/2021.eacl-main.283.pdf
eacl-2021-2
['multi-hop-question-answering']
['knowledge-base']
[-3.11267911e-04 8.96825671e-01 2.16082647e-01 -5.43411016e-01 -1.71129823e+00 -8.55966568e-01 3.30223233e-01 2.01252103e-01 -2.35096868e-02 9.42545652e-01 4.52566892e-01 -8.76403987e-01 -3.65354478e-01 -1.16954005e+00 -7.61239409e-01 8.48668963e-02 3.94460469e-01 1.14637756e+00 7.40037084e-01 -7.99730122e-01 2.49650761e-01 -1.60133451e-01 -1.24497867e+00 1.06352055e+00 1.31852853e+00 5.18164158e-01 6.94008619e-02 1.16207480e+00 -8.06747913e-01 1.27267754e+00 -6.95730805e-01 -7.33781815e-01 -1.99609324e-01 -1.00492907e+00 -1.91898596e+00 -4.09797847e-01 6.76274717e-01 -4.33103293e-01 -1.66541293e-01 6.21762514e-01 3.39306965e-02 2.61007994e-02 5.93469679e-01 -1.11466062e+00 -8.29136431e-01 6.16410017e-01 -2.24925932e-02 2.54649580e-01 9.47928488e-01 9.19699371e-02 1.56530702e+00 -6.62325025e-01 5.64983010e-01 1.55414319e+00 6.46508873e-01 6.40993953e-01 -9.18161273e-01 -3.32932211e-02 1.03663452e-01 7.85612524e-01 -7.99009085e-01 -1.13997921e-01 6.12688482e-01 -2.58126948e-02 1.14250708e+00 4.61031795e-01 2.01251507e-01 5.81616282e-01 4.39196318e-01 8.23929071e-01 1.08017027e+00 -4.65241551e-01 2.95361280e-02 -1.27449080e-01 8.39297473e-01 9.54448283e-01 7.73876533e-02 -4.67512220e-01 -2.82262176e-01 -3.55441183e-01 2.07321540e-01 -4.80416149e-01 -2.55629689e-01 3.07880938e-01 -1.07620943e+00 1.14602184e+00 4.92835879e-01 4.91355151e-01 -5.27132034e-01 1.49789685e-02 -2.25286260e-01 6.17660761e-01 -4.28872071e-02 9.92653251e-01 -9.65265512e-01 -1.10409223e-02 -7.33855665e-01 5.39716423e-01 1.48688900e+00 6.58845067e-01 9.67626214e-01 -6.10317469e-01 -4.12380576e-01 4.02058095e-01 2.58011937e-01 1.30258650e-01 2.01642618e-01 -1.58326530e+00 9.54231858e-01 9.55824792e-01 2.87461072e-01 -8.32475960e-01 -5.01618743e-01 -3.17054510e-01 -3.83639395e-01 -3.24790776e-01 8.59457612e-01 -2.31624246e-01 -7.13865519e-01 1.58697176e+00 3.98590624e-01 -3.44636589e-01 3.41140032e-01 6.50841534e-01 1.07403183e+00 9.86621439e-01 1.84576660e-02 1.32650375e-01 1.92950451e+00 -1.49822116e+00 -6.36010647e-01 -5.37830830e-01 6.77067399e-01 -6.29877508e-01 1.26773918e+00 2.14520901e-01 -1.34858012e+00 -7.36598074e-01 -7.56385028e-01 -8.09847236e-01 -3.04789573e-01 -8.47891271e-02 3.40411514e-01 1.59566507e-01 -1.09051013e+00 2.27446795e-01 -1.20988011e-01 -2.30212316e-01 9.30996239e-02 2.12566838e-01 -1.38827190e-01 -6.66735768e-01 -1.69732440e+00 1.18624878e+00 1.80530414e-01 -6.53788894e-02 -7.49514699e-01 -5.65034866e-01 -7.28763282e-01 3.82492483e-01 8.15680385e-01 -1.30822015e+00 1.75077128e+00 -3.12951714e-01 -1.09262681e+00 5.40780008e-01 -7.95654953e-01 -2.84070849e-01 1.03666939e-01 -2.90563464e-01 -2.24443078e-01 7.17317641e-01 5.29465377e-01 8.81918550e-01 5.46878397e-01 -1.30428028e+00 -6.67483807e-01 -4.05619502e-01 9.09094036e-01 1.57733243e-02 2.51748264e-01 -2.21425056e-01 -1.46599174e-01 -5.61141483e-02 2.96962976e-01 -6.31605327e-01 -2.59956688e-01 -3.44936162e-01 -4.68244225e-01 -6.45576894e-01 4.94704396e-01 -1.04584253e+00 1.26009297e+00 -1.19279099e+00 2.08737537e-01 -3.84251535e-01 5.55542171e-01 -2.57933941e-02 -6.42752409e-01 7.15259075e-01 2.11471051e-01 2.16755390e-01 -3.81952643e-01 -4.82610837e-02 1.90744489e-01 5.84605038e-01 -5.37584066e-01 -4.73287821e-01 5.91237724e-01 1.43122160e+00 -7.98781514e-01 -6.15187526e-01 -4.34437275e-01 -8.95881355e-02 -7.18692541e-01 3.71330380e-01 -1.06003380e+00 3.60938162e-01 -6.70327783e-01 4.87802297e-01 5.10554492e-01 -6.33796513e-01 -1.07106015e-01 1.55862933e-02 3.32294405e-01 8.76418769e-01 -6.43884003e-01 1.60950136e+00 -5.22179723e-01 5.89783192e-01 6.82658851e-02 -8.85959327e-01 6.69125855e-01 3.60871941e-01 -1.38791576e-01 -7.72874355e-01 -1.75598264e-01 3.36495608e-01 1.30304098e-01 -8.79271030e-01 6.28142715e-01 -3.69242579e-01 -1.45981506e-01 8.76549721e-01 9.12596881e-02 -3.96498501e-01 5.82696974e-01 5.60828090e-01 1.19855607e+00 -4.33484055e-02 1.35122448e-01 -3.85216773e-02 9.13479328e-01 6.81398630e-01 5.83059601e-02 1.02026606e+00 -4.67206500e-02 5.62312007e-01 8.42547059e-01 -5.12626827e-01 -6.47093832e-01 -9.18358743e-01 4.62717682e-01 1.17605388e+00 7.89385810e-02 -5.39960206e-01 -9.25110221e-01 -1.16905475e+00 -2.70692140e-01 1.29459548e+00 -5.68127215e-01 4.71091382e-02 -8.77926350e-01 -3.80109340e-01 7.27333486e-01 4.02092010e-01 5.60720623e-01 -1.08575070e+00 -6.31185710e-01 4.44646269e-01 -1.27694023e+00 -1.06204116e+00 -2.69132927e-02 4.72005866e-02 -9.92245793e-01 -1.45437109e+00 -6.38932884e-01 -8.57473493e-01 5.45090616e-01 2.89554507e-01 1.66420889e+00 5.34264207e-01 1.98087737e-01 5.47029376e-01 -4.64220941e-01 -2.57178456e-01 -5.76349378e-01 4.76793230e-01 -7.70025849e-01 -4.62590843e-01 5.37982941e-01 -2.78571337e-01 -5.32428503e-01 4.28388715e-01 -9.40044641e-01 -9.11985189e-02 6.01124644e-01 5.91347039e-01 3.55006635e-01 -4.39063571e-02 8.74310911e-01 -7.73836732e-01 1.26605558e+00 -8.02408695e-01 -1.72762364e-01 8.83938909e-01 -3.66799414e-01 5.89130998e-01 8.97578835e-01 -1.47468567e-01 -1.15891612e+00 -4.91291076e-01 -5.01854002e-01 4.90042269e-01 -2.81583995e-01 6.25566065e-01 -1.22113843e-02 2.93038934e-01 7.41582751e-01 1.12781085e-01 -3.30209076e-01 -4.39445496e-01 8.56034696e-01 2.95799077e-01 4.16624963e-01 -8.50310266e-01 8.54218602e-01 1.28783837e-01 -6.25487864e-02 -4.88442987e-01 -1.49395859e+00 -4.41872507e-01 -4.90036696e-01 7.26798326e-02 1.04439700e+00 -4.79614675e-01 -7.56157696e-01 -1.05355501e-01 -1.87627137e+00 -2.20522702e-01 -2.01125950e-01 -2.55653933e-02 -4.81009334e-01 5.65640628e-01 -6.99953318e-01 -4.85429138e-01 -3.52546662e-01 -8.71666193e-01 1.06431270e+00 4.34518546e-01 -5.63693166e-01 -9.67485845e-01 2.30968952e-01 1.21522284e+00 5.05178332e-01 -2.82635957e-01 1.72462916e+00 -9.10001814e-01 -9.32106137e-01 4.72037755e-02 -3.17188144e-01 -3.15286871e-03 -1.13538533e-01 -5.61796606e-01 -7.76449203e-01 2.88813591e-01 3.98811907e-01 -6.86557770e-01 8.72363508e-01 6.95293769e-02 8.47970068e-01 -6.69925213e-01 -1.37339337e-02 -1.45287380e-01 1.07913756e+00 -1.14431828e-01 7.47863770e-01 2.45902941e-01 2.61213005e-01 1.21324837e+00 5.36161125e-01 -3.09870869e-01 1.33061099e+00 1.43181644e-02 4.19170171e-01 2.18898371e-01 -1.14869684e-01 -3.87367934e-01 5.43730785e-05 8.83090019e-01 2.66031325e-01 -3.71982753e-01 -1.09916008e+00 8.29063177e-01 -1.72347641e+00 -8.96417439e-01 -8.10177922e-01 1.44000697e+00 9.27815735e-01 -7.02317059e-02 -7.78081417e-02 7.24847913e-02 2.62470156e-01 5.44903427e-02 -4.61111605e-01 -4.53650087e-01 -3.22006159e-02 4.15262163e-01 -5.22278011e-01 1.07448351e+00 -4.68086928e-01 1.02010512e+00 6.49739695e+00 4.18111742e-01 -3.09741467e-01 3.39494497e-02 5.20550013e-01 4.85787690e-01 -9.38013792e-01 4.12200004e-01 -8.24644804e-01 -2.09260434e-01 1.15136993e+00 -2.22473242e-03 4.11242813e-01 4.88447607e-01 -2.62181103e-01 -4.48907018e-01 -1.18918216e+00 2.83995509e-01 1.83098719e-01 -1.41476059e+00 5.45285225e-01 -2.62646645e-01 6.10073686e-01 -3.99538875e-01 -2.98349828e-01 6.77994967e-01 4.12780613e-01 -1.18652737e+00 3.86609972e-01 7.21063137e-01 1.75814047e-01 -4.95066255e-01 8.33675861e-01 9.11159515e-01 -1.03746116e+00 -3.16001505e-01 -4.17030513e-01 -2.65019387e-01 4.19081807e-01 2.87950158e-01 -7.63520896e-01 8.56003940e-01 4.32878107e-01 -1.98965847e-01 -9.88790333e-01 7.91981876e-01 -8.15445483e-01 5.65594494e-01 -6.22771233e-02 -2.83837855e-01 6.07224286e-01 -3.86069939e-02 1.38098016e-01 7.19241619e-01 4.38324302e-01 7.19668269e-01 -2.65130162e-01 1.11954474e+00 -1.68233559e-01 -2.21445024e-01 -1.16751716e-01 -1.84143618e-01 1.81264162e-01 1.02118492e+00 -2.28427947e-01 -5.44393301e-01 -5.34776151e-01 8.78953457e-01 5.80953121e-01 4.04141784e-01 -6.47733688e-01 -6.55830503e-01 1.36535138e-01 -1.58569422e-02 1.61224216e-01 -1.74564868e-01 -2.87418604e-01 -1.22090805e+00 2.95124382e-01 -1.43398118e+00 7.78423846e-01 -1.38910949e+00 -1.40060079e+00 9.54079151e-01 -4.07558568e-02 -6.42876744e-01 -7.11544156e-01 -5.06948650e-01 -6.74966335e-01 9.74891186e-01 -1.97829056e+00 -1.13839257e+00 -3.80474478e-01 6.69086337e-01 7.17626870e-01 4.20010000e-01 1.01008308e+00 -1.76754206e-01 1.02315567e-01 6.61188364e-02 -5.02575457e-01 5.40984347e-02 5.23064375e-01 -1.39688170e+00 5.49677849e-01 6.62037134e-01 3.07257652e-01 8.53576183e-01 6.21754944e-01 -4.56597596e-01 -1.31992722e+00 -6.61017418e-01 1.74355292e+00 -9.08938944e-01 6.73031807e-01 1.46181419e-01 -1.38062763e+00 7.25263774e-01 6.53229535e-01 -6.43206835e-01 7.42350936e-01 1.90381646e-01 -4.88888711e-01 1.39627695e-01 -1.06523061e+00 5.48315763e-01 6.61439478e-01 -7.49573052e-01 -1.85622072e+00 4.56952006e-01 1.19488716e+00 -3.99761088e-02 -5.11916101e-01 1.95365533e-01 2.36560941e-01 -1.12845290e+00 8.69866848e-01 -1.06011915e+00 9.83643353e-01 -3.25907916e-01 -9.25941467e-02 -1.17600513e+00 -1.39551640e-01 -5.87392449e-01 -3.75488549e-01 7.61155188e-01 8.79240811e-01 -6.35456622e-01 6.82852387e-01 6.88676775e-01 -8.80019367e-02 -8.75560164e-01 -8.89265895e-01 -2.04647824e-01 4.53737736e-01 -4.05857891e-01 7.29276001e-01 5.99208236e-01 -3.81775200e-02 9.56287265e-01 1.00921905e-02 3.32134992e-01 4.99430805e-01 6.34598076e-01 6.32280886e-01 -1.27774274e+00 -3.80339652e-01 -1.52493656e-01 6.10532343e-01 -1.71834445e+00 1.39646173e-01 -6.29713237e-01 6.77759647e-02 -2.46324873e+00 -9.01784152e-02 1.94932055e-02 2.49313965e-01 3.58845472e-01 -5.85099518e-01 -1.07499100e-01 2.14163274e-01 1.71588883e-02 -8.76388609e-01 4.06012803e-01 1.69196594e+00 -3.39324355e-01 2.29439616e-01 1.02912545e-01 -1.36150360e+00 7.11096764e-01 7.99331367e-01 -6.63022935e-01 -5.42939007e-01 -9.91878510e-01 5.86312592e-01 7.21952617e-01 4.62230235e-01 -7.34212220e-01 5.76431572e-01 -9.72795412e-02 1.46535486e-01 -8.55990291e-01 4.44921613e-01 -5.85247219e-01 -6.08476877e-01 5.97429335e-01 -5.28491855e-01 3.98194641e-01 1.04679346e-01 4.57805514e-01 -6.09747410e-01 -5.99084556e-01 2.56380290e-01 -7.16571569e-01 -3.64442885e-01 -1.20308094e-01 -5.26364744e-01 6.87654734e-01 6.78851664e-01 1.72856987e-01 -8.29194486e-01 -9.87477779e-01 -8.68846536e-01 7.57629991e-01 -2.69134283e-01 2.75606364e-01 6.55426919e-01 -8.25443387e-01 -9.22334075e-01 -2.55671829e-01 2.44649686e-02 6.68865740e-02 3.16005796e-01 4.68103528e-01 -5.84172368e-01 1.17821896e+00 3.19013745e-02 -2.06661001e-01 -1.01630509e+00 3.35554510e-01 4.51238006e-01 -9.00162816e-01 -4.93800566e-02 8.36278200e-01 6.96816444e-02 -9.19551134e-01 -1.12891316e-01 -6.25331461e-01 -4.11242068e-01 1.51515290e-01 5.94047725e-01 3.75631690e-01 -9.88741070e-02 -2.41077259e-01 -7.89847821e-02 8.30848813e-01 6.82471469e-02 -2.16479287e-01 1.17782044e+00 -3.04436952e-01 -5.12797594e-01 1.69933230e-01 9.06935513e-01 -6.71003833e-02 -6.14867926e-01 -3.39726090e-01 3.06154937e-01 -9.78231281e-02 -6.16735101e-01 -1.30976784e+00 -2.96095937e-01 1.33916306e+00 -4.15608764e-01 5.84346414e-01 1.07159400e+00 6.51217937e-01 1.41522622e+00 1.03195417e+00 4.25690830e-01 -6.37106240e-01 5.03078699e-01 1.00491512e+00 1.22755837e+00 -1.13339782e+00 -3.35169256e-01 -3.90165716e-01 -5.71175814e-01 1.32364964e+00 9.03376937e-01 2.26407126e-01 -1.56631824e-02 -2.64752388e-01 2.88224936e-01 -4.79381233e-01 -1.17799020e+00 -2.75109053e-01 4.91006881e-01 3.74479115e-01 1.48662508e-01 -2.93742687e-01 -2.27190688e-01 9.13200259e-01 -6.10459685e-01 -2.18373090e-01 5.67997098e-01 7.39182532e-01 -9.26135004e-01 -1.19473195e+00 -5.86067379e-01 2.91186959e-01 -3.48495811e-01 -2.78125376e-01 -9.83998299e-01 7.60595202e-01 -5.74252382e-02 1.71876395e+00 -4.49643046e-01 -1.45369381e-01 3.93116444e-01 7.15258002e-01 4.90316451e-01 -5.79864740e-01 -7.80855000e-01 -5.75264752e-01 4.26847160e-01 -4.42823738e-01 -1.75447434e-01 -2.21423848e-04 -1.41477513e+00 -2.24720344e-01 -1.20932966e-01 6.49384916e-01 2.54136890e-01 1.45108902e+00 4.93013144e-01 6.22726381e-01 1.72904298e-01 3.35785672e-02 -8.27706337e-01 -9.22764540e-01 1.89953759e-01 1.53861046e-01 5.96501827e-01 2.38684323e-02 -4.17289317e-01 -1.94511548e-01]
[11.105262756347656, 7.910545349121094]
82018524-de2d-4c44-be17-9fce7b49e546
chestx-ray8-hospital-scale-chest-x-ray
1705.02315
null
http://arxiv.org/abs/1705.02315v5
http://arxiv.org/pdf/1705.02315v5.pdf
ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseases
The chest X-ray is one of the most commonly accessible radiological examinations for screening and diagnosis of many lung diseases. A tremendous number of X-ray imaging studies accompanied by radiological reports are accumulated and stored in many modern hospitals' Picture Archiving and Communication Systems (PACS). On the other side, it is still an open question how this type of hospital-size knowledge database containing invaluable imaging informatics (i.e., loosely labeled) can be used to facilitate the data-hungry deep learning paradigms in building truly large-scale high precision computer-aided diagnosis (CAD) systems. In this paper, we present a new chest X-ray database, namely "ChestX-ray8", which comprises 108,948 frontal-view X-ray images of 32,717 unique patients with the text-mined eight disease image labels (where each image can have multi-labels), from the associated radiological reports using natural language processing. Importantly, we demonstrate that these commonly occurring thoracic diseases can be detected and even spatially-located via a unified weakly-supervised multi-label image classification and disease localization framework, which is validated using our proposed dataset. Although the initial quantitative results are promising as reported, deep convolutional neural network based "reading chest X-rays" (i.e., recognizing and locating the common disease patterns trained with only image-level labels) remains a strenuous task for fully-automated high precision CAD systems. Data download link: https://nihcc.app.box.com/v/ChestXray-NIHCC
['Mohammadhadi Bagheri', 'Xiaosong Wang', 'Le Lu', 'Yifan Peng', 'Ronald M. Summers', 'Zhiyong Lu']
2017-05-05
chestx-ray8-hospital-scale-chest-x-ray-1
http://openaccess.thecvf.com/content_cvpr_2017/html/Wang_ChestX-ray8_Hospital-Scale_Chest_CVPR_2017_paper.html
http://openaccess.thecvf.com/content_cvpr_2017/papers/Wang_ChestX-ray8_Hospital-Scale_Chest_CVPR_2017_paper.pdf
cvpr-2017-7
['multi-label-image-classification', 'lung-disease-classification']
['computer-vision', 'medical']
[ 1.93995774e-01 -4.47348803e-02 -4.44454670e-01 -5.26028335e-01 -1.63342977e+00 -4.56212580e-01 1.75710097e-01 2.91102529e-01 -4.51150447e-01 6.42343640e-01 1.32318020e-01 -7.44620502e-01 -5.88287711e-01 -7.69509494e-01 -5.68336487e-01 -7.92664707e-01 8.35687518e-02 1.03661025e+00 1.52799040e-01 3.76938283e-01 -3.67604166e-01 5.42849064e-01 -1.11552405e+00 5.74460328e-01 9.76667479e-02 1.12075174e+00 6.57012463e-01 1.14326775e+00 1.88594908e-01 1.35170138e+00 -3.27050239e-01 -1.53143093e-01 -1.57850347e-02 -3.50556672e-01 -1.11499989e+00 -1.01546809e-01 4.63964969e-01 -7.07225144e-01 -5.53086877e-01 6.93854332e-01 7.15564311e-01 -3.26467812e-01 6.49818540e-01 -3.46144915e-01 -7.38710046e-01 3.34735870e-01 -5.58500409e-01 6.62909746e-01 -1.94939319e-02 9.50921848e-02 7.81199157e-01 -7.52952337e-01 7.42511928e-01 5.33317447e-01 8.74568164e-01 5.77942491e-01 -3.49190414e-01 -3.69542360e-01 -6.93214059e-01 1.14717074e-01 -1.25052297e+00 2.50041842e-01 2.35906899e-01 -5.06901979e-01 5.54854453e-01 7.84357905e-01 4.55588251e-01 1.00218058e+00 6.19735777e-01 3.05961192e-01 1.21276665e+00 -5.59545934e-01 -5.60076050e-02 -3.31941545e-02 1.90977767e-01 1.33552468e+00 2.33275965e-01 2.59788837e-02 -1.31656304e-01 -3.24852586e-01 9.56631184e-01 7.23581433e-01 -4.32599306e-01 -6.41395003e-02 -1.62614644e+00 6.96005821e-01 7.84122765e-01 7.32472658e-01 -4.93339360e-01 2.02508241e-01 6.72981739e-01 2.50956491e-02 2.43475080e-01 2.92691618e-01 -4.60749507e-01 1.69952437e-01 -9.55459177e-01 -1.94006383e-01 4.13672000e-01 7.47226179e-01 1.35000110e-01 -6.20844662e-01 -2.73245633e-01 1.07288229e+00 1.75374314e-01 6.45063341e-01 8.63751054e-01 -7.91911662e-01 4.04643565e-01 4.44245815e-01 -5.30337274e-01 -4.90478188e-01 -1.07035112e+00 -5.60207665e-01 -1.38772333e+00 -1.20222956e-01 3.21952552e-01 -1.03877306e-01 -1.02350950e+00 1.17616260e+00 3.26804727e-01 -2.66797572e-01 -2.18331650e-01 1.10320199e+00 1.14833295e+00 2.25327447e-01 2.65664101e-01 -1.94710881e-01 2.10254145e+00 -9.19837117e-01 -5.24423063e-01 2.31006101e-01 1.20234513e+00 -5.71689665e-01 1.23744154e+00 1.39897689e-01 -1.02298176e+00 -4.20908540e-01 -7.95234025e-01 -2.14786783e-01 -4.43028480e-01 7.90431559e-01 5.63449085e-01 4.38015312e-01 -1.16048598e+00 3.11569214e-01 -9.98114228e-01 -5.53792179e-01 7.30110824e-01 4.85526264e-01 -5.77527642e-01 -3.84084612e-01 -9.47960556e-01 1.02079749e+00 2.51670331e-01 -2.26466253e-01 -8.42528045e-01 -9.25452173e-01 -4.07499224e-01 -1.49509773e-01 7.01007128e-01 -1.06532848e+00 1.61462379e+00 -2.66120315e-01 -9.31977272e-01 1.39994478e+00 2.54035503e-01 -1.60790861e-01 5.47803283e-01 -3.89174297e-02 -3.97447318e-01 7.59244204e-01 4.01819587e-01 6.42637551e-01 4.17855293e-01 -9.90795553e-01 -6.89038038e-01 -5.31123400e-01 -3.61015171e-01 1.51895717e-01 -3.47532302e-01 3.09223026e-01 -5.44627488e-01 -8.41492295e-01 1.08986944e-01 -1.00858402e+00 -2.51263678e-01 5.48274577e-01 -4.02118921e-01 -2.31937200e-01 8.47550988e-01 -7.87038267e-01 1.04880917e+00 -1.91577613e+00 -3.26431751e-01 8.87004510e-02 7.89102733e-01 7.54989237e-02 5.51547348e-01 6.96834642e-03 -4.95367348e-01 2.30660126e-01 -2.46785209e-01 -1.50263608e-01 -4.63729441e-01 4.20585424e-01 1.55027345e-01 6.58133805e-01 -8.67720246e-02 1.35069990e+00 -6.32019341e-01 -9.82260883e-01 5.37260413e-01 3.62975121e-01 -1.21170536e-01 3.03248972e-01 7.76830241e-02 7.30614126e-01 -5.93868792e-01 9.59148526e-01 1.05389632e-01 -1.16466296e+00 3.22020091e-02 -2.74897456e-01 1.17916144e-01 -2.00127482e-01 -4.68103856e-01 1.78021109e+00 -6.47998393e-01 3.32923353e-01 -4.05882262e-02 -6.82575285e-01 2.86346793e-01 6.14485383e-01 9.91052926e-01 -6.91689968e-01 3.33819866e-01 3.44149143e-01 -1.64017707e-01 -8.90692711e-01 -7.72165731e-02 -3.71314317e-01 -5.81677025e-03 8.20766926e-01 -8.21388438e-02 -1.44765273e-01 -2.21044630e-01 2.23942250e-01 1.60567439e+00 -7.36073494e-01 5.63404143e-01 -1.86942130e-01 5.42456508e-01 2.98195362e-01 -1.47524744e-01 1.04126143e+00 -2.74419099e-01 1.07042456e+00 -7.79946009e-03 -7.24601626e-01 -1.21144509e+00 -1.22851551e+00 -7.60679841e-01 9.61511135e-01 -3.14934552e-01 -2.37799287e-01 -4.03069854e-01 -7.72236884e-01 -2.31697276e-01 3.71947698e-02 -6.20729506e-01 1.87365010e-01 -9.25134540e-01 -6.61278427e-01 7.98883796e-01 8.54870558e-01 3.82073849e-01 -9.65584874e-01 -8.11421096e-01 -4.15071212e-02 -5.20322472e-03 -8.56345832e-01 -3.17275822e-01 5.96489668e-01 -7.54213214e-01 -1.32831383e+00 -1.42088032e+00 -9.44827497e-01 6.35398924e-01 1.39409959e-01 1.22733533e+00 4.10255373e-01 -8.61793935e-01 8.01042318e-01 -2.60288239e-01 -1.68748200e-01 -5.42089403e-01 2.08066627e-01 -5.58353066e-02 -6.71092272e-01 3.75023782e-02 -1.23302840e-01 -8.08260381e-01 2.66274437e-03 -9.76973474e-01 9.66100991e-02 1.07763803e+00 1.01223004e+00 1.26225448e+00 -3.90102826e-02 4.38698232e-01 -1.25513387e+00 5.72181761e-01 -6.30580425e-01 -5.36365323e-02 7.01242983e-01 -4.25348401e-01 -3.87061417e-01 7.46724665e-01 -8.58998522e-02 -7.86329925e-01 -3.71690579e-02 -4.30482298e-01 -5.46844184e-01 -6.91592753e-01 5.63679874e-01 6.20371759e-01 -1.64468244e-01 8.25175881e-01 1.73869386e-01 -1.74493492e-01 -4.19336766e-01 3.71222228e-01 9.10960793e-01 9.48155105e-01 -4.05444980e-01 3.32640469e-01 6.21050060e-01 3.59317183e-01 -6.37914002e-01 -1.15349889e+00 -8.09109867e-01 -8.68721783e-01 -4.11948413e-01 1.40596032e+00 -9.17662799e-01 -4.70871627e-01 1.46521851e-01 -9.01102960e-01 1.54102622e-02 -5.52326858e-01 7.21025109e-01 -5.47000289e-01 4.20127600e-01 -9.64866042e-01 -9.24373046e-02 -6.57001138e-01 -1.31334364e+00 1.56946814e+00 -1.15271039e-01 -2.04085916e-01 -8.61678064e-01 2.19901398e-01 5.64481795e-01 3.38949531e-01 8.24191645e-02 1.24536407e+00 -6.93432033e-01 -4.80154961e-01 -3.73297572e-01 -7.03485489e-01 2.37294853e-01 2.98333675e-01 -3.55024487e-01 -6.75375879e-01 -2.02747256e-01 4.59044248e-01 -6.41684949e-01 7.42789030e-01 6.26434922e-01 1.86108708e+00 6.11596555e-02 -4.56539184e-01 7.04361320e-01 1.46122968e+00 1.91202909e-01 6.04412593e-02 2.17859566e-01 9.78139281e-01 1.22427337e-01 1.75981730e-01 3.91226590e-01 4.05153006e-01 2.26344734e-01 2.78652072e-01 -3.67169857e-01 -4.63128299e-01 1.55387595e-01 -5.94565868e-01 1.16366744e+00 -5.48144318e-02 -1.96689323e-01 -1.73594475e+00 2.76917338e-01 -1.26210296e+00 -4.13317531e-01 -3.35616171e-01 1.50162506e+00 8.00305247e-01 -1.70385495e-01 -3.01131606e-01 3.16213444e-02 6.35489821e-01 3.19174491e-02 -5.25413275e-01 2.89909303e-01 6.42156228e-02 6.85846388e-01 6.65112793e-01 -9.94161442e-02 -1.46624804e+00 1.50356069e-01 5.85159254e+00 7.93160141e-01 -1.27061296e+00 7.35136151e-01 9.54573870e-01 -2.79527098e-01 1.97540730e-01 -6.90115035e-01 -3.58421355e-01 3.47161114e-01 9.85743225e-01 3.29422981e-01 -2.29749188e-01 9.76340830e-01 7.62390643e-02 -3.54768515e-01 -1.02123249e+00 1.26716065e+00 8.48641619e-02 -1.81667984e+00 -4.67848852e-02 7.49206692e-02 5.56962907e-01 6.10748649e-01 2.45119870e-01 -5.59601486e-02 -7.27022812e-02 -1.02214730e+00 2.60807276e-01 5.82556903e-01 1.64743245e+00 -1.92182764e-01 9.28556561e-01 3.05830568e-01 -9.19101655e-01 -4.03783135e-02 -8.25182647e-02 5.87038338e-01 1.24943487e-01 4.95440781e-01 -8.48195970e-01 6.90720439e-01 1.01556754e+00 5.48662782e-01 -7.83131778e-01 7.76822150e-01 1.12052560e-01 7.59194195e-01 -4.26821202e-01 2.52909988e-01 3.12500000e-01 3.14231664e-01 -4.81063388e-02 1.08586800e+00 4.15512323e-01 6.83805466e-01 1.73921153e-01 4.28499848e-01 -4.48261708e-01 1.30448893e-01 -6.92331851e-01 2.84972787e-01 6.62247613e-02 1.32641530e+00 -1.00200999e+00 -7.17227399e-01 -7.13599086e-01 5.33816040e-01 6.81053996e-02 -1.53153345e-01 -9.95774865e-01 2.98879057e-01 -3.78810763e-01 3.53698730e-01 -1.16115272e-01 -1.90932024e-02 -6.07533634e-01 -1.03653765e+00 1.92402024e-02 -6.56993091e-01 8.34279895e-01 -1.07926631e+00 -1.41814446e+00 7.32959926e-01 -1.78703487e-01 -1.21662033e+00 -1.92758799e-01 -8.35090876e-01 -3.03352535e-01 5.22528827e-01 -1.45020199e+00 -1.33599699e+00 -5.13166785e-01 8.40755641e-01 6.10043526e-01 -2.19654426e-01 1.06085587e+00 7.06201434e-01 -4.26132053e-01 4.29364145e-01 4.39749300e-01 2.29003996e-01 7.44832873e-01 -1.38290381e+00 -4.62460369e-01 1.57859057e-01 7.52231255e-02 2.00079679e-01 -2.65013158e-01 -5.29072046e-01 -1.13294041e+00 -1.37272251e+00 6.30098641e-01 -7.50683606e-01 5.43898702e-01 1.47631437e-01 -9.68105316e-01 6.60687506e-01 3.82398698e-03 5.20333827e-01 9.04200613e-01 -5.38158238e-01 2.32197382e-02 3.07477117e-01 -1.10658789e+00 1.17038392e-01 8.67762089e-01 -8.54900122e-01 -5.40965140e-01 1.17473161e+00 5.51995337e-01 -6.37444675e-01 -1.34611273e+00 6.58505559e-01 2.51159102e-01 -6.37319505e-01 1.15719867e+00 -4.64484721e-01 7.56223798e-01 1.92241594e-02 -2.03872845e-01 -5.32177508e-01 -2.45684296e-01 3.83019537e-01 4.29075863e-03 4.55513895e-01 1.74115151e-01 -3.85594547e-01 8.95179272e-01 3.58528823e-01 -4.26120669e-01 -1.32427037e+00 -1.17990649e+00 -3.11228275e-01 2.43648991e-01 -5.73706269e-01 1.04584359e-01 1.00107205e+00 -4.56272602e-01 -8.34141895e-02 1.35772645e-01 2.13535368e-01 3.54235709e-01 8.26365687e-03 8.43350068e-02 -9.39276159e-01 -4.13364142e-01 -2.85608470e-01 -2.15370134e-01 -6.47482693e-01 -3.16734940e-01 -1.18572581e+00 -1.96429744e-01 -1.52496147e+00 5.54579377e-01 -6.54870927e-01 -5.57934225e-01 5.36335170e-01 6.94810227e-02 4.04218853e-01 -1.07378058e-01 9.75052118e-01 -8.07137609e-01 3.32801379e-02 1.47708023e+00 -2.23619461e-01 4.95483398e-01 1.06793888e-01 -3.83807778e-01 9.89517629e-01 6.83110356e-01 -7.77683079e-01 -1.58949539e-01 -6.38626933e-01 2.85598785e-02 5.59521019e-01 6.59277201e-01 -1.24890602e+00 5.23034275e-01 2.33673960e-01 5.85788429e-01 -1.10958505e+00 2.02215403e-01 -1.00161612e+00 1.00465849e-01 7.32094049e-01 -4.70658630e-01 2.43432954e-01 -1.87450230e-01 5.16233802e-01 -4.66547459e-01 -1.62902802e-01 9.47878063e-01 -7.97672331e-01 -4.80886042e-01 4.49045360e-01 -4.17106360e-01 -8.47943947e-02 1.37413514e+00 8.09403732e-02 -4.82970893e-01 7.16522485e-02 -8.70840251e-01 -3.87527384e-02 4.43650857e-02 7.14944676e-02 6.84305310e-01 -1.07514346e+00 -6.04774296e-01 7.23197609e-02 2.46668592e-01 5.48465669e-01 6.50655210e-01 1.06493056e+00 -1.10231197e+00 9.13313150e-01 1.12343214e-01 -1.02130163e+00 -1.28283572e+00 5.48298359e-01 5.00123024e-01 -8.27278554e-01 -9.43264067e-01 8.93421710e-01 3.01700354e-01 -4.98956382e-01 1.58128902e-01 -7.01887250e-01 1.05239123e-01 -2.63356417e-01 4.01793391e-01 4.43846621e-02 5.24886489e-01 -6.13902390e-01 -4.02104646e-01 7.31720626e-01 -4.63929892e-01 2.51588315e-01 1.47929049e+00 -5.74377291e-02 -1.55860662e-01 3.47396910e-01 1.45120609e+00 -5.29951513e-01 -2.94964433e-01 -2.36728728e-01 -2.10036244e-02 -6.84901848e-02 1.69884682e-01 -1.08808768e+00 -1.13382113e+00 1.10757101e+00 1.10078919e+00 6.13502823e-02 1.17768133e+00 8.73110294e-01 8.89812291e-01 6.97988868e-01 3.20940465e-01 -6.24311447e-01 1.24998301e-01 1.03649795e-01 6.83412075e-01 -1.59090793e+00 1.98293462e-01 -2.66134441e-01 -5.30232489e-01 1.18752599e+00 5.37030280e-01 1.80404395e-01 8.78735363e-01 3.15052897e-01 3.23422879e-01 -9.74222124e-01 -6.37624264e-01 4.22984846e-02 2.10079830e-02 4.06188875e-01 4.57164377e-01 4.11223710e-01 -1.59963176e-01 7.07709193e-01 7.19665457e-03 1.25132784e-01 1.56705230e-01 1.19868565e+00 -4.04298365e-01 -8.28334868e-01 -5.84685683e-01 1.21086133e+00 -9.03874695e-01 -1.05439641e-01 -2.74584919e-01 1.01441038e+00 4.01716262e-01 1.98433205e-01 -1.30347893e-01 -8.06225017e-02 2.63787121e-01 -8.23488981e-02 1.80362448e-01 -8.68078828e-01 -6.63745105e-01 -2.52857506e-01 -3.23918849e-01 -2.59308636e-01 -4.59506333e-01 -3.71783823e-01 -1.35128701e+00 1.65938288e-02 -3.58568013e-01 -2.78211385e-01 5.31186104e-01 8.97730887e-01 7.65574723e-02 9.43508804e-01 3.38738024e-01 -2.25745991e-01 -5.70084512e-01 -6.89598620e-01 -5.09813845e-01 2.72833318e-01 3.87025744e-01 -5.47571063e-01 -6.29097000e-02 1.74261019e-01]
[15.255321502685547, -2.0474720001220703]
546a01ca-2fbd-4c76-bc2a-bbde5e73cec3
reinforcement-learning
2005.14419
null
https://arxiv.org/abs/2005.14419v2
https://arxiv.org/pdf/2005.14419v2.pdf
Reinforcement Learning
Reinforcement learning (RL) is a general framework for adaptive control, which has proven to be efficient in many domains, e.g., board games, video games or autonomous vehicles. In such problems, an agent faces a sequential decision-making problem where, at every time step, it observes its state, performs an action, receives a reward and moves to a new state. An RL agent learns by trial and error a good policy (or controller) based on observations and numeric reward feedback on the previously performed action. In this chapter, we present the basic framework of RL and recall the two main families of approaches that have been developed to learn a good policy. The first one, which is value-based, consists in estimating the value of an optimal policy, value from which a policy can be recovered, while the other, called policy search, directly works in a policy space. Actor-critic methods can be seen as a policy search technique where the policy value that is learned guides the policy improvement. Besides, we give an overview of some extensions of the standard RL framework, notably when risk-averse behavior needs to be taken into account or when rewards are not available or not known.
['Olivier Buffet', 'Paul Weng', 'Olivier Pietquin']
2020-05-29
null
null
null
null
['board-games']
['playing-games']
[ 1.31315172e-01 3.59738052e-01 -8.11640680e-01 1.23405538e-01 -5.32727778e-01 -5.54308116e-01 7.20537543e-01 1.83717966e-01 -7.61623800e-01 1.37344193e+00 -1.68151975e-01 -3.15807879e-01 -3.08365941e-01 -7.80697882e-01 -5.26603401e-01 -1.11814332e+00 -1.37354836e-01 4.73056704e-01 1.62627429e-01 -4.15053248e-01 4.37438846e-01 4.51587230e-01 -1.44423378e+00 -4.35100168e-01 6.46718919e-01 1.06818700e+00 3.41415524e-01 8.61572027e-01 1.13986887e-01 1.14705074e+00 -6.36478066e-01 3.63405049e-02 3.21382165e-01 -5.18961191e-01 -7.94460833e-01 1.32433891e-01 -5.95018685e-01 -5.53812444e-01 -1.40137643e-01 1.09309757e+00 2.62601912e-01 5.78347087e-01 5.13068616e-01 -1.30009115e+00 -3.86351906e-02 4.09067303e-01 -1.90275878e-01 7.78371543e-02 3.31766784e-01 5.61418235e-01 9.00262594e-01 -2.48011038e-01 6.08089089e-01 1.34513259e+00 -5.77782504e-02 8.98674965e-01 -1.18837297e+00 -3.35171282e-01 5.67750156e-01 2.97382414e-01 -8.15877557e-01 -1.86828494e-01 4.87519205e-01 -2.70884931e-01 8.11712027e-01 -9.92473662e-02 9.57934618e-01 9.38985765e-01 4.44646358e-01 9.73130584e-01 1.30595207e+00 -5.16084492e-01 1.01130152e+00 -1.99885461e-02 -5.91210723e-01 4.19586509e-01 6.33082092e-02 8.72592270e-01 -1.04296403e-02 -2.63036728e-01 8.94194782e-01 -1.22187905e-01 -1.81840025e-02 -6.57001317e-01 -1.00936186e+00 1.07393384e+00 2.61091799e-01 2.05782503e-01 -9.65023577e-01 4.87053245e-01 4.48559582e-01 7.30952621e-01 6.57385588e-02 6.78584158e-01 -4.49396670e-01 -3.48860711e-01 -4.37003493e-01 7.02087998e-01 6.46358490e-01 2.48227388e-01 7.04203725e-01 5.81014574e-01 -3.85118961e-01 3.97114009e-01 3.13553065e-01 7.01127827e-01 4.66035008e-01 -1.45904970e+00 3.09478700e-01 1.74804270e-01 8.18116128e-01 -4.07971889e-01 -1.75038338e-01 5.35790250e-03 -4.11676407e-01 1.18268359e+00 3.78555894e-01 -6.07609510e-01 -5.23474574e-01 1.66874468e+00 4.75880921e-01 3.53910983e-01 4.59592879e-01 7.83636212e-01 -6.84834123e-02 6.70086384e-01 6.59869164e-02 -8.24498594e-01 8.05235744e-01 -8.20765913e-01 -8.15066755e-01 -3.44864637e-01 1.57138839e-01 -2.47270033e-01 6.34824157e-01 7.10049450e-01 -1.17173064e+00 -2.53824115e-01 -9.76925671e-01 7.81798422e-01 -8.05728734e-02 -1.77877411e-01 2.86813766e-01 3.61509293e-01 -1.01603889e+00 1.01309609e+00 -9.38922226e-01 -1.13895901e-01 4.61387224e-02 4.47186112e-01 7.44789243e-02 1.80255979e-01 -1.18002474e+00 1.25838661e+00 6.52949810e-01 -1.67766303e-01 -1.57166946e+00 -1.36897698e-01 -6.40598953e-01 -1.28389060e-01 1.25571549e+00 -3.93204182e-01 2.07966113e+00 -1.18806708e+00 -2.52802038e+00 2.46488616e-01 1.14468202e-01 -7.34456599e-01 6.36722147e-01 -1.72685459e-01 -2.27072090e-01 -6.58866293e-06 -9.86109488e-03 2.99174964e-01 1.22767699e+00 -1.09644961e+00 -9.86517072e-01 -1.64608717e-01 5.14863193e-01 6.01822495e-01 3.07412028e-01 1.15233995e-02 1.54705331e-01 -3.26411784e-01 -6.06368542e-01 -1.07849216e+00 -8.38348627e-01 -1.32691264e-01 -1.92104638e-01 -4.12777573e-01 5.60651422e-01 -7.55711347e-02 1.04791117e+00 -1.64375114e+00 4.85713780e-01 3.69652331e-01 -1.80800289e-01 4.59496528e-01 -1.13763198e-01 7.03036606e-01 1.54211149e-01 -1.64870620e-01 -2.82974057e-02 1.04952164e-01 -3.18432152e-02 5.97523808e-01 -3.35734278e-01 4.12500739e-01 4.97056171e-02 6.91382587e-01 -1.32439387e+00 3.03037055e-02 5.19720793e-01 -3.37900333e-02 -3.72145861e-01 5.04135489e-01 -7.44331181e-01 9.05679524e-01 -1.05287921e+00 1.86886519e-01 -5.27341617e-03 6.58405200e-02 4.15417373e-01 8.27448905e-01 -5.18050075e-01 2.53355682e-01 -1.43878305e+00 1.10724103e+00 -5.05881965e-01 1.57459497e-01 2.66630232e-01 -1.25222516e+00 8.41783643e-01 6.47859156e-01 7.32882679e-01 -7.36558616e-01 3.36143017e-01 1.80800214e-01 -1.23533018e-01 -4.32565331e-01 4.11772132e-01 -1.43895268e-01 3.44017558e-02 6.70242906e-01 -2.96995103e-01 -2.78270900e-01 2.45315701e-01 -2.72361904e-01 1.16819513e+00 2.54562259e-01 1.13907039e+00 3.90892737e-02 8.70836318e-01 9.67440233e-02 6.04756474e-01 9.87116396e-01 -3.03960979e-01 -3.48602623e-01 7.15134680e-01 -5.02957761e-01 -9.25372481e-01 -7.18806624e-01 5.76236963e-01 1.17367089e+00 1.53734401e-01 -1.03692506e-02 -4.26354825e-01 -7.90912271e-01 1.10507302e-01 8.64497900e-01 -6.37044072e-01 -2.80349791e-01 -7.30248570e-01 -2.20610216e-01 -1.22122139e-01 2.92483687e-01 4.27616388e-01 -1.60111153e+00 -1.08350015e+00 5.77310860e-01 2.97034681e-01 -6.10741079e-01 -1.12713285e-01 1.23727471e-01 -9.87719476e-01 -1.13452613e+00 -7.01524913e-01 -1.83692142e-01 4.29597199e-01 -3.12471427e-02 9.31746662e-01 3.60699184e-02 1.98244408e-01 6.54667735e-01 -2.86734432e-01 -6.70574248e-01 -6.73824430e-01 -2.75617987e-01 3.36693138e-01 4.45440523e-02 -1.96882471e-01 -1.76691368e-01 -5.58309793e-01 2.32889459e-01 -5.90563655e-01 -4.03432697e-01 5.48449397e-01 8.89606595e-01 8.48196864e-01 1.73406690e-01 6.35999501e-01 -6.93846345e-01 9.71284628e-01 -4.32184428e-01 -1.17221367e+00 2.55969256e-01 -7.84003973e-01 4.95430589e-01 8.84576738e-01 -5.88901699e-01 -9.21918035e-01 1.39926955e-01 -1.27691850e-01 -4.09371376e-01 -1.63088635e-01 2.64857858e-01 7.53071010e-02 2.77377665e-03 5.67940474e-01 2.40426049e-01 3.68400812e-01 -2.94981867e-01 4.03193861e-01 3.49998444e-01 2.87345231e-01 -6.89466953e-01 6.07630670e-01 1.10387862e-01 8.86928514e-02 -3.90779257e-01 -6.09641254e-01 -3.32618207e-01 -4.19415355e-01 -4.99139190e-01 7.13782072e-01 -5.65045774e-01 -1.22751307e+00 1.50715306e-01 -8.32505465e-01 -9.61083770e-01 -8.67956519e-01 5.87601423e-01 -1.22905838e+00 -1.63034305e-01 -1.40360221e-01 -1.36064279e+00 -4.75780517e-02 -1.19833291e+00 5.75976849e-01 6.19330287e-01 6.84807524e-02 -1.06199431e+00 4.32259053e-01 -3.03272724e-01 4.52854455e-01 3.50860506e-01 6.36832714e-01 -5.27037323e-01 -6.03812039e-01 4.38764580e-02 5.43459296e-01 2.42105633e-01 8.12451094e-02 -2.26793677e-01 -5.82417488e-01 -6.47008598e-01 4.58272025e-02 -5.48374772e-01 3.39297146e-01 6.32541656e-01 8.62223506e-01 -7.08461761e-01 -1.71651900e-01 -6.79211915e-02 1.49322438e+00 8.90933633e-01 4.12124962e-01 7.45257139e-01 -2.20760964e-02 2.57254094e-01 1.20334303e+00 8.55206907e-01 1.53068036e-01 7.18590081e-01 8.92626882e-01 2.71020472e-01 4.91334975e-01 -2.63003826e-01 5.97139716e-01 1.56048657e-02 -4.54591602e-01 -8.75435024e-02 -4.11681473e-01 1.12096667e-01 -2.23241925e+00 -1.13260615e+00 7.31511533e-01 2.56505585e+00 7.65497684e-01 1.60691038e-01 5.76978743e-01 1.09393410e-01 5.76892018e-01 3.57502326e-02 -1.29844928e+00 -7.56530583e-01 3.97407889e-01 8.96472409e-02 6.29931569e-01 6.94011152e-01 -8.17279220e-01 1.13718748e+00 6.95058012e+00 6.53233051e-01 -1.26981139e+00 -8.95492956e-02 4.63106841e-01 3.66391502e-02 1.52948245e-01 -5.38473390e-02 -6.79538786e-01 2.88784772e-01 1.05962169e+00 -5.26330292e-01 1.12950420e+00 1.11394680e+00 7.88435102e-01 -5.12262344e-01 -9.70456362e-01 6.29188895e-01 -5.60070992e-01 -1.12494981e+00 -4.89550889e-01 1.60174102e-01 8.38879704e-01 -1.64647207e-01 2.18809415e-02 7.25960374e-01 1.11275840e+00 -9.12450492e-01 6.69581890e-01 5.08210123e-01 5.65982282e-01 -1.02416313e+00 5.35125792e-01 7.36824155e-01 -1.05691373e+00 -6.13148153e-01 -3.20822805e-01 -3.87281120e-01 4.66032103e-02 -1.99166499e-03 -8.38650942e-01 3.07289034e-01 2.39370033e-01 6.14242613e-01 1.05023019e-01 1.11327970e+00 -8.71357203e-01 4.47921813e-01 6.19569002e-03 -5.07148385e-01 6.80391490e-01 -3.13985765e-01 7.41398215e-01 4.66276497e-01 2.06227973e-01 2.37434223e-01 7.01806247e-01 5.37102282e-01 3.22641432e-01 1.03070721e-01 -7.31260419e-01 1.08583480e-01 3.80325824e-01 1.04853153e+00 -6.21524513e-01 -3.42394739e-01 -1.22019149e-01 6.47441566e-01 3.18915844e-01 3.93279314e-01 -5.64855099e-01 -1.67402253e-02 1.04318082e+00 -2.89747924e-01 3.45372438e-01 -1.87118933e-01 3.02479625e-01 -8.11681330e-01 -4.55402106e-01 -9.67359006e-01 4.20424074e-01 -4.86275941e-01 -6.99728727e-01 3.82794768e-01 1.01739682e-01 -1.39187324e+00 -1.11336327e+00 -4.06771839e-01 -7.41859794e-01 5.45131385e-01 -1.68460476e+00 -9.72105339e-02 3.34290683e-01 6.87767625e-01 7.99360514e-01 -4.52227712e-01 6.97325706e-01 -4.37467903e-01 -5.27176321e-01 8.60562697e-02 5.62267005e-01 -2.07771972e-01 2.82144338e-01 -1.47154748e+00 -7.08694011e-02 5.28192878e-01 -3.35580349e-01 5.97924516e-02 8.87463748e-01 -5.18047035e-01 -1.36751509e+00 -8.50359797e-01 2.94364512e-01 1.63587287e-01 7.85221815e-01 1.97724521e-01 -6.22192860e-01 4.93133187e-01 1.32707834e-01 -1.25023695e-02 -3.67330387e-02 -3.52832973e-01 3.54432523e-01 -1.86403051e-01 -1.29371130e+00 7.44018197e-01 4.34809327e-01 -1.01412423e-01 -4.02048796e-01 1.95567459e-01 5.38571894e-01 -5.34066021e-01 -6.71175838e-01 5.16967140e-02 3.94337744e-01 -9.51141000e-01 8.22317958e-01 -9.91788745e-01 -7.81516582e-02 -2.42643788e-01 1.68176830e-01 -2.01086640e+00 -3.35280806e-01 -1.24804890e+00 -3.73396754e-01 5.29212415e-01 1.31715551e-01 -7.65329123e-01 7.65435338e-01 6.59827650e-01 1.93807214e-01 -9.38739002e-01 -1.03895116e+00 -1.10143864e+00 1.28041595e-01 -3.48260313e-01 6.88308954e-01 3.50371122e-01 1.57590918e-02 8.79006609e-02 -5.74985504e-01 -4.03761342e-02 4.20206577e-01 -2.96679456e-02 7.15970516e-01 -9.26455915e-01 -4.55529988e-01 -5.97608864e-01 -5.26407920e-02 -1.10795927e+00 3.35558742e-01 -2.36956939e-01 3.64785314e-01 -1.40444338e+00 -3.84033084e-01 -4.16726470e-01 -3.71964008e-01 5.18819809e-01 -4.16476885e-03 -5.30057847e-01 4.68375623e-01 2.06128165e-01 -8.50020885e-01 6.69564188e-01 1.51010776e+00 1.75352804e-02 -6.67474329e-01 7.66423941e-01 -4.88213718e-01 7.80153692e-01 1.13667750e+00 -5.05085111e-01 -4.98772979e-01 1.28857777e-01 1.95385665e-01 8.80921781e-01 -1.34200361e-02 -7.00977504e-01 1.12998523e-01 -1.02309811e+00 -8.44758376e-02 -1.14563433e-02 3.09877187e-01 -7.86821604e-01 -2.14993745e-01 1.06524754e+00 -7.25862801e-01 2.57478774e-01 -1.78328335e-01 8.98719907e-01 -4.01486978e-02 -6.20226860e-01 9.29462612e-01 -3.49980175e-01 -9.66291249e-01 3.54389012e-01 -9.80089486e-01 1.38408676e-01 1.35822368e+00 1.35965273e-01 5.49595691e-02 -7.53999233e-01 -8.51872742e-01 4.87741917e-01 1.49448484e-01 3.53671163e-01 6.07200027e-01 -1.29181457e+00 -5.38769662e-01 -2.10845605e-01 -2.80180424e-01 -3.20447177e-01 -1.92549750e-01 5.59292734e-01 -2.18244847e-02 5.06686270e-01 -2.50628322e-01 -2.05588371e-01 -1.02714026e+00 7.11833954e-01 5.17484248e-01 -8.03759992e-01 -5.92240334e-01 2.73426622e-01 -2.10217386e-01 -7.26053193e-02 3.75010490e-01 -3.16544086e-01 -7.03296185e-01 -2.83167720e-01 9.31600332e-01 6.40120625e-01 -3.18236828e-01 -4.02664989e-01 -9.33430996e-03 4.34197634e-01 2.47869313e-01 -5.87797284e-01 1.12863338e+00 -1.32602766e-01 2.39711612e-01 6.33137047e-01 5.39275646e-01 -4.82793778e-01 -1.83415639e+00 -3.44184279e-01 1.72783762e-01 -4.28701580e-01 3.50247361e-02 -8.49141061e-01 -9.39064562e-01 4.27828014e-01 5.58105469e-01 4.64460641e-01 1.01408279e+00 -3.73150945e-01 3.96999568e-01 7.16916442e-01 7.97764659e-01 -1.50475335e+00 3.25844258e-01 7.53110349e-01 9.31918561e-01 -1.27802885e+00 -4.67101596e-02 2.52862513e-01 -1.09904182e+00 1.14148247e+00 5.16280234e-01 -4.63151634e-01 4.96524781e-01 1.02607369e-01 -1.03462622e-01 2.17066079e-01 -1.09336054e+00 -6.27804518e-01 -1.89126115e-02 8.36071730e-01 -1.87501863e-01 1.98299453e-01 -2.49550849e-01 -1.36010826e-01 7.49004781e-02 2.08357915e-01 5.68917394e-01 1.07485914e+00 -8.61162066e-01 -1.41246617e+00 -5.73831618e-01 2.72400767e-01 -3.92811298e-01 5.10153651e-01 -1.90852463e-01 6.97014272e-01 -1.83497861e-01 1.09782469e+00 -5.72572201e-02 -3.85485925e-02 3.87785256e-01 -2.40739465e-01 4.91398692e-01 -6.58078611e-01 -4.91310984e-01 5.68823293e-02 -2.54165884e-02 -9.94946599e-01 -4.96381611e-01 -7.52411723e-01 -1.51263011e+00 -1.05770744e-01 8.47289264e-02 3.65952283e-01 6.73141778e-01 1.14091372e+00 -6.68300167e-02 5.07129312e-01 1.10379803e+00 -9.71476674e-01 -9.95959461e-01 -4.79605496e-01 -5.95367432e-01 -1.19576901e-01 5.84019065e-01 -9.70194697e-01 -6.26206249e-02 -4.69162613e-01]
[4.168280124664307, 2.1000585556030273]
28a2dae4-c332-47ed-a6d2-dc8ad3207b7a
clickbait-detection-using-word-embeddings
1710.02861
null
http://arxiv.org/abs/1710.02861v1
http://arxiv.org/pdf/1710.02861v1.pdf
Clickbait detection using word embeddings
Clickbait is a pejorative term describing web content that is aimed at generating online advertising revenue, especially at the expense of quality or accuracy, relying on sensationalist headlines or eye-catching thumbnail pictures to attract click-throughs and to encourage forwarding of the material over online social networks. We use distributed word representations of the words in the title as features to identify clickbaits in online news media. We train a machine learning model using linear regression to predict the cickbait score of a given tweet. Our methods achieve an F1-score of 64.98\% and an MSE of 0.0791. Compared to other methods, our method is simple, fast to train, does not require extensive feature engineering and yet moderately effective.
['Vijayasaradhi Indurthi', 'Subba Reddy Oota']
2017-10-08
null
null
null
null
['clickbait-detection']
['natural-language-processing']
[-2.99427301e-01 -5.35656102e-02 -8.50564361e-01 -4.09833014e-01 -1.08619404e+00 -6.03771985e-01 8.86527836e-01 4.34323221e-01 -4.90259171e-01 5.06138265e-01 2.89306343e-01 -7.45767474e-01 -1.15073398e-01 -7.31603444e-01 -6.73495173e-01 -7.76400790e-02 -2.23362163e-01 1.96028844e-01 3.62440050e-01 -2.28538007e-01 6.73783898e-01 4.41519823e-03 -1.33832252e+00 2.94314265e-01 4.10938799e-01 1.67294002e+00 -3.21954712e-02 4.59088713e-01 -4.93115723e-01 9.78049457e-01 -6.16894841e-01 -8.57487977e-01 2.68360917e-02 -1.58314630e-01 -3.22123349e-01 -1.14704728e-01 6.61852241e-01 -2.49582097e-01 -5.20158589e-01 1.05225599e+00 -2.29808185e-02 -3.80929947e-01 4.06741470e-01 -7.93219686e-01 -1.01427722e+00 5.79448164e-01 -8.93228352e-01 6.83955550e-01 3.05666745e-01 -2.11095333e-01 1.67291760e+00 -7.88827360e-01 6.76838994e-01 9.92251039e-01 3.21626991e-01 4.90857102e-02 -1.40616345e+00 -7.93356061e-01 1.38337269e-01 -6.56509697e-02 -1.05264950e+00 -1.49621278e-01 7.76900172e-01 -5.76759219e-01 2.22302467e-01 5.59917450e-01 7.77377784e-01 1.28107679e+00 5.00742316e-01 1.07476199e+00 1.19190180e+00 -4.12806273e-02 2.40027234e-01 5.21148086e-01 3.77109438e-01 5.77484787e-01 2.24949747e-01 -7.32230321e-02 -5.93535900e-01 -5.24332225e-01 4.97894287e-01 2.19381660e-01 2.39577055e-01 7.14971796e-02 -7.22315669e-01 1.75636518e+00 9.91049469e-01 1.99040204e-01 -5.37596524e-01 2.70845711e-01 2.70908531e-02 3.18787128e-01 8.91760647e-01 7.26660073e-01 -1.61369309e-01 -1.06012151e-01 -8.44961822e-01 6.45163357e-01 7.90694058e-01 4.16013569e-01 5.54107487e-01 -1.66293964e-01 1.84508488e-02 8.96222949e-01 1.88621834e-01 6.20379090e-01 6.97624862e-01 -4.88921165e-01 2.54959285e-01 5.04182458e-01 2.65239477e-01 -1.48871040e+00 -1.69815853e-01 -7.95523703e-01 1.06472597e-02 -1.81805849e-01 3.16403508e-01 -1.46786511e-01 -6.57795548e-01 1.19217193e+00 -1.14698268e-01 -5.57570279e-01 -7.25120544e-01 8.95638287e-01 4.97146040e-01 9.21722293e-01 2.83167750e-01 -9.98737365e-02 1.46681559e+00 -7.01852381e-01 -7.95621455e-01 -5.75335026e-01 4.60601985e-01 -1.08396482e+00 1.32319379e+00 3.79883200e-01 -1.17754292e+00 2.34633833e-02 -9.40902710e-01 8.75141174e-02 -7.70928860e-01 -1.15658537e-01 1.04116666e+00 5.37963986e-01 -2.56380856e-01 5.16467690e-01 -2.95926958e-01 9.41350162e-02 7.70042479e-01 -1.95293576e-01 2.42514804e-01 1.96894005e-01 -1.05160785e+00 7.43914723e-01 -1.57614350e-01 -5.49471796e-01 -2.71197647e-01 -8.26588750e-01 -4.24726665e-01 -5.97646013e-02 5.04588425e-01 -1.17564946e-02 1.50255895e+00 -1.14674437e+00 -1.16605759e+00 6.37495875e-01 5.98776229e-02 -7.28638172e-01 4.16820288e-01 -4.04796124e-01 -9.21815574e-01 4.88042869e-02 1.91407263e-01 3.68711293e-01 1.02394617e+00 -5.83168149e-01 -7.97076821e-01 -2.60042250e-01 -5.67112193e-02 -3.60564291e-01 -5.08733869e-01 3.18252146e-01 -1.33814022e-01 -8.15090299e-01 2.97708839e-01 -7.71970868e-01 -1.01997241e-01 -9.03861895e-02 -3.97025675e-01 -3.47501427e-01 8.02338004e-01 -8.30076039e-01 1.55944324e+00 -1.98649788e+00 -5.84832370e-01 5.19506931e-01 4.75844085e-01 9.03883129e-02 1.48000151e-01 4.38872814e-01 4.26335514e-01 4.43921149e-01 6.94783807e-01 7.79227793e-01 1.60625339e-01 -4.72342670e-01 -4.28076744e-01 2.11042330e-01 1.24753378e-01 1.02265894e+00 -8.33071530e-01 -1.41091153e-01 -2.72387505e-01 2.04417348e-01 -6.61809325e-01 -2.44684502e-01 -5.21621644e-01 -2.68449873e-01 -9.38929200e-01 6.15629494e-01 3.17540586e-01 -8.80914211e-01 -5.05943075e-02 -8.44250545e-02 -1.76543742e-01 9.50616658e-01 -5.39020836e-01 8.04448485e-01 -4.79884207e-01 9.27208424e-01 -2.52751589e-01 -3.40371430e-01 1.04619622e+00 -3.07028919e-01 1.67523369e-01 -1.29548860e+00 3.07163775e-01 3.55146319e-01 -4.86175045e-02 -4.02250201e-01 4.57198769e-01 -7.30420500e-02 -4.01628107e-01 5.11840463e-01 -4.40503657e-01 3.52077901e-01 8.49056095e-02 5.14845192e-01 1.05084527e+00 -5.56681633e-01 2.35211566e-01 -3.06814015e-01 5.13074081e-03 5.11850901e-02 -9.09644812e-02 6.33613288e-01 6.09064475e-02 -1.39590353e-01 6.16511643e-01 -6.96196914e-01 -1.01171851e+00 -1.01683331e+00 -1.40431598e-01 1.45560420e+00 -8.61166269e-02 -6.04371130e-01 -2.26421073e-01 -6.27090871e-01 3.91313672e-01 1.07064104e+00 -5.67005754e-01 9.26134437e-02 -2.37415791e-01 -4.36829567e-01 -1.08552560e-01 1.77476451e-01 4.32826996e-01 -4.55094993e-01 -2.24506229e-01 3.60589445e-01 -2.93017190e-04 -8.83028150e-01 -6.67621553e-01 -9.40006152e-02 -5.76304078e-01 -7.58631408e-01 -7.80159354e-01 -6.97476387e-01 3.70215893e-01 2.74956584e-01 1.10315549e+00 -2.38319919e-01 -3.17692459e-01 -1.24240950e-01 -2.22335100e-01 -4.58512366e-01 -1.16102599e-01 1.56729877e-01 -2.43771791e-01 2.19795421e-01 8.52117181e-01 -1.23384319e-01 -8.16609383e-01 4.64371145e-01 -6.95658028e-01 -2.72074074e-01 5.20081162e-01 6.67926788e-01 1.42906532e-01 -2.72979289e-01 6.29163265e-01 -1.00467205e+00 1.04666960e+00 -7.57253528e-01 -7.70064890e-01 -1.91176951e-01 -7.10164428e-01 -1.23161599e-01 2.30823070e-01 -8.32184792e-01 -4.57215935e-01 -3.37520450e-01 3.29413004e-02 2.83893257e-01 6.06622756e-01 7.52674043e-01 7.24265575e-01 -1.30952016e-01 8.92888069e-01 1.41876880e-02 1.05276845e-01 -6.15549088e-01 3.65079135e-01 7.99590290e-01 -5.76984882e-02 2.68913835e-01 8.30439568e-01 2.35252380e-01 -5.68594575e-01 -8.50657105e-01 -1.31877840e+00 -5.06924987e-01 4.75574970e-01 -2.61452109e-01 4.28440630e-01 -7.77784824e-01 -1.00722969e+00 -2.16631174e-01 -5.72659492e-01 4.49269116e-01 2.05201715e-01 5.80647588e-01 -1.32319421e-01 -1.79440096e-01 -7.31314182e-01 -7.76953995e-01 -2.24116564e-01 -5.93775511e-01 4.64732230e-01 2.02250645e-01 -5.07454991e-01 -7.52265632e-01 -2.27224499e-01 5.52927673e-01 9.75890279e-01 5.87993972e-02 7.52144158e-01 -1.21638644e+00 -5.23602128e-01 -1.16071200e+00 -5.02420962e-01 1.11459509e-01 1.20269675e-02 -1.95490152e-01 -6.34556592e-01 3.84236500e-02 -1.98545203e-01 -5.01881003e-01 6.55513704e-01 5.11732459e-01 1.39420438e+00 -1.11743140e+00 -2.88543314e-01 -1.05849482e-01 1.34017444e+00 2.73950752e-02 4.95380402e-01 6.03109777e-01 -1.46458954e-01 5.58701038e-01 5.24466455e-01 6.66833520e-01 1.15761414e-01 7.02212691e-01 3.42585713e-01 1.22254111e-01 3.33847523e-01 -7.90811002e-01 3.47000211e-01 4.60526794e-01 3.06361586e-01 1.73325777e-01 -4.94497389e-01 2.70683318e-01 -1.39520860e+00 -1.06396353e+00 -1.84335023e-01 2.09458828e+00 6.47146225e-01 7.17564762e-01 5.47574520e-01 -1.83035880e-01 7.20005453e-01 5.19868374e-01 -3.39078009e-01 -6.00766301e-01 1.71805531e-01 5.19970953e-02 1.00641310e+00 4.82263327e-01 -1.03532326e+00 8.53422165e-01 6.77276325e+00 9.44836974e-01 -1.13815451e+00 -3.74560058e-02 8.66535783e-01 -3.92722249e-01 -3.48833472e-01 -3.67313743e-01 -9.12160754e-01 8.31037343e-01 1.16486812e+00 -3.64082038e-01 4.15376574e-01 1.26186728e+00 4.79202509e-01 -5.69527932e-02 -5.59308648e-01 7.86256313e-01 3.76678705e-02 -1.81031489e+00 4.09976579e-02 3.86080801e-01 4.17170763e-01 -4.51560393e-02 7.23823786e-01 4.49290186e-01 5.87487519e-01 -7.18938053e-01 6.11964464e-01 1.92353815e-01 2.14493319e-01 -5.36187172e-01 4.19246286e-01 1.61235124e-01 -4.63967890e-01 -3.88341844e-01 -3.03038538e-01 -2.22261831e-01 2.60573268e-01 7.56871343e-01 -8.21695328e-01 -4.79469478e-01 5.59093654e-01 3.33219290e-01 -8.80471826e-01 1.09069395e+00 -7.71951687e-04 9.12057698e-01 -2.59983450e-01 -1.08410525e+00 9.71364975e-01 1.18905917e-01 3.65479410e-01 9.72516716e-01 2.41508871e-01 -3.97167593e-01 7.27984682e-03 7.95790553e-01 -2.42175743e-01 5.44882238e-01 -3.56598228e-01 -8.25522423e-01 3.58102262e-01 1.32086682e+00 -4.55756009e-01 -3.06923926e-01 -4.88486111e-01 5.90666175e-01 1.26138672e-01 6.99634552e-02 -1.01285815e+00 -5.23975015e-01 2.86335409e-01 9.09054637e-01 4.42237794e-01 -1.17702983e-01 -3.94174904e-02 -8.79558563e-01 -1.07445121e-02 -7.36301303e-01 2.53910780e-01 -7.75251448e-01 -1.67661512e+00 5.74741185e-01 -4.54854637e-01 -9.24151301e-01 -1.43425345e-01 -6.94649875e-01 -4.79415745e-01 5.10063350e-01 -1.43570590e+00 -8.63260865e-01 -9.73838009e-03 2.34621614e-01 5.35419345e-01 -9.83766615e-02 4.13633436e-01 2.85364330e-01 -1.54830202e-01 4.35901940e-01 1.22259036e-01 3.14974710e-02 5.62110722e-01 -1.05662799e+00 4.35236067e-01 -4.61504459e-02 2.14856192e-01 6.76685572e-01 1.04699302e+00 -6.89157605e-01 -1.33319855e+00 -7.30976760e-01 1.41469347e+00 -2.40630388e-01 1.79224610e+00 -4.17041957e-01 -5.36359429e-01 5.40182412e-01 -5.80506632e-03 -2.47467086e-01 1.00441623e+00 6.03170395e-01 -8.91702294e-01 -2.41173655e-01 -9.27427113e-01 7.04811931e-01 3.03902805e-01 -2.50671476e-01 -5.25350749e-01 9.15769100e-01 5.61717570e-01 2.73956239e-01 -6.64295673e-01 -4.07945335e-01 9.03219581e-01 -5.01252472e-01 9.79803085e-01 -9.79255676e-01 6.86831355e-01 4.45347607e-01 -2.86555201e-01 -1.03318691e+00 -6.50185704e-01 -5.64867854e-01 1.65203691e-01 8.16959977e-01 1.17559826e+00 -6.57764018e-01 9.18434143e-01 6.27608120e-01 3.80804539e-01 -6.43302321e-01 -6.46478653e-01 -8.08008373e-01 -1.40791297e-01 -2.88692176e-01 2.84277290e-01 6.42259121e-01 3.10124874e-01 6.55994475e-01 -6.21448636e-01 -3.93215179e-01 6.30070090e-01 8.30152407e-02 6.36474252e-01 -1.34102798e+00 -3.15552801e-01 -6.38526559e-01 -3.20623219e-01 -1.29091513e+00 -4.61316109e-01 -9.26184654e-01 -5.16080260e-01 -9.50383782e-01 1.74736738e-01 -1.29214436e-01 -2.83431292e-01 6.14314787e-02 2.79208899e-01 5.62866330e-01 -1.24998428e-02 3.55212241e-01 -7.60995090e-01 8.90523568e-02 1.13100863e+00 -2.80025572e-01 -1.09618567e-01 4.30075884e-01 -1.13207388e+00 7.21957088e-01 8.18092644e-01 -4.31472719e-01 3.42768617e-02 9.99465659e-02 8.96209538e-01 -1.69426024e-01 4.72533375e-01 -3.44461620e-01 2.23905537e-02 -1.89061195e-01 3.75766933e-01 -5.60330033e-01 4.05553341e-01 -6.51634753e-01 -2.45930389e-01 4.48279142e-01 -1.08465958e+00 1.27364546e-01 -1.45098522e-01 9.22350526e-01 -1.46838993e-01 -3.15231651e-01 7.35137224e-01 -9.96131897e-02 -3.75016987e-01 1.30799085e-01 -6.03484690e-01 -2.25886963e-02 8.63778472e-01 1.16018318e-01 -5.54449856e-01 -8.85427475e-01 -6.11131430e-01 -9.48017612e-02 -1.29720688e-01 7.21641362e-01 3.81399870e-01 -1.40711427e+00 -6.77442908e-01 1.44800663e-01 3.55299652e-01 -1.29712081e+00 -7.22004324e-02 5.06812751e-01 -4.41926003e-01 5.41834652e-01 1.07165836e-01 -1.80108234e-01 -9.66386437e-01 4.14324194e-01 -3.04786265e-01 -7.46685490e-02 -5.74653506e-01 8.20034206e-01 -1.45505220e-01 3.33922535e-01 2.65023261e-01 1.08532861e-01 -1.23408243e-01 2.90936679e-01 1.00758195e+00 2.36021921e-01 -9.13893208e-02 -3.45423669e-01 -2.08041310e-01 2.52650138e-02 -7.86766231e-01 -1.71158120e-01 1.31192386e+00 6.80888072e-02 2.57380068e-01 4.98308718e-01 1.64084172e+00 2.00342819e-01 -6.98297739e-01 -4.44716632e-01 4.63892847e-01 -9.83236134e-01 5.65225780e-01 -1.10028744e+00 -9.14423585e-01 3.52353096e-01 3.33418608e-01 9.60268199e-01 1.84833214e-01 4.16270018e-01 1.05211282e+00 3.42180312e-01 3.20633757e-03 -1.33925021e+00 4.86463457e-01 -8.12207684e-02 9.40319180e-01 -1.39192939e+00 1.74750403e-01 -2.69248635e-01 -8.02034736e-01 9.76170540e-01 4.74067628e-02 -4.27445114e-01 9.10590053e-01 -2.49697529e-02 2.08616033e-02 -5.28886199e-01 -8.52979481e-01 4.43424769e-02 4.45760161e-01 1.40941534e-02 5.62626004e-01 1.12430170e-01 -8.04166019e-01 4.50667560e-01 -3.20072323e-01 -1.56738162e-01 2.54353255e-01 6.77121043e-01 -1.17154527e+00 -5.36050439e-01 1.06052142e-02 1.13899553e+00 -1.12516463e+00 -1.43227413e-01 -4.23292100e-01 7.06580281e-01 -7.69096017e-01 9.53557372e-01 1.86250195e-01 -4.33713138e-01 -4.84665670e-02 4.10314426e-02 -2.05833241e-01 -1.18693136e-01 -4.28899944e-01 4.71796662e-01 4.77969885e-01 -5.14429390e-01 1.90440621e-02 -4.85229582e-01 -3.41332883e-01 -5.76853871e-01 -6.26360655e-01 3.55499566e-01 1.30166626e+00 4.71264839e-01 4.16757971e-01 1.16783753e-01 1.03679907e+00 -2.00230554e-01 -8.84294152e-01 -8.80870044e-01 -7.33199239e-01 3.93789470e-01 5.05751185e-02 -5.76798081e-01 -6.72312737e-01 -5.51057935e-01]
[7.797877311706543, 9.79059886932373]
bcdacc24-0cbf-4617-9b51-183831a7e6ea
native-language-identification-using-stacked
1703.06541
null
http://arxiv.org/abs/1703.06541v1
http://arxiv.org/pdf/1703.06541v1.pdf
Native Language Identification using Stacked Generalization
Ensemble methods using multiple classifiers have proven to be the most successful approach for the task of Native Language Identification (NLI), achieving the current state of the art. However, a systematic examination of ensemble methods for NLI has yet to be conducted. Additionally, deeper ensemble architectures such as classifier stacking have not been closely evaluated. We present a set of experiments using three ensemble-based models, testing each with multiple configurations and algorithms. This includes a rigorous application of meta-classification models for NLI, achieving state-of-the-art results on three datasets from different languages. We also present the first use of statistical significance testing for comparing NLI systems, showing that our results are significantly better than the previous state of the art. We make available a collection of test set predictions to facilitate future statistical tests.
['Mark Dras', 'Shervin Malmasi']
2017-03-19
null
null
null
null
['native-language-identification']
['natural-language-processing']
[ 3.27798575e-01 -4.99664038e-01 -2.70263076e-01 -6.10661268e-01 -9.35975730e-01 -8.57671976e-01 1.10685790e+00 3.29877809e-02 -4.80020106e-01 9.52240527e-01 1.88360572e-01 -7.17965782e-01 -2.24478796e-01 -2.14416265e-01 -3.73650402e-01 -4.59384978e-01 -2.16778100e-01 7.84811974e-01 -3.46259803e-01 -2.48366535e-01 3.33749264e-01 2.02601686e-01 -1.93838608e+00 6.03543758e-01 1.16651154e+00 6.83775485e-01 -4.57284123e-01 7.91055977e-01 -3.23570848e-01 5.22304237e-01 -9.27800894e-01 -6.39829218e-01 -1.47254780e-01 -2.12089300e-01 -7.25004137e-01 -7.38756776e-01 9.17875350e-01 1.40380070e-01 -4.20642160e-02 8.21991324e-01 8.63883853e-01 -5.43920090e-03 8.39903831e-01 -1.13217127e+00 -6.48181200e-01 8.97494137e-01 -1.29132986e-01 3.08204710e-01 7.65522897e-01 -6.56373873e-02 9.21175122e-01 -8.32193673e-01 4.91968632e-01 1.37480497e+00 9.48943555e-01 7.20744193e-01 -1.43068135e+00 -1.23362231e+00 8.48601535e-02 1.81906402e-01 -1.34846294e+00 -8.79193485e-01 4.38219309e-01 -3.69533092e-01 1.50413013e+00 4.00654674e-01 1.92082956e-01 1.47171545e+00 2.86625791e-02 1.02349555e+00 1.88857138e+00 -8.43738079e-01 -6.95092678e-02 5.23732781e-01 6.73611403e-01 3.26528311e-01 4.02203083e-01 4.26324040e-01 -6.27842784e-01 -3.09445024e-01 -7.82540590e-02 -6.53485239e-01 -8.25501326e-03 6.65017441e-02 -1.44571424e+00 8.81780326e-01 -2.46840194e-01 8.61090958e-01 9.59738791e-02 -3.00198406e-01 4.31453139e-01 4.20527756e-01 8.34841371e-01 5.72094142e-01 -8.28576922e-01 -2.56965727e-01 -1.19963837e+00 3.95977855e-01 1.24797976e+00 7.48471022e-01 2.38249704e-01 -1.53609803e-02 -1.12587437e-01 9.30920124e-01 2.06684113e-01 4.87326980e-01 5.12558818e-01 -5.47445953e-01 6.28657341e-01 6.35181844e-01 -1.63619429e-01 -5.43722272e-01 -2.77512342e-01 -3.81053925e-01 -8.51731062e-01 3.71105552e-01 6.80476069e-01 -1.81384563e-01 -5.90926111e-01 1.66637659e+00 -2.58591622e-01 2.86599725e-01 3.62325400e-01 2.27785975e-01 1.06118917e+00 3.61796767e-01 4.55482066e-01 -9.08286721e-02 1.13154173e+00 -6.04567289e-01 -6.69769049e-01 -1.40037984e-01 8.69678855e-01 -7.79787600e-01 5.07497191e-01 6.21475935e-01 -7.34028935e-01 -7.07466781e-01 -8.67353678e-01 3.23815405e-01 -7.41658568e-01 3.33639175e-01 9.39419329e-01 1.30512691e+00 -1.01236928e+00 4.89073396e-01 -3.72961998e-01 -4.82701778e-01 1.61187753e-01 6.84510589e-01 -8.85309517e-01 8.27285349e-02 -1.24226642e+00 1.12452793e+00 5.07702231e-01 -1.72484964e-01 -3.90745819e-01 -7.30174422e-01 -7.84919083e-01 -4.09976482e-01 -6.21308535e-02 -4.93568510e-01 1.06615222e+00 -8.99046004e-01 -1.36763966e+00 1.03771842e+00 -5.66750407e-01 -4.43515301e-01 3.23447376e-01 -1.03870250e-01 -8.78505707e-01 -6.82261705e-01 -1.01271875e-01 4.92124975e-01 4.25810933e-01 -1.27011168e+00 -7.78096616e-01 -6.37273490e-01 -2.54209936e-01 4.59194183e-02 -3.35615069e-01 5.85464299e-01 1.77931845e-01 -6.76694810e-01 -3.21838796e-01 -9.73924875e-01 1.15109466e-01 -1.00077260e+00 -3.72076184e-01 -6.49352908e-01 6.36484861e-01 -9.63072896e-01 1.76428032e+00 -1.80941999e+00 1.50218636e-01 -4.36782837e-04 -5.15985712e-02 4.73504215e-01 -1.85002878e-01 2.74175793e-01 -3.78003240e-01 6.75783575e-01 -2.08170101e-01 -6.57611549e-01 -9.17110890e-02 -8.42000172e-02 -2.29091689e-01 5.07342676e-03 3.27307195e-03 7.78101921e-01 -6.70390368e-01 -3.67448956e-01 2.32704699e-01 4.67757612e-01 -4.12899494e-01 1.45712689e-01 -1.98555812e-02 7.73060203e-01 3.75059992e-02 6.98327422e-01 3.83578211e-01 1.54739037e-01 2.87729949e-01 -6.30579367e-02 -4.10890430e-01 4.29394990e-01 -1.27764153e+00 1.46671045e+00 -4.58871603e-01 6.92358136e-01 -7.98685700e-02 -1.00327277e+00 9.50813651e-01 3.49800795e-01 1.86987668e-01 -2.61769503e-01 -3.60712297e-02 6.27350509e-01 5.78070223e-01 5.51086431e-03 3.02112132e-01 3.47824007e-01 -3.02021354e-01 4.51181918e-01 3.20396990e-01 1.34347081e-01 3.51318926e-01 -6.27746806e-02 7.98971891e-01 2.84118295e-01 5.92878938e-01 -4.39496666e-01 1.07428014e+00 1.04922624e-02 2.23260701e-01 1.25166309e+00 -1.99115202e-01 1.83745205e-01 -1.30928919e-01 -4.00363982e-01 -9.54833806e-01 -7.46241152e-01 -5.26430666e-01 1.30047846e+00 -6.11633837e-01 -3.29519838e-01 -8.74627292e-01 -7.17546046e-01 1.65071890e-01 1.07377982e+00 -4.14466262e-01 3.18433195e-01 -5.46486914e-01 -1.12837315e+00 8.60754073e-01 3.94570500e-01 6.08567059e-01 -1.13470256e+00 9.73618478e-02 1.97720081e-01 -2.12155372e-01 -9.54181790e-01 -1.03785865e-01 1.04368351e-01 -7.93538749e-01 -9.32649076e-01 -3.79551351e-01 -6.75712049e-01 9.01340246e-02 -1.20564051e-01 1.29275143e+00 1.21640250e-01 -2.09481359e-01 4.73588049e-01 -2.45122075e-01 -6.97816133e-01 -9.85738158e-01 6.25990331e-01 5.28785408e-01 -2.94029057e-01 8.43603194e-01 -3.05653334e-01 -1.17266484e-01 2.94394642e-01 -3.71316940e-01 -1.28573298e-01 2.72122890e-01 9.96878386e-01 1.33046150e-01 6.79991022e-03 5.42360663e-01 -8.13647747e-01 7.66448498e-01 -2.71311492e-01 -3.85522842e-01 6.26527667e-01 -7.22219706e-01 2.57501751e-01 2.88031727e-01 -6.40100300e-01 -1.14418817e+00 -1.91893190e-01 -3.16514730e-01 1.76864445e-01 -7.23540962e-01 4.19181734e-01 -1.44948915e-01 -4.25122976e-01 3.68243486e-01 1.49354771e-01 1.40523808e-02 -7.95411766e-01 1.06746502e-01 1.15681660e+00 4.08187389e-01 -6.93873703e-01 2.55136549e-01 -1.43730193e-01 -4.08349931e-01 -8.67513239e-01 -7.58433819e-01 -4.70060676e-01 -9.93036807e-01 -1.32910460e-01 6.26672864e-01 -8.93177986e-01 -7.40801096e-01 7.25066841e-01 -1.12332749e+00 -2.00468436e-01 4.69639003e-01 5.43662131e-01 -3.61577272e-01 2.90489018e-01 -2.25489274e-01 -1.21531796e+00 -5.33953667e-01 -1.33798039e+00 8.89616907e-01 3.55963595e-02 -7.75432110e-01 -1.14760375e+00 4.51681882e-01 7.05333710e-01 6.23303592e-01 -1.02347702e-01 9.17736888e-01 -1.48159134e+00 -3.85315418e-02 -1.88331008e-01 1.94134768e-02 1.97094545e-01 -1.24272220e-01 2.17105791e-01 -1.32484412e+00 -4.67133999e-01 -4.30238783e-01 -1.71337500e-01 9.69981492e-01 1.79340094e-01 1.13813114e+00 -1.16690174e-01 -6.82382882e-01 4.64697033e-01 1.14807236e+00 2.03748420e-01 4.92726386e-01 4.40498352e-01 5.99827170e-01 7.74471641e-01 2.57650733e-01 4.30725105e-02 4.93406534e-01 8.84710729e-01 -2.51875728e-01 4.27923530e-01 -2.21712649e-01 -1.27793938e-01 4.46989894e-01 8.69116783e-01 -2.37837911e-01 -3.78854662e-01 -1.36799896e+00 3.69997501e-01 -1.57044518e+00 -1.29906058e+00 -1.31136909e-01 2.52539134e+00 7.49129593e-01 -1.04214020e-01 2.05200940e-01 4.50673491e-01 5.65769792e-01 -1.17687449e-01 -3.00240546e-01 -5.92176795e-01 -3.13343108e-01 4.72945094e-01 2.25735262e-01 7.37267137e-01 -1.51732695e+00 1.06679070e+00 7.86842442e+00 8.77948105e-01 -9.01392281e-01 1.75743923e-01 7.18264222e-01 9.85027775e-02 -2.34719977e-01 1.10990323e-01 -1.58183861e+00 6.04623914e-01 1.36838591e+00 -4.59922403e-02 5.63868523e-01 5.43397725e-01 -4.92353767e-01 -1.64779887e-01 -1.12876940e+00 9.59406674e-01 3.09824437e-01 -1.00960505e+00 -1.26954243e-01 2.39500538e-01 8.83527040e-01 2.60979533e-01 -1.94876101e-02 7.22589612e-01 4.55805629e-01 -1.30891562e+00 4.34570879e-01 6.10469460e-01 9.38699424e-01 -7.36683190e-01 8.56811345e-01 5.45166612e-01 -8.93318534e-01 -2.79466152e-01 1.80501968e-01 2.19813678e-02 -2.67788082e-01 4.10484672e-01 -4.75331932e-01 6.97786629e-01 8.03959846e-01 5.26626348e-01 -8.91948998e-01 7.20408678e-01 1.13072939e-01 9.29126620e-01 -5.41610062e-01 -3.90423894e-01 -1.84158459e-01 1.34298846e-01 7.11282074e-01 1.50475204e+00 3.48034918e-01 -1.77079812e-01 1.34016037e-01 4.35188115e-01 2.57673860e-01 2.67998338e-01 -1.01764870e+00 -2.67409205e-01 5.92373908e-01 1.15534210e+00 -8.12316164e-02 -3.65706295e-01 -4.74853516e-01 9.04164672e-01 5.45931399e-01 1.43338859e-01 -2.94696003e-01 -2.22980633e-01 8.21135342e-01 -3.93181443e-01 -1.72841370e-01 -3.69361252e-01 -5.81501961e-01 -1.30077767e+00 -3.03520858e-01 -1.40668130e+00 6.77359879e-01 -3.15963447e-01 -1.60229993e+00 7.23333657e-01 3.71115625e-01 -8.74750733e-01 -7.71646440e-01 -1.22244465e+00 -5.22205055e-01 1.09558105e+00 -1.14659214e+00 -1.17490220e+00 -1.38772830e-01 2.16791660e-01 2.93841630e-01 -1.05810845e+00 1.50517642e+00 1.84709296e-01 -7.70806313e-01 1.02488708e+00 3.11469436e-01 1.74278677e-01 1.01493013e+00 -1.16480088e+00 4.47000980e-01 5.25375366e-01 4.18220043e-01 8.56771767e-01 5.84326148e-01 -6.63095176e-01 -1.27278483e+00 -5.38098395e-01 1.37410069e+00 -1.17017293e+00 6.20331526e-01 -4.00477439e-01 -7.46110380e-01 7.22971261e-01 3.75829130e-01 -5.76843202e-01 1.14472818e+00 7.31101036e-01 -5.07223010e-01 -6.64913729e-02 -1.20930922e+00 3.77019227e-01 1.16965747e+00 -4.05877799e-01 -6.13572001e-01 8.93075094e-02 2.02876180e-01 -1.83042079e-01 -1.12143970e+00 7.94062555e-01 8.78974915e-01 -1.03659987e+00 1.12674546e+00 -6.28974915e-01 1.48164704e-01 2.46511437e-02 -2.70228446e-01 -1.49014008e+00 -2.52910942e-01 -3.41171384e-01 -2.48553865e-02 1.69981813e+00 8.40139985e-01 -1.07775140e+00 4.15602326e-01 6.49823725e-01 1.75988704e-01 -5.94796598e-01 -1.04413497e+00 -8.53865623e-01 4.26690072e-01 -9.36115742e-01 7.46994734e-01 9.79708016e-01 2.75207534e-02 3.88090521e-01 -1.04440667e-01 -1.11487716e-01 7.89178908e-01 -1.17095619e-01 9.26783681e-01 -1.43163502e+00 -6.58910424e-02 -8.71959567e-01 -4.57658172e-01 -1.98897898e-01 8.75458717e-01 -1.39028180e+00 -4.08500314e-01 -1.09305811e+00 4.67051476e-01 -4.88873333e-01 -2.77537704e-01 5.08010328e-01 -3.41194481e-01 4.00123000e-01 1.92275167e-01 3.13875377e-01 -3.20498019e-01 5.56563884e-02 5.26087999e-01 -4.23573762e-01 -1.09823458e-01 -1.01315724e-02 -7.60969520e-01 6.11380637e-01 9.80578005e-01 -2.00760931e-01 -1.41940460e-01 -3.01288694e-01 -5.27408421e-02 -3.02277178e-01 8.14474523e-02 -1.30688083e+00 2.60503083e-01 3.71593118e-01 6.80942714e-01 -7.01414704e-01 1.17878228e-01 -4.17648226e-01 2.10410833e-01 2.03879893e-01 -4.62270588e-01 3.72086793e-01 6.28425062e-01 9.31361392e-02 -1.90698326e-01 -2.85395503e-01 5.08269906e-01 1.87864266e-02 -8.94183278e-01 1.55240089e-01 -3.06137472e-01 5.35214804e-02 7.26591349e-01 -1.65653974e-01 -2.93070860e-02 -1.58559889e-01 -6.32968605e-01 -1.31929033e-02 2.20306441e-01 7.58547246e-01 2.58783609e-01 -1.10676122e+00 -1.25988638e+00 5.02653658e-01 3.02490294e-01 -8.50340128e-01 1.02822501e-02 6.02890790e-01 -1.98435020e-02 7.90688872e-01 9.75995213e-02 -5.83862543e-01 -1.72094631e+00 3.29898775e-01 3.49234134e-01 -7.41103590e-01 -1.15560040e-01 6.75407767e-01 -2.56594956e-01 -1.38774025e+00 2.51588255e-01 4.34258759e-01 -5.44142604e-01 6.18752912e-02 8.31020892e-01 4.49260384e-01 3.82323600e-02 -8.43635023e-01 -6.38954818e-01 5.27065814e-01 -2.93409199e-01 -3.48672420e-01 9.48171675e-01 1.61690712e-01 -4.49751496e-01 6.46255612e-01 9.95605111e-01 -4.58463058e-02 -8.66527781e-02 -1.44873289e-02 3.07033122e-01 -1.06571585e-01 -1.66817769e-01 -1.53687012e+00 -2.90416449e-01 7.94128060e-01 8.35913241e-01 -1.28151014e-01 7.57785261e-01 -3.57212663e-01 2.11712062e-01 3.50797713e-01 7.45022655e-01 -9.39424574e-01 -5.58294892e-01 8.83726060e-01 8.21480513e-01 -1.70153534e+00 2.29629083e-03 -1.81695774e-01 -3.82292807e-01 1.15592790e+00 4.87106264e-01 3.03128481e-01 7.46112168e-01 4.86396134e-01 9.61136147e-02 3.99038553e-01 -8.20738137e-01 -1.64969210e-02 8.00420225e-01 6.90399706e-01 1.28271163e+00 4.49983656e-01 -4.17349190e-01 7.18639076e-01 -4.95585322e-01 -7.66698867e-02 -2.61385888e-01 2.95373976e-01 9.42217559e-02 -1.58058822e+00 -6.65522635e-01 7.17607975e-01 -6.25694454e-01 -3.67128909e-01 -7.05790162e-01 9.14109945e-01 1.34608030e-01 1.28300965e+00 1.26370117e-02 -6.25031829e-01 1.91713676e-01 8.56302559e-01 7.34438658e-01 -3.51206630e-01 -1.09567964e+00 -5.51965833e-01 5.92313468e-01 -1.89694807e-01 -4.62348580e-01 -1.07170105e+00 -4.99042690e-01 -6.71131611e-01 -3.48965317e-01 3.31234820e-02 9.22753394e-01 9.69931066e-01 4.08284217e-01 1.34301424e-01 2.89438188e-01 -9.11160707e-01 -6.16214871e-01 -1.31386113e+00 -1.20554574e-01 4.18057233e-01 -1.13369087e-02 -5.55752218e-01 -6.11896396e-01 -4.70973879e-01]
[10.394888877868652, 10.555804252624512]
c4b845b9-0d0b-48be-a4ec-33f4b18817bc
self-supervised-occupancy-grid-learning-from
1904.00415
null
https://arxiv.org/abs/1904.00415v2
https://arxiv.org/pdf/1904.00415v2.pdf
Road Scene Understanding by Occupancy Grid Learning from Sparse Radar Clusters using Semantic Segmentation
Occupancy grid mapping is an important component in road scene understanding for autonomous driving. It encapsulates information of the drivable area, road obstacles and enables safe autonomous driving. Radars are an emerging sensor in autonomous vehicle vision, becoming more widely used due to their long range sensing, low cost, and robustness to severe weather conditions. Despite recent advances in deep learning technology, occupancy grid mapping from radar data is still mostly done using classical filtering approaches.In this work, we propose learning the inverse sensor model used for occupancy grid mapping from clustered radar data. This is done in a data driven approach that leverages computer vision techniques. This task is very challenging due to data sparsity and noise characteristics of the radar sensor. The problem is formulated as a semantic segmentation task and we show how it can be learned using lidar data for generating ground truth. We show both qualitatively and quantitatively that our learned occupancy net outperforms classic methods by a large margin using the recently released NuScenes real-world driving data.
['Gilad Cohen', 'Shaul Oron', 'Liat Sless', 'Bat El Shlomo']
2019-03-31
null
null
null
null
['road-scene-understanding']
['computer-vision']
[ 5.42667210e-01 2.15063188e-02 -1.59997240e-01 -6.96778119e-01 -8.84549379e-01 -5.07411182e-01 8.26879621e-01 -2.83747986e-02 -6.61855280e-01 9.18980896e-01 4.39137258e-02 -3.58854860e-01 -2.24028006e-01 -1.04655659e+00 -8.24588001e-01 -6.80072427e-01 -1.07525907e-01 9.27702844e-01 3.42946619e-01 -3.01688850e-01 1.52448758e-01 5.85191786e-01 -2.08311582e+00 -3.54025036e-01 9.95673239e-01 8.88577640e-01 3.82556081e-01 5.44488788e-01 -1.55702427e-01 4.46036577e-01 7.26013482e-02 2.47421876e-01 5.13652742e-01 6.98730275e-02 -1.38157234e-01 4.08292040e-02 8.87181580e-01 -3.29007626e-01 -4.11382854e-01 1.11664212e+00 2.28334814e-01 3.03483427e-01 7.13518441e-01 -1.23428106e+00 1.67039439e-01 3.19175720e-01 -5.81724465e-01 8.06966126e-02 -2.95308441e-01 2.07420234e-02 8.44136775e-01 -6.75972939e-01 3.40131372e-01 1.08682644e+00 6.92458093e-01 2.03994900e-01 -1.25606096e+00 -8.33125949e-01 8.66009891e-02 2.99369127e-01 -1.60657048e+00 -4.51354623e-01 5.73134542e-01 -7.75106072e-01 5.98712265e-01 2.25320365e-02 6.84931457e-01 7.31330216e-01 3.24122816e-01 5.55425346e-01 1.57201195e+00 6.65998608e-02 6.61874413e-01 -6.33747801e-02 2.35877454e-01 5.39360821e-01 5.22911966e-01 6.38266921e-01 -3.66641015e-01 3.06151807e-01 4.37012494e-01 1.28882736e-01 -3.25785801e-02 -8.28444898e-01 -8.01074088e-01 1.22859931e+00 6.49410903e-01 -2.05022156e-01 -4.21065211e-01 4.75631267e-01 7.20218197e-02 4.38691396e-03 3.97069871e-01 6.90263733e-02 -5.96399419e-02 8.46499056e-02 -1.29820693e+00 6.72372162e-01 6.62432849e-01 7.23783433e-01 1.39755082e+00 2.43491903e-01 3.20514441e-01 5.11360109e-01 4.94626760e-01 1.22115076e+00 -1.06949657e-01 -1.09381163e+00 8.08836594e-02 2.26180226e-01 3.36393088e-01 -8.57489944e-01 -5.43079793e-01 -5.02330124e-01 -9.34895873e-01 7.34031320e-01 1.56488985e-01 -1.24333531e-01 -1.38316035e+00 1.58935940e+00 3.33401471e-01 5.76687276e-01 7.92214647e-02 9.39552605e-01 5.75540900e-01 5.21500885e-01 5.60442032e-03 1.33801758e-01 1.14642239e+00 -4.30233687e-01 -7.77771592e-01 -8.30304801e-01 1.98246002e-01 -2.04581454e-01 3.87636125e-01 3.11832011e-01 -2.18052506e-01 -6.10089123e-01 -1.44686079e+00 7.15323389e-02 -7.63951004e-01 -3.30527455e-01 6.05758309e-01 6.72267079e-01 -9.74344313e-01 1.65510252e-01 -7.46911287e-01 -2.84515351e-01 6.79358840e-01 2.08188802e-01 -1.75660372e-01 -4.21995819e-01 -1.31206787e+00 1.06288826e+00 3.91023040e-01 1.39634430e-01 -9.81225252e-01 -7.27671742e-01 -1.33274186e+00 -2.66506523e-01 5.37823796e-01 -4.32925969e-01 8.85253966e-01 -3.58601391e-01 -9.62093055e-01 6.79832757e-01 -1.66863784e-01 -1.18555295e+00 4.79655743e-01 -2.94334412e-01 -2.32550040e-01 -2.89280415e-01 4.20280874e-01 1.13328528e+00 9.58541274e-01 -1.38149822e+00 -1.08682609e+00 -5.57716131e-01 -2.16841310e-01 1.33509621e-01 6.41455710e-01 -1.02493846e+00 -1.68311492e-01 -6.56204745e-02 1.99581936e-01 -1.03466403e+00 -5.28115451e-01 -2.69266576e-01 -6.60165399e-02 8.85692704e-03 9.35189664e-01 -4.10620153e-01 5.40463567e-01 -1.85383368e+00 -2.84181148e-01 4.99893099e-01 3.76421630e-01 1.83981046e-01 1.27682999e-01 1.66504487e-01 5.45969903e-01 -2.31249899e-01 -7.06218481e-01 -1.01046771e-01 3.20202172e-01 5.19439340e-01 -7.09380805e-01 5.87165773e-01 1.92823201e-01 1.08328283e+00 -7.24544644e-01 -1.53938830e-01 9.17295933e-01 8.02493453e-01 -2.39489108e-01 2.48764660e-02 -4.64008301e-01 5.38259089e-01 -2.82571077e-01 3.72349113e-01 1.00625491e+00 4.23102826e-01 -2.04235867e-01 5.31005226e-02 -5.20457506e-01 7.07656518e-02 -1.38971972e+00 1.55879498e+00 -3.69454414e-01 1.03246999e+00 2.96249479e-01 -1.06853783e+00 1.18438506e+00 -2.95116276e-01 3.19955528e-01 -1.18679011e+00 -8.40088539e-03 2.55724549e-01 -2.38593847e-01 -1.78973794e-01 6.37784541e-01 -3.25622112e-01 -4.44577843e-01 1.41978025e-01 -4.36368942e-01 -6.37234151e-01 -8.15424509e-03 -3.64675820e-02 1.02094316e+00 7.89575800e-02 1.56699225e-01 -4.49294806e-01 1.44458026e-01 5.18129528e-01 5.84734261e-01 1.13373196e+00 -1.38725907e-01 4.47073549e-01 -1.99933767e-01 -7.45163560e-01 -1.03043795e+00 -1.21622133e+00 -4.62238759e-01 5.42771399e-01 3.67258430e-01 1.15751766e-01 -6.58307493e-01 -2.04336584e-01 4.39030796e-01 5.77255189e-01 -6.89846456e-01 1.14713445e-01 -4.81461525e-01 -5.94946563e-01 2.92887330e-01 5.13972759e-01 6.88759208e-01 -8.23180914e-01 -9.96356905e-01 3.84255499e-01 -9.41762608e-03 -1.36151290e+00 1.90373212e-01 3.19551408e-01 -4.55934584e-01 -1.13832426e+00 -2.12718681e-01 -2.79579580e-01 1.86990857e-01 6.18750095e-01 1.12382448e+00 -2.40034357e-01 -7.40688562e-01 2.99663633e-01 1.39513016e-02 -8.16307187e-01 5.66920564e-02 1.04341812e-01 9.67375785e-02 1.68932080e-01 7.96715796e-01 -6.31195843e-01 -4.08143699e-01 7.64795393e-02 -6.95281684e-01 2.56479204e-01 7.45969713e-01 4.72841948e-01 9.29020703e-01 3.05379659e-01 5.34414530e-01 -8.91460657e-01 1.67376384e-01 -5.26601791e-01 -1.20144880e+00 -4.73816544e-01 -4.85666245e-01 7.70969093e-02 3.27471830e-02 4.17053074e-01 -6.62382960e-01 5.40939927e-01 -1.49428710e-01 -2.35516652e-01 -6.90435052e-01 3.41909617e-01 -2.19826818e-01 -1.47813603e-01 4.69426423e-01 1.75134450e-01 2.16257125e-02 -1.51813835e-01 4.84742582e-01 5.71433008e-01 7.65892446e-01 -1.25876129e-01 1.26602077e+00 1.08399332e+00 1.93845838e-01 -1.23661196e+00 -1.14214242e+00 -6.67081594e-01 -7.63099074e-01 -3.51419568e-01 1.00933182e+00 -1.25406194e+00 -5.12094557e-01 1.35814950e-01 -6.99123681e-01 -4.24302936e-01 -2.97894716e-01 3.50580812e-01 -7.80914545e-01 -8.89010429e-02 3.68764490e-01 -1.03740633e+00 -1.10061243e-01 -9.31788743e-01 1.23842239e+00 2.37479851e-01 4.41954918e-02 -8.37555110e-01 3.97318095e-01 4.75436836e-01 5.93875051e-01 5.31245351e-01 4.65691268e-01 -1.23308271e-01 -1.08558738e+00 -1.47727475e-01 -5.00214756e-01 1.21253259e-01 -8.86492506e-02 -5.04534483e-01 -1.11889434e+00 -1.65636897e-01 -1.05564751e-01 -1.62764087e-01 1.55521953e+00 7.37074018e-01 8.63761067e-01 6.35724366e-02 -4.92052436e-01 6.24296665e-01 1.78664744e+00 -2.89643705e-02 6.22507274e-01 1.29690334e-01 7.88929939e-01 9.97305095e-01 8.94141138e-01 1.04833744e-01 6.41632199e-01 5.50921917e-01 9.10348535e-01 -2.00993448e-01 -6.33976012e-02 -2.50096500e-01 -5.20278402e-02 -5.07788882e-02 3.89130682e-01 1.11688368e-01 -1.23918366e+00 9.02482688e-01 -2.22494125e+00 -1.08027744e+00 -1.87200725e-01 2.25920701e+00 1.57524049e-01 3.57328832e-01 -1.73878655e-01 5.35012446e-02 4.53760505e-01 2.57855356e-01 -7.05173969e-01 -7.54109696e-02 -2.05038562e-01 2.90360034e-01 1.17557931e+00 1.04487574e+00 -1.45984042e+00 1.12571526e+00 5.67062235e+00 6.97839022e-01 -9.82550681e-01 6.58918023e-02 2.13317022e-01 1.07727565e-01 -3.99612665e-01 -2.03321412e-01 -1.32225132e+00 3.21262300e-01 9.20948684e-01 1.84721515e-01 2.51729548e-01 5.55253506e-01 4.69878614e-01 -5.74619293e-01 -6.29209936e-01 9.45309758e-01 6.58787191e-02 -1.69248736e+00 -2.68695623e-01 4.84394163e-01 7.75404453e-01 6.61050439e-01 3.71051878e-02 3.91437680e-01 8.10224891e-01 -1.45346451e+00 4.43235666e-01 5.91606498e-01 5.14577866e-01 -9.94852901e-01 8.36626232e-01 5.65644443e-01 -1.26312637e+00 7.09840506e-02 -5.19895792e-01 -2.94857442e-01 3.47176254e-01 1.02085507e+00 -8.53566229e-01 2.10516050e-01 6.47380173e-01 8.23170483e-01 -2.01936409e-01 1.02735794e+00 -1.84127390e-01 5.37597120e-01 -5.89059830e-01 3.74811947e-01 6.71622515e-01 -4.04932261e-01 5.58735013e-01 1.04805708e+00 1.33814171e-01 -1.69177756e-01 6.51427150e-01 1.00716472e+00 2.18109325e-01 -3.68632406e-01 -1.31610262e+00 3.35712940e-01 5.39984822e-01 1.45207775e+00 -6.26128912e-01 -1.51503040e-02 -1.01780742e-01 1.84370399e-01 8.03953260e-02 1.81763500e-01 -7.78781235e-01 -3.16360205e-01 1.06588221e+00 4.64958876e-01 6.42566323e-01 -5.65369964e-01 -3.63168687e-01 -5.04524827e-01 -2.37260297e-01 -2.59864718e-01 -2.35457495e-01 -4.97643620e-01 -9.97858226e-01 3.03631097e-01 1.36943892e-01 -1.06989396e+00 -2.75259882e-01 -5.10579169e-01 -4.03078824e-01 8.54155779e-01 -2.35063386e+00 -1.12649608e+00 -7.57348239e-01 2.75804460e-01 5.17584264e-01 -1.08352818e-01 5.88773012e-01 5.70297204e-02 -4.92374562e-02 -2.94165134e-01 2.62766778e-01 3.05948984e-02 3.46378088e-01 -1.34174383e+00 6.80612564e-01 9.00584519e-01 2.05218866e-01 -2.61557624e-02 1.01046062e+00 -7.69905031e-01 -1.28302324e+00 -1.70339370e+00 6.94898427e-01 -3.37370068e-01 4.74429846e-01 -7.99977064e-01 -8.41862977e-01 4.20934618e-01 -6.78031594e-02 4.63497750e-02 4.20140266e-01 -1.52718484e-01 -2.30134264e-01 -3.45143497e-01 -1.08279085e+00 3.68637174e-01 8.61305296e-01 -4.52303559e-01 -5.09144843e-01 9.62988809e-02 5.29597521e-01 -2.28225991e-01 -2.81584084e-01 7.51419902e-01 5.09402454e-01 -1.02402067e+00 7.25925028e-01 5.40212989e-02 -2.95469747e-03 -6.59977138e-01 -4.53908801e-01 -1.24655962e+00 -2.64747441e-01 -2.93694139e-01 1.58807278e-01 7.20439434e-01 2.34227851e-01 -6.04698837e-01 1.18232131e+00 1.39162675e-01 -1.72357038e-01 -1.75365090e-01 -1.23496974e+00 -6.18929207e-01 -8.43262374e-02 -7.55064011e-01 3.49032313e-01 5.82715929e-01 -8.27102184e-01 4.29278642e-01 -3.31255913e-01 4.89722699e-01 1.39733350e+00 2.29139253e-01 9.17270601e-01 -1.81102931e+00 3.56727093e-01 -9.44629908e-02 -7.06457853e-01 -1.01458359e+00 5.07865250e-01 -7.13255763e-01 6.86168134e-01 -1.76284099e+00 -1.82296872e-01 -7.78720260e-01 1.17681660e-01 2.68849999e-01 1.53648630e-01 8.08482885e-01 -1.48483709e-01 -1.14703896e-02 -5.42326033e-01 6.43866241e-01 5.15226424e-01 -3.65879387e-01 1.74673628e-02 -9.42134634e-02 -4.57263887e-01 7.21240044e-01 1.02154315e+00 -4.76546258e-01 -3.88091743e-01 -3.29041481e-01 2.37953842e-01 -2.86982447e-01 5.97399175e-01 -1.47249973e+00 3.55781734e-01 -1.38937145e-01 2.15175956e-01 -1.25483370e+00 6.45712912e-01 -9.93347406e-01 2.00225681e-01 4.33931679e-01 1.19709380e-01 -4.04584497e-01 3.60017449e-01 9.43898559e-01 -1.90595075e-01 8.55642408e-02 7.91667402e-01 -7.96615630e-02 -1.26637793e+00 5.37396908e-01 -6.79590285e-01 -8.84065311e-03 1.08243585e+00 -5.67153871e-01 -1.17245063e-01 -4.58902866e-01 -5.65087199e-01 7.10302889e-01 4.32735205e-01 5.47099829e-01 6.05522454e-01 -1.17537749e+00 -8.99646640e-01 6.44118130e-01 2.42917731e-01 3.42574984e-01 3.90099138e-01 6.04268193e-01 -2.88629889e-01 7.14441180e-01 -1.94161266e-01 -1.00687003e+00 -8.95478189e-01 5.33032641e-02 3.49793136e-01 5.69705255e-02 -8.35802674e-01 3.97289425e-01 3.18679184e-01 -6.71168745e-01 1.69866979e-01 -8.24968368e-02 -2.88675457e-01 1.46580368e-01 5.77012777e-01 1.18993461e-01 1.76461294e-01 -8.54098916e-01 -2.18410134e-01 7.72529483e-01 4.58832756e-02 -3.00938070e-01 1.28029692e+00 -4.42076653e-01 2.25097835e-01 6.20499134e-01 7.02432692e-01 -2.60552257e-01 -1.68980610e+00 -3.88297766e-01 -8.55559949e-03 -3.94458920e-01 6.85313165e-01 -4.72205073e-01 -8.65622342e-01 1.11421490e+00 9.05415237e-01 1.01205684e-01 4.45701033e-01 -1.37772560e-01 6.99555218e-01 5.52454948e-01 6.71010375e-01 -1.28290045e+00 -4.47900981e-01 8.93312216e-01 4.63397235e-01 -1.72647727e+00 -7.39466250e-02 -2.73210764e-01 -4.06396478e-01 5.55918813e-01 3.56845886e-01 -5.01679063e-01 8.51059973e-01 5.73880613e-01 2.07076490e-01 -3.52077633e-01 -4.56957281e-01 -8.88440251e-01 7.83339143e-02 9.85632837e-01 -2.24841818e-01 4.28784847e-01 3.10187548e-01 1.53187722e-01 -4.80460286e-01 -2.04168215e-01 3.97105962e-01 7.96872795e-01 -1.32051313e+00 -7.92247474e-01 -4.20108289e-01 7.55648613e-01 1.21930473e-01 6.46738429e-03 -7.08226413e-02 7.98869610e-01 3.69859695e-01 8.67462456e-01 5.01702428e-01 -1.88869998e-01 1.03542306e-01 -9.01171789e-02 3.03070903e-01 -5.82298756e-01 1.56303421e-01 -1.52315319e-01 -9.62727591e-02 -5.76747894e-01 -4.17062074e-01 -7.31275499e-01 -1.28413951e+00 -2.63666838e-01 4.41182181e-02 1.25395253e-01 1.20320106e+00 1.25585973e+00 2.82668382e-01 5.69038749e-01 4.36460674e-01 -1.19986081e+00 -1.60940349e-01 -5.73713899e-01 -7.56137550e-01 -7.42210299e-02 6.27568781e-01 -1.08877063e+00 -3.18048894e-01 -2.98236221e-01]
[8.106411933898926, -2.05146861076355]
b20bd91a-aea6-46dc-b1bb-63a4b2afb7d0
graph-fairing-convolutional-networks-for
2010.10274
null
https://arxiv.org/abs/2010.10274v1
https://arxiv.org/pdf/2010.10274v1.pdf
Graph Fairing Convolutional Networks for Anomaly Detection
Graph convolution is a fundamental building block for many deep neural networks on graph-structured data. In this paper, we introduce a simple, yet very effective graph convolutional network with skip connections for semi-supervised anomaly detection. The proposed multi-layer network architecture is theoretically motivated by the concept of implicit fairing in geometry processing, and comprises a graph convolution module for aggregating information from immediate node neighbors and a skip connection module for combining layer-wise neighborhood representations. In addition to capturing information from distant graph nodes through skip connections between the network's layers, our approach exploits both the graph structure and node features for learning discriminative node representations. The effectiveness of our model is demonstrated through extensive experiments on five benchmark datasets, achieving better or comparable anomaly detection results against strong baseline methods.
['A. Ben Hamza', 'Mahsa Mesgaran']
2020-10-20
null
null
null
null
['supervised-anomaly-detection', 'semi-supervised-anomaly-detection']
['computer-vision', 'computer-vision']
[-2.76908487e-01 4.79686797e-01 2.73858034e-03 -5.36183178e-01 1.99521050e-01 -5.40993027e-02 4.63760525e-01 8.20557475e-01 -2.94609994e-01 1.41824812e-01 9.46382955e-02 -4.07322407e-01 1.10304520e-01 -1.13423598e+00 -6.77307785e-01 -4.42408115e-01 -7.54893720e-01 2.68930703e-01 3.41096044e-01 -2.79922694e-01 5.18607832e-02 1.04969692e+00 -9.59067166e-01 6.20282069e-02 8.01006079e-01 1.27606595e+00 -5.01503706e-01 6.34848773e-01 -4.46741521e-01 1.22257578e+00 -1.71163246e-01 -6.66938126e-01 3.12452883e-01 -7.00306660e-03 -7.94411957e-01 -8.30796361e-02 7.46926963e-01 -5.39206743e-01 -1.06360030e+00 1.14076638e+00 3.73128772e-01 2.71814764e-01 5.98556280e-01 -1.24683642e+00 -7.76032746e-01 6.55613780e-01 -5.96015573e-01 5.12694240e-01 1.30803466e-01 6.96570426e-02 1.43085182e+00 -8.30294490e-01 1.78721711e-01 1.16884959e+00 9.19248700e-01 1.83437228e-01 -1.07757473e+00 -3.23185742e-01 4.98057753e-01 3.14618796e-01 -1.55242050e+00 -7.02506304e-02 1.23654556e+00 -2.92358339e-01 1.05431616e+00 -8.99749324e-02 5.64125478e-01 8.27251971e-01 3.34836245e-01 8.24270129e-01 -1.77676659e-02 3.24463062e-02 3.81296873e-02 -5.24156988e-01 2.97963649e-01 1.20042431e+00 4.17224228e-01 -1.25616848e-01 -6.33007511e-02 -2.89953023e-01 8.80306304e-01 2.33711421e-01 1.13281153e-01 -6.80670977e-01 -7.59780049e-01 9.67944562e-01 1.34165192e+00 1.84640110e-01 -4.40105408e-01 5.83669126e-01 6.22390151e-01 4.07923192e-01 6.61306500e-01 2.94308424e-01 -7.97566846e-02 5.42346537e-01 -7.70073652e-01 6.92226051e-04 6.58983469e-01 8.80868852e-01 9.02922571e-01 5.41053832e-01 -3.62614721e-01 6.12727284e-01 4.49630141e-01 7.69590363e-02 6.14001937e-02 -3.80846471e-01 3.05863708e-01 1.11577308e+00 -5.79741478e-01 -1.52811372e+00 -7.33719945e-01 -7.60372579e-01 -1.31294000e+00 1.96015537e-01 2.74540126e-01 6.27509356e-02 -1.16992092e+00 1.52076256e+00 1.99862897e-01 4.88625318e-01 -3.90627325e-01 7.02108860e-01 1.19803107e+00 2.38758802e-01 1.02770135e-01 4.19687003e-01 6.59215510e-01 -1.00979853e+00 -4.25508499e-01 -1.70313776e-01 9.89279568e-01 -3.43875021e-01 7.94901907e-01 -4.31200266e-02 -9.64352727e-01 -4.81692016e-01 -1.02619195e+00 -3.15805256e-01 -6.52206361e-01 -1.45277068e-01 9.81129289e-01 3.07889670e-01 -1.38922024e+00 8.85229409e-01 -8.92816663e-01 -2.88670242e-01 9.29071665e-01 5.11135459e-01 -4.73403424e-01 -1.20193407e-01 -9.26659346e-01 2.94830918e-01 4.77014661e-01 3.25252175e-01 -7.51928687e-01 -4.81275827e-01 -1.51881015e+00 4.23487455e-01 2.61181593e-01 -5.87894440e-01 6.91754758e-01 -8.48165035e-01 -1.05570555e+00 7.28882551e-01 1.34804726e-01 -8.62550974e-01 2.48184666e-01 -2.35945627e-01 -5.49100101e-01 2.73800105e-01 -3.62838618e-02 3.40771914e-01 8.36194336e-01 -7.57339776e-01 -1.36414021e-01 -4.86280322e-01 1.73084140e-01 -7.63337761e-02 -3.57549936e-01 -2.54700243e-01 -4.86487210e-01 -9.98631239e-01 3.70807499e-01 -5.49055457e-01 -6.72556341e-01 2.38847002e-01 -7.88563251e-01 -3.36043268e-01 9.38715577e-01 -5.53738296e-01 1.28554368e+00 -2.07278275e+00 -1.16229631e-01 8.11173081e-01 1.00171494e+00 3.36065561e-01 -4.27947015e-01 4.59503919e-01 -3.75128359e-01 -2.78662704e-02 -4.43109125e-01 -4.93009418e-01 -1.21473044e-01 2.78600097e-01 -2.55206048e-01 6.87213361e-01 6.88138008e-01 1.34783840e+00 -1.04410994e+00 -3.36513728e-01 4.28837836e-01 3.68816644e-01 -6.99502051e-01 1.66078463e-01 -1.37823224e-01 1.49839580e-01 -5.36860645e-01 8.28386068e-01 9.13083196e-01 -4.82252389e-01 -9.74412784e-02 -5.91699742e-02 5.21579742e-01 3.11874509e-01 -1.06395304e+00 1.89107001e+00 -6.20372687e-03 4.21394408e-01 1.87493101e-01 -1.45503855e+00 1.06928742e+00 -1.01690166e-01 4.42581266e-01 -7.11539149e-01 3.25345844e-01 -9.71840881e-03 2.75643349e-01 -1.23089897e-02 3.64861876e-01 4.08461988e-01 -1.23899691e-01 4.06341374e-01 4.04579520e-01 2.81837761e-01 1.99242905e-01 8.10673952e-01 1.66534066e+00 -3.08435172e-01 1.90759823e-01 -4.97624546e-01 8.02198052e-01 -4.35999632e-01 4.42790270e-01 7.75228977e-01 -2.78821617e-01 5.83739817e-01 7.59847224e-01 -1.09662282e+00 -7.37277985e-01 -1.35888398e+00 1.38527304e-01 9.16650891e-01 1.38765767e-01 -8.45668793e-01 -4.68167603e-01 -1.15324020e+00 2.51668334e-01 4.27204698e-01 -8.56315076e-01 -6.17073476e-01 -7.42457271e-01 -3.88433099e-01 6.77228093e-01 8.77916813e-01 6.40627563e-01 -1.00137389e+00 2.06788778e-01 2.34438688e-01 3.86658907e-01 -1.17270124e+00 -4.66761380e-01 -3.57213430e-02 -8.78816485e-01 -1.27169883e+00 -2.46222615e-01 -7.59627044e-01 9.49024618e-01 1.86023101e-01 1.45251441e+00 8.75662148e-01 -4.68270689e-01 3.71912062e-01 -2.12910801e-01 -2.05210328e-01 -1.11291446e-01 3.47286642e-01 -7.30779096e-02 2.98898816e-01 3.01237106e-01 -9.80177045e-01 -6.23717248e-01 -4.23887633e-02 -8.03858578e-01 -3.43866557e-01 6.06550217e-01 7.83095837e-01 5.36967814e-01 -1.11450762e-01 1.86830699e-01 -1.10339630e+00 4.88656998e-01 -5.64615905e-01 -5.76577663e-01 -9.05407220e-02 -3.34603131e-01 3.16570342e-01 1.00750506e+00 2.08854482e-01 -4.91281420e-01 -5.17183132e-02 -4.27116007e-01 -6.88750088e-01 -1.68545753e-01 5.03831327e-01 -3.04990619e-01 -5.20250082e-01 4.40567702e-01 -7.80672356e-02 4.20920774e-02 -4.63804275e-01 4.21356112e-01 -2.13567629e-01 6.61001503e-01 -5.23093164e-01 1.19577801e+00 5.31952322e-01 4.98825610e-01 -8.47211361e-01 -6.90620303e-01 -3.89962137e-01 -1.00605130e+00 -3.35962102e-02 7.87771225e-01 -7.65357852e-01 -3.87675434e-01 6.84147596e-01 -1.02668869e+00 -2.96798766e-01 -4.45379734e-01 2.43809849e-01 -2.64645368e-02 6.36333168e-01 -8.05730045e-01 -4.40077752e-01 -3.93723637e-01 -9.38304484e-01 9.71301496e-01 1.28183931e-01 3.67290936e-02 -1.53781688e+00 7.22554885e-03 -3.36165965e-01 5.28372467e-01 5.30416191e-01 9.64116633e-01 -1.10268807e+00 -7.19821930e-01 -5.72730184e-01 -5.99999428e-01 4.60007936e-01 3.19061093e-02 1.49884671e-02 -7.60831952e-01 -4.62871730e-01 -6.62360251e-01 -1.82063133e-01 1.31985402e+00 2.83786505e-01 1.74577475e+00 -1.25910923e-01 -3.39387000e-01 1.16263819e+00 1.33922851e+00 -5.36999524e-01 7.04393148e-01 -2.11574376e-01 1.60947776e+00 2.10490763e-01 -1.21514328e-01 1.85623780e-01 3.92248809e-01 2.48854563e-01 1.08368444e+00 -4.38030452e-01 -1.97757632e-01 -2.63713509e-01 -3.72231528e-02 5.73177457e-01 4.46321123e-04 -1.70714393e-01 -9.73777115e-01 5.76422095e-01 -1.85718763e+00 -8.99057686e-01 -4.23408866e-01 1.98098695e+00 -1.80924132e-01 2.82742679e-01 2.03598052e-01 3.39136422e-02 6.32583559e-01 6.77204132e-01 -5.76654971e-01 -7.51025856e-01 -2.28285372e-01 6.29556656e-01 5.23584247e-01 4.43295389e-01 -1.47357392e+00 1.01766610e+00 6.15624285e+00 6.57224357e-01 -8.56558800e-01 -2.71272182e-01 7.90879309e-01 1.05573714e-01 -3.19855869e-01 -1.61876366e-01 -1.09693751e-01 1.66741818e-01 5.84512651e-01 8.66170228e-03 1.22383513e-01 1.05482578e+00 -2.85972357e-01 4.25861716e-01 -1.12656009e+00 7.62109399e-01 9.55229625e-03 -1.48496974e+00 4.88208115e-01 1.82783589e-01 5.54692149e-01 3.95862550e-01 -5.39016612e-02 2.09538072e-01 6.04128003e-01 -1.40620279e+00 2.74507970e-01 3.02302480e-01 4.45408791e-01 -9.92259562e-01 8.90508533e-01 -1.74983025e-01 -1.75698090e+00 1.31352982e-02 -4.77387995e-01 -3.76092434e-01 -5.25064170e-02 8.23028803e-01 -6.59522533e-01 9.32876110e-01 6.94528401e-01 1.17688644e+00 -8.87130618e-01 1.05319178e+00 -4.53532815e-01 6.30458295e-01 -3.85473818e-01 3.44664037e-01 6.70566261e-01 -2.59199470e-01 6.49252892e-01 1.20249939e+00 -3.19000743e-02 -1.42614394e-01 3.35233420e-01 1.04699111e+00 -4.74618345e-01 1.71375632e-01 -1.09643924e+00 -6.09989278e-02 2.10171461e-01 1.55825114e+00 -8.70576441e-01 -1.17250592e-01 -6.79811716e-01 1.21419740e+00 9.32023883e-01 4.19970214e-01 -6.78834677e-01 -8.05369616e-01 9.12487030e-01 1.60417870e-01 4.71679389e-01 -4.29093480e-01 -2.74396688e-01 -1.00623643e+00 1.16918698e-01 -2.41458356e-01 9.25273478e-01 -1.46279857e-01 -1.63710952e+00 6.12176001e-01 -4.28185642e-01 -8.22850287e-01 -5.76085746e-02 -9.33994293e-01 -1.54352951e+00 9.31895971e-01 -1.45565879e+00 -1.33655632e+00 -4.61879551e-01 9.95431602e-01 -1.74019888e-01 -3.18788290e-01 7.69370794e-01 3.09614897e-01 -7.59986877e-01 1.00967491e+00 -2.69271076e-01 9.64327216e-01 1.83693528e-01 -1.64489532e+00 1.18559849e+00 1.21858978e+00 5.01247883e-01 6.12802505e-01 8.92966613e-02 -6.20474160e-01 -1.17938566e+00 -1.49166489e+00 4.37328517e-01 -1.53513283e-01 8.05848420e-01 -4.99738544e-01 -1.32933378e+00 9.28605795e-01 -9.48159620e-02 8.14732432e-01 6.69544518e-01 3.12906623e-01 -7.18652666e-01 -2.56174803e-02 -1.14255273e+00 4.20562953e-01 1.46151555e+00 -6.13846004e-01 -2.93162793e-01 2.44470775e-01 8.24919641e-01 -4.35409337e-01 -7.13566422e-01 5.83323896e-01 7.43653849e-02 -1.00601232e+00 1.14304662e+00 -8.97676706e-01 3.65597963e-01 -1.58394009e-01 5.23701198e-02 -1.23747802e+00 -6.24540687e-01 -6.11113191e-01 -5.31083465e-01 8.56840491e-01 3.19957674e-01 -6.34445548e-01 1.17669022e+00 3.80578011e-01 -5.28981268e-01 -9.10505950e-01 -8.25792968e-01 -6.33565962e-01 5.55925444e-02 -5.26314974e-01 7.31473088e-01 9.65323269e-01 -3.43648940e-01 3.41456056e-01 -7.02082142e-02 5.68285108e-01 7.64840722e-01 -2.42674183e-02 8.41919839e-01 -1.45491588e+00 -1.13341242e-01 -6.29210532e-01 -1.19901967e+00 -1.00277746e+00 3.82841796e-01 -1.38209736e+00 -5.10793209e-01 -1.57560456e+00 -2.68223077e-01 -2.28307739e-01 -5.78266919e-01 3.39820504e-01 -2.90214092e-01 2.77852446e-01 -2.68982977e-01 -3.57834518e-01 -7.90919840e-01 7.64626384e-01 1.06341755e+00 -2.70573825e-01 -9.60462540e-02 -5.55581376e-02 -5.63227296e-01 9.73864257e-01 7.29574561e-01 -9.92749408e-02 -4.30595040e-01 -6.23244047e-01 1.08990282e-01 -7.02036142e-01 6.69563532e-01 -1.18021154e+00 2.44805411e-01 4.37019855e-01 5.34448564e-01 -7.28180707e-01 -4.66773957e-02 -8.31679821e-01 -4.76901650e-01 3.85042906e-01 -1.14101112e-01 3.69047999e-01 2.97757924e-01 8.61510813e-01 -2.68263429e-01 2.47627601e-01 5.95779002e-01 -1.03615962e-01 -9.91299510e-01 1.13446808e+00 1.23866521e-01 2.08377793e-01 8.54846418e-01 -3.18695381e-02 -1.41842574e-01 -4.21643645e-01 -8.40219915e-01 5.46737194e-01 4.08524275e-01 3.57767791e-01 8.76697958e-01 -1.70556426e+00 -5.59375226e-01 5.15732765e-01 3.09828639e-01 4.76759791e-01 2.49138415e-01 6.26086473e-01 -9.50513065e-01 -3.69923818e-03 -1.90985933e-01 -5.95472038e-01 -8.03943217e-01 6.30196273e-01 4.81775194e-01 -4.51961547e-01 -9.93585825e-01 1.20335102e+00 3.89981508e-01 -5.75216591e-01 3.25975984e-01 -4.81666416e-01 1.90413669e-02 -5.38731217e-01 4.08789277e-01 3.16031069e-01 1.86260864e-01 -7.09275067e-01 -4.90124375e-01 2.73744553e-01 -4.42218840e-01 5.40332139e-01 1.23993206e+00 1.65875688e-01 -3.64342541e-01 2.62447774e-01 1.27664340e+00 -1.35436669e-01 -9.43619967e-01 -6.52466893e-01 9.83138084e-02 -3.08077425e-01 1.39186218e-01 -1.22575626e-01 -1.65856957e+00 1.11066127e+00 7.67749771e-02 1.80050001e-01 9.74746823e-01 1.17956817e-01 9.60327625e-01 6.65037632e-01 1.83578823e-02 -8.82684946e-01 2.87595332e-01 6.71528995e-01 7.32516885e-01 -1.42539775e+00 5.97068034e-02 -5.67471325e-01 -5.22596650e-02 1.09931076e+00 1.03814042e+00 -8.27451169e-01 1.08831692e+00 -2.65533533e-02 -2.74210721e-01 -8.50007653e-01 -2.87249029e-01 -2.85800189e-01 6.51823401e-01 5.74557543e-01 3.53841543e-01 1.69081371e-02 1.59305409e-01 4.31554675e-01 -4.56593558e-03 -7.00232208e-01 1.98512167e-01 6.06413305e-01 -2.67612547e-01 -9.49453533e-01 3.39256585e-01 8.72666299e-01 -4.43165243e-01 -2.98416287e-01 -7.07417369e-01 9.37243581e-01 -2.49989927e-01 5.39878726e-01 5.97622454e-01 -5.21964908e-01 2.67971188e-01 -1.17781445e-01 1.47189125e-01 -7.09484279e-01 -4.79351521e-01 -5.13121486e-01 -9.96931177e-03 -1.16650474e+00 4.68244627e-02 -1.46843120e-01 -1.32233298e+00 -6.33728862e-01 3.01044341e-02 9.89514589e-03 1.01737361e-02 8.87762845e-01 5.72451413e-01 7.93985188e-01 5.91246009e-01 -7.68463552e-01 -4.88508306e-03 -7.79803276e-01 -7.65608370e-01 6.55060232e-01 4.08129960e-01 -5.89839697e-01 -3.22869420e-01 -8.38368595e-01]
[7.030810356140137, 6.2321271896362305]
13ebc249-5f1b-4dd6-804d-1e41d6648518
progressive-hint-prompting-improves-reasoning
2304.09797
null
https://arxiv.org/abs/2304.09797v4
https://arxiv.org/pdf/2304.09797v4.pdf
Progressive-Hint Prompting Improves Reasoning in Large Language Models
The performance of Large Language Models (LLMs) in reasoning tasks depends heavily on prompt design, with Chain-of-Thought (CoT) and self-consistency being critical methods that enhance this ability. However, these methods do not fully exploit the answers generated by the LLM to guide subsequent responses. This paper proposes a new prompting method, named Progressive-Hint Prompting (PHP), that enables automatic multiple interactions between users and LLMs by using previously generated answers as hints to progressively guide toward the correct answers. PHP is orthogonal to CoT and self-consistency, making it easy to combine with state-of-the-art techniques to further improve performance. We conducted extensive and comprehensive experiments on seven benchmarks. The results show that PHP significantly improves accuracy while remaining highly efficient. For instance, with text-davinci-003, we observed a 4.2% improvement on GSM8K with greedy decoding compared to Complex CoT, and a 46.17% reduction in sample paths with self-consistency. With GPT-4 and PHP, we achieve state-of-the-art performances on SVAMP (89.1% -> 91.9%), GSM8K (92% -> 95.5%), AQuA (76.4% -> 79.9%) and MATH (50.3% -> 53.9%).
['Yu Li', 'Zhenguo Li', 'Enze Xie', 'Zhengying Liu', 'Chuanyang Zheng']
2023-04-19
https-arxiv-org-abs-2304-09797
https://arxiv.org/abs/2304.09797
https://arxiv.org/pdf/2304.09797
null
['math-word-problem-solving', 'gsm8k', 'arithmetic-reasoning', 'math-word-problem-solving', 'math-word-problem-solving']
['knowledge-base', 'natural-language-processing', 'reasoning', 'reasoning', 'time-series']
[-3.07857454e-01 7.47094080e-02 -8.69417787e-02 -4.86152828e-01 -1.04237127e+00 -6.79254174e-01 7.20232785e-01 4.21216011e-01 -4.89347875e-01 5.43736875e-01 3.83014679e-01 -7.67667949e-01 -1.12638481e-01 -6.24232888e-01 -5.00366867e-01 -1.66282982e-01 -3.81977856e-02 4.84899104e-01 4.87550735e-01 -4.91854668e-01 6.33771420e-01 8.95312428e-02 -1.04427922e+00 9.60468590e-01 9.61740315e-01 5.76380610e-01 1.72956243e-01 9.50682878e-01 -5.75581193e-01 1.51356828e+00 -6.16703212e-01 -4.21381503e-01 -1.41038999e-01 -1.72669932e-01 -1.29933953e+00 -5.66144884e-01 3.57950360e-01 -4.32289988e-01 -2.75683522e-01 6.25116408e-01 5.81325293e-01 1.70635536e-01 3.54104757e-01 -1.04643130e+00 -4.69271868e-01 1.01606858e+00 -4.78218168e-01 1.49361864e-01 8.69309902e-01 3.29213172e-01 1.03087497e+00 -9.95430231e-01 3.22083056e-01 1.33227670e+00 5.07794976e-01 5.33569694e-01 -1.11722314e+00 -6.07678831e-01 1.78553835e-01 3.84814262e-01 -1.36423767e+00 -5.19340515e-01 3.34612131e-01 -2.04692245e-01 1.57110000e+00 6.78303361e-01 2.06393287e-01 8.78692210e-01 1.77979931e-01 1.00892830e+00 1.45604920e+00 -5.78861952e-01 2.70699471e-01 2.61212945e-01 6.56553209e-01 8.44049990e-01 -2.69627869e-01 -2.04971150e-01 -1.07306445e+00 -3.91955346e-01 4.42784756e-01 -4.13978428e-01 -2.03849137e-01 4.75560844e-01 -1.22707438e+00 6.33858204e-01 2.57113725e-01 1.53645724e-01 -4.68542486e-01 -4.81649376e-02 2.70342618e-01 3.97461891e-01 1.55414850e-01 6.87883615e-01 -6.10489666e-01 -6.18551612e-01 -9.58824635e-01 6.10064924e-01 1.24672604e+00 8.63543332e-01 4.47849870e-01 -1.29637033e-01 -6.89276695e-01 8.78492773e-01 3.75197738e-01 5.60479522e-01 5.50110877e-01 -1.07973671e+00 8.00330937e-01 5.99642634e-01 3.16445917e-01 -1.10756242e+00 -6.46254241e-01 -7.10494936e-01 -6.80362821e-01 -2.25609124e-01 5.89195549e-01 -1.98854581e-02 -4.54225957e-01 1.70511889e+00 1.77720785e-01 2.26115640e-02 3.80516201e-01 6.86580241e-01 8.56412828e-01 8.76296103e-01 1.50277883e-01 -1.07070193e-01 1.33129287e+00 -1.24509716e+00 -4.77005064e-01 -5.86663008e-01 9.40078557e-01 -8.94471586e-01 1.62440419e+00 7.88360775e-01 -1.29687750e+00 -5.16082346e-01 -6.93133116e-01 -9.03851092e-02 5.15527576e-02 -2.69798394e-02 5.78811109e-01 2.45716587e-01 -1.17690003e+00 4.75490659e-01 -5.83146930e-01 -2.66836762e-01 -7.45246857e-02 2.02922866e-01 2.58018747e-02 -1.91727892e-01 -1.29272652e+00 8.06982696e-01 1.56395331e-01 -2.34187052e-01 -5.26992023e-01 -9.05594885e-01 -4.66644496e-01 2.19164640e-01 5.57107210e-01 -3.87561172e-01 1.96482229e+00 -1.63231000e-01 -1.72269082e+00 5.71011186e-01 -2.58863747e-01 -5.69690108e-01 6.63693190e-01 -7.19846129e-01 -3.81033391e-01 1.92164704e-01 5.90755045e-02 7.01762795e-01 3.45639467e-01 -7.58348942e-01 -7.00471044e-01 1.05723888e-01 1.65973306e-01 1.33385792e-01 -3.66249293e-01 2.72326440e-01 -4.90943462e-01 -3.73899698e-01 7.44774193e-02 -8.22057664e-01 -2.14483783e-01 -5.20468235e-01 -4.82751936e-01 -6.08061612e-01 4.43898559e-01 -8.31851721e-01 1.53680849e+00 -1.74091220e+00 -2.21217930e-01 1.46000743e-01 3.54704469e-01 5.36966980e-01 -2.62946814e-01 7.15767503e-01 2.86542565e-01 -2.42181960e-02 8.39068070e-02 -2.37748653e-01 2.02498451e-01 4.19479620e-04 -5.14108181e-01 -4.04971465e-02 -8.18842091e-03 1.06441247e+00 -9.74282742e-01 -2.64900774e-01 1.24849856e-01 1.48237750e-01 -8.31802547e-01 4.26272094e-01 -4.89609390e-01 1.96511567e-01 -3.00338298e-01 4.33853954e-01 3.82658094e-01 -6.76979005e-01 2.08065823e-01 3.70348126e-01 -4.01584199e-03 8.50164294e-01 -1.22340393e+00 1.51016808e+00 -5.26016414e-01 5.15244842e-01 -1.72325954e-01 -3.61459196e-01 1.03549111e+00 2.10019723e-01 -7.97643512e-02 -1.22522175e+00 -2.38289118e-01 3.91437918e-01 9.20198932e-02 -6.18994474e-01 7.81599581e-01 3.90623629e-01 -1.04861178e-01 5.42527139e-01 -2.69110501e-01 1.51378587e-01 3.30766201e-01 9.00730014e-01 1.36545956e+00 -4.24662590e-01 3.84970844e-01 -3.46783310e-01 5.91204941e-01 -6.46510199e-02 2.87898481e-01 1.22241962e+00 1.95390299e-01 2.12709308e-01 6.52145088e-01 -2.02681482e-01 -5.27382374e-01 -7.57016838e-01 3.41855884e-01 1.29279649e+00 -2.50079066e-01 -1.14302957e+00 -6.60196841e-01 -6.92254841e-01 -2.44006924e-02 1.28007841e+00 -1.61302790e-01 -8.34636912e-02 -5.36163628e-01 -3.22126150e-01 6.94661081e-01 6.98423147e-01 7.29518473e-01 -1.05403566e+00 -6.27367079e-01 4.32185024e-01 -5.74083149e-01 -1.30406606e+00 -9.26871225e-02 1.73897132e-01 -6.97390378e-01 -9.04623508e-01 -4.36525732e-01 -2.63844103e-01 3.59212905e-01 3.09040129e-01 1.33058584e+00 3.65416825e-01 3.58586282e-01 3.05718899e-01 -5.57266712e-01 3.96988429e-02 -6.01491690e-01 2.39010319e-01 -7.22605139e-02 -3.65383387e-01 2.16753706e-01 -3.73789877e-01 -4.06844437e-01 4.03660476e-01 -3.90107542e-01 4.85296577e-01 5.57241201e-01 6.90101624e-01 2.19275296e-01 -2.64193326e-01 4.52215791e-01 -1.05764568e+00 7.73887396e-01 -4.31063861e-01 -3.57146531e-01 2.82402903e-01 -8.24623048e-01 2.38257915e-01 8.83233011e-01 -4.28135842e-01 -9.49195981e-01 -5.61355829e-01 -4.34287786e-01 2.77686030e-01 -1.16319850e-01 6.22785270e-01 2.41928607e-01 2.13139374e-02 9.63148057e-01 2.32479766e-01 -5.58653295e-01 -4.83765990e-01 3.26932847e-01 9.37720239e-01 7.11625218e-01 -8.93685043e-01 3.16968650e-01 -1.66076481e-01 -3.97470862e-01 -5.59734583e-01 -9.19222116e-01 -3.83098334e-01 6.75538853e-02 -1.04446150e-01 2.20969319e-01 -8.24208677e-01 -1.00059843e+00 5.18505871e-01 -1.17327905e+00 -9.78096008e-01 5.90025522e-02 3.11203331e-01 -2.89938182e-01 4.06918824e-01 -1.15138376e+00 -8.24954391e-01 -8.13949227e-01 -1.06159449e+00 7.11212039e-01 3.22238922e-01 -9.06643033e-01 -7.20461130e-01 -2.76254475e-01 5.59343457e-01 7.01074600e-01 -4.65589523e-01 1.00335014e+00 -1.03052199e+00 -4.69527841e-01 -6.81569874e-02 -2.30333552e-01 1.66880533e-01 -3.47786337e-01 -3.11722964e-01 -9.45708513e-01 -1.17760196e-01 -1.86429814e-01 -5.13691247e-01 5.18112898e-01 -1.51564196e-01 9.89476025e-01 -5.70881724e-01 -1.93207741e-01 2.20854014e-01 9.88196611e-01 2.13223383e-01 6.49444938e-01 5.30172229e-01 4.16541338e-01 4.70987558e-01 7.08790720e-01 6.54452860e-01 8.13662469e-01 6.34901643e-01 4.79586422e-02 4.36722338e-01 3.01982480e-04 -4.03420001e-01 6.01910114e-01 9.89616036e-01 4.12429452e-01 -4.37768221e-01 -1.41913378e+00 3.16346526e-01 -1.97885466e+00 -5.69820166e-01 -7.04679668e-01 1.88946152e+00 1.15425646e+00 5.06839871e-01 -7.89423138e-02 3.72062117e-01 1.25818476e-01 1.38739005e-01 -2.71105409e-01 -5.57352602e-01 5.97443134e-02 2.61526495e-01 1.28645971e-01 8.41647923e-01 -4.71124828e-01 1.18764555e+00 5.84588623e+00 9.89110410e-01 -1.00160158e+00 3.80992107e-02 5.35971582e-01 -1.16237909e-01 -3.16727579e-01 -4.56714667e-02 -1.13613808e+00 5.54820180e-01 1.36397433e+00 -1.74128767e-02 4.70968068e-01 6.49541676e-01 1.71002164e-01 -4.81863320e-01 -1.05915201e+00 9.73712802e-01 -2.06431687e-01 -1.37206388e+00 -2.95918018e-01 -4.67681199e-01 6.02383971e-01 2.74188556e-02 -1.45285517e-01 6.97858810e-01 5.14201343e-01 -7.45929956e-01 8.73642445e-01 5.95708966e-01 4.71643329e-01 -6.35619402e-01 5.99767387e-01 9.28585172e-01 -8.54913294e-01 -3.95186990e-02 -1.59791902e-01 -4.66480851e-01 2.91053634e-02 6.45213842e-01 -1.03267002e+00 4.16074187e-01 7.08253860e-01 3.09566677e-01 -7.74121940e-01 8.20344925e-01 -8.10915947e-01 1.06478441e+00 -4.95749503e-01 -3.15368474e-01 3.29423487e-01 4.65189546e-01 3.05348247e-01 1.48860157e+00 3.28793228e-02 3.02504182e-01 3.70039523e-01 5.61795771e-01 -2.07760669e-02 3.36385429e-01 1.31463930e-01 -2.89994385e-02 7.60943353e-01 1.19114614e+00 -5.92334747e-01 -7.69022048e-01 -2.88429141e-01 8.70056748e-01 6.03050947e-01 3.14694703e-01 -8.73245001e-01 -2.70104200e-01 3.71160626e-01 1.63398206e-01 -6.90210164e-02 -2.78798908e-01 -4.15079176e-01 -9.50238168e-01 1.12883270e-01 -1.41942799e+00 5.37731767e-01 -1.06543350e+00 -9.91198719e-01 5.84510088e-01 -1.08383499e-01 -5.37048161e-01 -4.58759159e-01 -3.03481787e-01 -4.88587916e-01 9.61301267e-01 -1.54478168e+00 -6.64452553e-01 -4.02534783e-01 3.44763100e-01 6.22825623e-01 -1.68050751e-02 9.73848224e-01 2.49429286e-01 -2.68123388e-01 7.60004222e-01 -1.17619187e-01 -1.36040077e-01 9.28666055e-01 -1.32571530e+00 7.27819085e-01 8.30098212e-01 2.05733046e-01 8.84055853e-01 6.37260497e-01 -6.93959832e-01 -1.46219563e+00 -8.05200458e-01 1.49421334e+00 -4.35345292e-01 5.46013474e-01 -2.55349934e-01 -1.20954859e+00 5.87207019e-01 3.04239124e-01 -2.61830300e-01 4.71853077e-01 3.66520822e-01 -6.02300227e-01 1.07663395e-02 -8.67416501e-01 8.88738036e-01 7.78551280e-01 -6.33078218e-01 -6.20803118e-01 5.17517805e-01 8.83806288e-01 -1.00321317e+00 -6.40509009e-01 1.43312111e-01 3.51847410e-01 -1.03162265e+00 8.61876786e-01 -5.61734200e-01 4.04968202e-01 -1.13999143e-01 -2.50586271e-01 -1.04306376e+00 -5.94090223e-01 -9.47447002e-01 -4.99505341e-01 1.44494629e+00 4.49963391e-01 -7.39979684e-01 6.64822936e-01 8.63642275e-01 -1.11403570e-01 -9.77479398e-01 -2.19008729e-01 -5.63132346e-01 -2.07396727e-02 -1.02613032e+00 6.06090784e-01 8.73759866e-01 3.65627378e-01 5.29531002e-01 -4.29018259e-01 1.18126370e-01 1.46137059e-01 1.86520685e-02 8.88630211e-01 -8.84063244e-01 -7.11331427e-01 -2.92748600e-01 3.11571300e-01 -1.45808351e+00 5.32099940e-02 -1.05584061e+00 -1.54474661e-01 -1.70127392e+00 1.30900398e-01 -4.29383874e-01 -2.56280661e-01 9.75534499e-01 -4.30512339e-01 -1.92843020e-01 4.04981554e-01 1.11152872e-01 -9.52683926e-01 1.81013972e-01 9.07696426e-01 1.05789833e-01 -3.75504047e-01 -9.85875726e-02 -9.35818017e-01 6.59273446e-01 1.05897927e+00 -6.50430560e-01 -3.77567261e-01 -4.81187761e-01 5.68219006e-01 3.78973067e-01 4.17835712e-01 -1.05886722e+00 7.71140814e-01 -1.00609317e-01 1.12100691e-01 -6.98388398e-01 1.73294738e-01 -1.69931948e-01 -2.93139040e-01 5.43871820e-01 -7.74574518e-01 2.72356838e-01 3.97881687e-01 1.86268508e-01 6.03795052e-02 -9.36420485e-02 4.35812920e-01 -3.23061757e-02 -1.03796947e+00 -3.84803265e-01 -5.74399710e-01 3.92946154e-01 5.02499640e-01 3.88151824e-01 -7.05355227e-01 -4.60968345e-01 -2.12923273e-01 6.44744873e-01 -7.36649483e-02 3.56159776e-01 6.52744710e-01 -9.70726073e-01 -7.50860333e-01 1.49724558e-01 1.06652025e-02 -1.53215334e-01 3.72933954e-01 1.00486732e+00 -4.01649833e-01 7.70741701e-01 2.22675294e-01 -5.94831765e-01 -1.37787259e+00 1.34162635e-01 9.84880179e-02 -6.22546554e-01 -6.86984062e-01 1.17395556e+00 -4.88495409e-01 -5.34111440e-01 4.62157667e-01 -5.39509833e-01 -2.12524403e-02 -2.24856198e-01 8.46965909e-01 5.33984184e-01 3.69607538e-01 1.87170923e-01 -4.91962224e-01 7.09038824e-02 -4.40547138e-01 -2.79770702e-01 1.07323289e+00 9.69313085e-02 -9.42342058e-02 2.71506190e-01 7.80480921e-01 2.37133920e-01 -8.68390322e-01 -7.19547272e-01 3.52357984e-01 -4.34467077e-01 -1.04246788e-01 -1.55422950e+00 -5.53214908e-01 7.92762458e-01 -6.15330338e-02 2.00515568e-01 8.41325223e-01 -1.33642748e-01 1.00397861e+00 7.46069252e-01 5.10401487e-01 -8.65225554e-01 2.98227131e-01 8.97228599e-01 8.26119483e-01 -1.07746851e+00 -3.25791925e-01 -1.50891393e-01 -7.56838739e-01 1.08417773e+00 8.79768491e-01 2.99566865e-01 5.33215031e-02 4.78164822e-01 1.26890287e-01 -1.07357696e-01 -1.38594675e+00 2.60235190e-01 2.75908858e-01 -6.05331175e-02 7.10696518e-01 1.00387186e-01 -3.13344598e-01 1.03360808e+00 -4.40114528e-01 6.05143309e-02 4.09983933e-01 1.07329655e+00 -5.82747221e-01 -1.13405538e+00 -4.48061407e-01 4.41746503e-01 -3.25880140e-01 -3.79738659e-01 -3.75199467e-01 6.42698944e-01 -4.64120895e-01 1.47892988e+00 -3.33349824e-01 -6.93334043e-01 4.04091239e-01 3.53665709e-01 2.78698504e-01 -6.06029868e-01 -9.89642739e-01 -4.47061211e-02 4.11579877e-01 -8.95504177e-01 2.34937772e-01 -4.17702556e-01 -1.69217861e+00 -8.25708985e-01 -2.98769474e-02 4.46558923e-01 3.27081233e-01 1.05839515e+00 5.51096916e-01 6.79724693e-01 5.86900227e-02 -1.51039967e-02 -9.67968762e-01 -9.82403457e-01 -1.26552805e-01 -8.90102237e-02 5.72128268e-03 -1.36639446e-01 -1.02970071e-01 -4.03771251e-01]
[9.828478813171387, 7.487346649169922]
8bee9891-7550-4feb-a647-95deac13edd9
a-study-on-angular-based-embedding-learning
1908.0399
null
https://arxiv.org/abs/1908.03990v1
https://arxiv.org/pdf/1908.03990v1.pdf
A Study on Angular Based Embedding Learning for Text-independent Speaker Verification
Learning a good speaker embedding is important for many automatic speaker recognition tasks, including verification, identification and diarization. The embeddings learned by softmax are not discriminative enough for open-set verification tasks. Angular based embedding learning target can achieve such discriminativeness by optimizing angular distance and adding margin penalty. We apply several different popular angular margin embedding learning strategies in this work and explicitly compare their performance on Voxceleb speaker recognition dataset. Observing the fact that encouraging inter-class separability is important when applying angular based embedding learning, we propose an exclusive inter-class regularization as a complement for angular based loss. We verify the effectiveness of these methods for learning a discriminative embedding space on ASV task with several experiments. These methods together, we manage to achieve an impressive result with 16.5% improvement on equal error rate (EER) and 18.2% improvement on minimum detection cost function comparing with baseline softmax systems.
['Shugong Xu', 'Zongze Ren', 'Zhiyong Chen']
2019-08-12
null
null
null
null
['text-independent-speaker-verification']
['speech']
[-8.16960260e-02 3.46720368e-01 -3.81481051e-01 -8.32934082e-01 -1.15693402e+00 -6.29490316e-01 5.46968877e-01 3.11467471e-03 -5.33827364e-01 3.87039870e-01 2.56518215e-01 -3.75201166e-01 -2.24530138e-02 -1.14064902e-01 -5.00250757e-01 -7.69618690e-01 -1.64607704e-01 1.34257033e-01 -1.60378039e-01 6.96771895e-05 -3.03651188e-02 6.42036974e-01 -1.24590755e+00 -6.54957891e-02 7.54603386e-01 1.10758555e+00 -2.79350638e-01 6.07012808e-01 7.60169998e-02 2.29095832e-01 -7.23494828e-01 -4.29934919e-01 2.11816311e-01 -1.65181700e-02 -6.19987428e-01 -2.08592862e-01 7.71292746e-01 -2.11739689e-01 -5.56899369e-01 8.71131837e-01 9.42784965e-01 1.58652335e-01 8.26861978e-01 -1.27533114e+00 -6.92448199e-01 8.59342158e-01 -4.43277746e-01 2.69694299e-01 3.23250443e-01 -1.57376453e-01 1.32127261e+00 -1.05324233e+00 1.52047142e-01 1.25182998e+00 5.61497509e-01 7.87284911e-01 -1.29885149e+00 -7.39712715e-01 1.40529722e-01 4.35545564e-01 -1.69962430e+00 -1.05451405e+00 1.03628445e+00 -2.03668043e-01 8.65022242e-01 6.68884814e-01 4.81960326e-02 1.25396347e+00 -4.21223134e-01 8.82966578e-01 7.96740651e-01 -5.89405596e-01 3.90805602e-02 7.50612020e-01 4.86425698e-01 5.75232327e-01 -9.72050056e-02 3.23511958e-01 -7.47560918e-01 -7.87353739e-02 2.37559006e-01 -2.85618901e-01 -4.83072191e-01 -2.56810546e-01 -8.97970080e-01 9.99972761e-01 5.47586322e-01 3.56061786e-01 1.58433467e-01 2.08048865e-01 3.07128876e-01 4.68146771e-01 4.10910219e-01 1.01617686e-01 -3.63764226e-01 -1.39296129e-01 -1.06239414e+00 -2.75482416e-01 6.59617305e-01 7.16160417e-01 2.86019266e-01 5.11481106e-01 -3.51624757e-01 1.26227677e+00 8.99034917e-01 6.13952816e-01 6.14562809e-01 -4.12365973e-01 4.50365573e-01 1.62744582e-01 -2.74843246e-01 -5.35120785e-01 -2.93375343e-01 -4.74324942e-01 -5.60594440e-01 4.55788970e-01 4.04455453e-01 -2.16863587e-01 -8.47695947e-01 1.87688982e+00 3.65380406e-01 3.10452521e-01 1.70687228e-01 7.98232198e-01 1.04133260e+00 6.13231361e-01 -2.13236734e-01 -1.39398143e-01 1.46502125e+00 -8.50538135e-01 -8.20495546e-01 -1.23213798e-01 5.00318527e-01 -9.59938765e-01 9.34763730e-01 2.01374382e-01 -7.26379037e-01 -4.13993210e-01 -1.41006899e+00 1.76986661e-02 -2.96982020e-01 4.90687877e-01 3.47645670e-01 1.28611815e+00 -9.43627059e-01 2.73237735e-01 -8.43855798e-01 9.07251146e-03 2.98685133e-01 6.07741773e-01 -5.62452435e-01 3.98131102e-01 -1.08788133e+00 7.94488728e-01 1.07317001e-01 1.15503088e-01 -7.93264270e-01 -7.17831612e-01 -1.07339001e+00 2.45694444e-01 -5.51691577e-02 -4.43760343e-02 9.46340919e-01 -5.87494314e-01 -1.90939188e+00 8.93431604e-01 -4.06132519e-01 -5.34277380e-01 4.69227821e-01 -1.64573982e-01 -5.78488290e-01 -2.87837505e-01 -5.33029199e-01 5.43074906e-01 1.19762516e+00 -8.93477380e-01 -1.08844429e-01 -4.74246860e-01 -2.52999008e-01 -1.31762838e-02 -8.41956496e-01 2.46449023e-01 -1.14141814e-01 -7.26012707e-01 2.10449174e-01 -8.94724667e-01 2.78440237e-01 6.04855828e-02 -4.78508502e-01 -6.28444135e-01 1.16161358e+00 -9.54795361e-01 1.15868211e+00 -2.50132823e+00 9.52487215e-02 2.47982278e-01 4.39455062e-02 3.04484218e-01 -7.47389719e-02 -1.14867695e-01 -1.60182610e-01 1.57426178e-01 -9.28359255e-02 -7.98138678e-01 5.50698996e-01 -8.23481679e-02 -3.84980649e-01 7.92573571e-01 3.51078242e-01 6.46286190e-01 -2.14979023e-01 -4.51829553e-01 2.52765745e-01 1.19178808e+00 -5.46911895e-01 1.69088528e-01 3.60618263e-01 9.91066992e-02 4.51135263e-02 7.64421999e-01 7.58614659e-01 2.90035576e-01 -7.36988187e-02 -1.66174978e-01 2.34586000e-01 5.59811831e-01 -1.46110046e+00 1.50028527e+00 -6.48692846e-01 1.31580973e+00 3.00869584e-01 -1.03229880e+00 9.94452596e-01 5.33864915e-01 7.22493902e-02 -2.10242450e-01 1.45649418e-01 1.31024897e-01 7.18807504e-02 -2.10523263e-01 2.61854798e-01 -2.25178748e-01 1.15208872e-01 1.86597556e-01 5.24662614e-01 1.96098834e-01 -3.99981171e-01 -6.09268807e-03 5.11047602e-01 -3.61738175e-01 2.33693849e-02 -3.57725650e-01 7.91371644e-01 -9.83201146e-01 4.01203603e-01 3.60986322e-01 -4.23105329e-01 5.84529221e-01 4.08230513e-01 -1.94495618e-02 -6.44787729e-01 -1.19051611e+00 -8.57050598e-01 1.18427610e+00 -1.43118724e-01 -2.98735946e-01 -5.81664622e-01 -9.12216902e-01 1.69739842e-01 7.29086459e-01 -6.73533142e-01 -6.27886280e-02 -6.39070988e-01 -6.73980236e-01 8.80559623e-01 8.01672995e-01 -6.29665032e-02 -5.33052206e-01 2.98601002e-01 -1.50482953e-01 1.17750257e-01 -1.04482853e+00 -9.17873085e-01 4.56960052e-01 -4.90398943e-01 -5.50085425e-01 -7.03985751e-01 -1.03902721e+00 5.14789701e-01 -1.56580970e-01 5.66748857e-01 -3.65829468e-01 -2.19523713e-01 1.40819237e-01 -9.56770703e-02 -4.54989970e-01 -3.07754189e-01 1.87977776e-01 5.38790643e-01 2.15695575e-01 4.22683537e-01 -3.42209399e-01 -4.22373861e-01 4.25709724e-01 -4.59683895e-01 -8.72576654e-01 3.32669735e-01 1.02796173e+00 1.23677067e-01 -5.26318967e-01 8.39106262e-01 -1.34686992e-01 5.09014189e-01 3.16818915e-02 -7.73849964e-01 3.06918591e-01 -7.81653464e-01 4.82651591e-01 2.11291656e-01 -5.65347135e-01 -7.48886764e-01 -5.26323840e-02 -5.59579730e-01 -3.87100101e-01 -2.82951947e-02 8.45996961e-02 -4.14614290e-01 -3.08157384e-01 5.52887619e-01 2.42195874e-01 1.79439500e-01 -5.60517013e-01 3.31163764e-01 1.22159731e+00 1.20653749e-01 -2.59075373e-01 8.23920727e-01 2.40390629e-01 -5.39829850e-01 -1.20350134e+00 -2.55782247e-01 -5.47599196e-01 -4.43252325e-01 1.04467504e-01 8.03938687e-01 -9.23824847e-01 -1.11638618e+00 5.82743473e-02 -1.04057622e+00 -1.71935335e-02 -1.92567661e-01 8.46144497e-01 -6.56093508e-02 4.49339330e-01 -4.71197397e-01 -1.07612014e+00 -5.14159739e-01 -1.40877950e+00 1.19134879e+00 1.95807919e-01 -1.50792971e-01 -1.14430845e+00 1.55975670e-01 4.44987714e-01 6.33176267e-01 -2.95519620e-01 3.11811507e-01 -1.05388606e+00 -1.68426141e-01 -3.91725183e-01 -5.49994186e-02 6.39639258e-01 1.40740499e-01 8.14076811e-02 -1.68597746e+00 -5.33369243e-01 -2.23424792e-01 -1.48165047e-01 1.10387695e+00 3.49477142e-01 1.03611219e+00 -3.12936157e-01 -2.55977482e-01 7.42061675e-01 1.05880523e+00 -3.77889872e-02 3.22877973e-01 -4.33666669e-02 5.24608791e-01 3.87493968e-01 1.24836400e-01 3.36929500e-01 1.21351415e-02 1.25413787e+00 1.85723230e-01 6.90957606e-02 -3.10662866e-01 -3.06561999e-02 9.74169075e-01 7.63185740e-01 2.19892845e-01 -7.01309964e-02 -6.44762337e-01 5.50838947e-01 -1.32050419e+00 -1.04804075e+00 2.67107964e-01 2.51484203e+00 8.44473064e-01 4.87701371e-02 2.58591473e-01 3.56876343e-01 5.99412560e-01 2.42048413e-01 -2.98605025e-01 -5.99934757e-01 -1.88365892e-01 2.52839476e-01 3.23117793e-01 1.10415053e+00 -1.18957686e+00 7.77181625e-01 6.18315983e+00 9.42557275e-01 -1.59259105e+00 3.20967883e-01 4.67226416e-01 -2.67103702e-01 -2.32789397e-01 -4.20564711e-01 -1.27472961e+00 3.85088354e-01 9.94560301e-01 1.03459433e-01 2.25146323e-01 9.48703229e-01 -5.33424206e-02 6.16670847e-01 -1.30953217e+00 1.41582370e+00 3.72531980e-01 -9.84837532e-01 -4.59634751e-01 9.72167403e-02 4.01585251e-01 3.30014676e-02 4.35273290e-01 5.50406575e-01 -1.30920291e-01 -1.28969860e+00 6.17413461e-01 -9.86200124e-02 8.63613725e-01 -6.69879138e-01 8.06004345e-01 -6.12362362e-02 -1.07933652e+00 2.10568547e-01 -3.03041220e-01 4.59927291e-01 2.64156848e-01 4.83651936e-01 -1.10196698e+00 2.45028362e-01 5.26421487e-01 2.82776415e-01 -3.59545380e-01 9.26788270e-01 -4.27188873e-01 9.89267468e-01 -7.09000647e-01 -1.87560096e-01 -9.39094350e-02 -5.39632067e-02 7.98720002e-01 1.40097666e+00 4.28176485e-02 -5.89479685e-01 -3.37923050e-01 6.70379877e-01 -3.78395289e-01 1.69679821e-01 -3.75003457e-01 9.49700829e-03 5.05211890e-01 1.13662493e+00 -7.16983527e-02 -1.56733230e-01 -3.05078536e-01 1.06083560e+00 2.54413486e-01 3.94916862e-01 -1.24082255e+00 -6.20123565e-01 1.11089396e+00 -1.30878240e-01 5.14597058e-01 -1.78815737e-01 -3.29717785e-01 -1.20284331e+00 3.52010757e-01 -5.91730833e-01 2.64104456e-01 6.24275999e-04 -1.08945680e+00 7.92686522e-01 -2.55733728e-01 -1.09168708e+00 -1.89449742e-01 -8.39860141e-01 -7.84237921e-01 8.62519979e-01 -1.66084433e+00 -1.21648681e+00 4.81025130e-03 4.37262416e-01 4.50494021e-01 -4.22885120e-01 8.98871362e-01 6.05661035e-01 -8.54737222e-01 1.63764250e+00 3.34519953e-01 3.17680478e-01 1.03660214e+00 -1.45353317e+00 4.80329655e-02 6.97457850e-01 7.41322815e-01 6.14089906e-01 6.34975493e-01 2.54716843e-01 -1.30907893e+00 -7.94417202e-01 1.02332449e+00 -5.54078043e-01 5.44148505e-01 -7.39009559e-01 -8.02242756e-01 8.51215184e-01 3.73782098e-01 3.05748999e-01 1.04585123e+00 5.43208122e-01 -8.12365651e-01 -4.42956179e-01 -1.23013842e+00 2.44960502e-01 6.45652950e-01 -8.16210270e-01 -5.75517833e-01 3.64533365e-01 7.93134272e-01 -2.00872272e-01 -9.41440046e-01 3.63939643e-01 7.56038368e-01 -5.11657238e-01 1.29917610e+00 -6.47393525e-01 -1.89278349e-01 -2.30652183e-01 -4.32305872e-01 -1.28688169e+00 1.18453326e-02 -6.87496483e-01 -2.51166344e-01 1.65237200e+00 8.86979282e-01 -9.73464906e-01 7.97325134e-01 4.64581609e-01 -1.38782188e-01 -6.40630662e-01 -1.49078763e+00 -1.04651499e+00 1.68179750e-01 -5.60261488e-01 3.53444815e-01 1.14811492e+00 3.86529684e-01 4.73287523e-01 -4.28334147e-01 5.63437521e-01 7.43127048e-01 -1.16645783e-01 6.29398704e-01 -8.86450827e-01 -4.57644910e-01 -6.99195147e-01 -8.85064423e-01 -1.20698476e+00 5.00446856e-01 -1.09419930e+00 -1.72447711e-01 -9.28968370e-01 -6.84416145e-02 -4.68648225e-01 -6.77064478e-01 5.00379503e-01 -1.83219954e-01 3.10385704e-01 9.22654867e-02 -2.41437644e-01 -2.45183572e-01 8.09562325e-01 4.88949835e-01 -6.11197174e-01 -3.23071867e-01 1.20177850e-01 -4.91010934e-01 3.88833165e-01 8.21159124e-01 -3.86992246e-01 -6.72112182e-02 -3.10085624e-01 -4.90427941e-01 -1.98628783e-01 1.81598887e-01 -7.65478313e-01 4.77191098e-02 2.73440421e-01 1.38323396e-01 -1.56043380e-01 7.36487031e-01 -7.68326283e-01 -4.70108569e-01 3.03014725e-01 -4.76617724e-01 -5.47570527e-01 3.95138681e-01 3.34513336e-01 -5.17577410e-01 -2.46116057e-01 9.17636037e-01 9.60812628e-01 -4.72390294e-01 1.53834388e-01 1.84077844e-02 -2.19944701e-01 9.72289205e-01 -1.62850395e-01 -3.21427584e-02 -3.38427663e-01 -1.10579133e+00 4.36868519e-02 -1.00232013e-01 6.21640861e-01 6.46493673e-01 -1.45793426e+00 -1.09833360e+00 5.43961823e-01 2.52305150e-01 -5.08553684e-01 3.21083777e-02 9.73154187e-01 -1.41644731e-01 5.89724183e-01 2.42635176e-01 -8.23395669e-01 -1.92170203e+00 2.58532256e-01 4.54901665e-01 7.02434778e-02 -2.10821807e-01 1.43976498e+00 9.41701308e-02 -6.92813158e-01 7.16529191e-01 -3.64676982e-01 -8.25373530e-02 1.09379336e-01 6.80184484e-01 2.62421131e-01 3.32229614e-01 -7.07819700e-01 -8.70493472e-01 4.95834529e-01 -1.82243019e-01 -2.17490777e-01 1.24260545e+00 1.11232981e-01 4.07589972e-01 2.69041687e-01 1.57722640e+00 6.05156958e-01 -1.09381831e+00 -2.55816638e-01 -1.92895845e-01 -3.88787746e-01 2.83133060e-01 -5.02295911e-01 -9.83915567e-01 1.23396766e+00 1.17901623e+00 2.81153738e-01 6.01986706e-01 2.48009995e-01 4.56341088e-01 1.13469116e-01 -2.54119635e-01 -8.42369497e-01 2.46578213e-02 1.85901105e-01 1.14468145e+00 -1.74884629e+00 -2.48951003e-01 -3.46619427e-01 -5.64605474e-01 1.05882716e+00 2.95076638e-01 1.03778340e-01 9.98093307e-01 3.81439775e-01 2.26118237e-01 1.70400873e-01 -4.20439601e-01 -3.39402035e-02 7.47489333e-01 4.54823166e-01 8.17363441e-01 3.65639627e-01 -1.41330138e-01 5.41215301e-01 -3.12498361e-01 -9.13420320e-01 -5.50027266e-02 3.70218128e-01 -3.69861513e-01 -1.22212255e+00 -5.75059474e-01 6.53949827e-02 -4.86024916e-01 -7.42624328e-02 -3.58629525e-01 5.04155159e-01 -3.87391210e-01 1.02215195e+00 -6.59998953e-02 -4.31621999e-01 1.30035847e-01 3.97651583e-01 4.68821466e-01 -2.24654496e-01 -3.68514121e-01 6.77094311e-02 1.18142376e-02 -1.57147333e-01 3.82298529e-02 -7.00610280e-01 -1.05981410e+00 -1.62347779e-01 -9.69204903e-01 1.61513969e-01 1.07395196e+00 7.67380595e-01 2.58400857e-01 5.16662121e-01 7.94017673e-01 -5.19297183e-01 -1.15610600e+00 -1.38430655e+00 -5.22120059e-01 2.92950243e-01 8.20672631e-01 -6.55880094e-01 -7.41256058e-01 -3.39491330e-02]
[14.303635597229004, 6.077650547027588]
e57c94a7-05bd-4fab-8767-4690a6694165
margin-aware-unsupervised-domain-adaptation
null
null
https://aclanthology.org/2020.findings-emnlp.315
https://aclanthology.org/2020.findings-emnlp.315.pdf
Margin-aware Unsupervised Domain Adaptation for Cross-lingual Text Labeling
Unsupervised domain adaptation addresses the problem of leveraging labeled data in a source domain to learn a well-performing model in a target domain where labels are unavailable. In this paper, we improve upon a recent theoretical work (Zhang et al., 2019b) and adopt the Margin Disparity Discrepancy (MDD) unsupervised domain adaptation algorithm to solve the cross-lingual text labeling problems. Experiments on cross-lingual document classification and NER demonstrate the proposed domain adaptation approach advances the state-of-the-art results by a large margin. Specifically, we improve MDD by efficiently optimizing the margin loss on the source domain via Virtual Adversarial Training (VAT). This bridges the gap between theory and the loss function used in the original work Zhang et al.(2019b), and thereby significantly boosts the performance. Our numerical results also indicate that VAT can remarkably improve the generalization performance of both domains for various domain adaptation approaches.
['Bing Xiang', 'Kathleen McKeown', 'Cicero Nogueira dos santos', 'Feng Nan', 'Henghui Zhu', 'Ramesh Nallapati', 'Dejiao Zhang']
2020-11-01
null
null
null
findings-of-the-association-for-computational
['cross-lingual-document-classification']
['natural-language-processing']
[ 1.75053552e-01 2.96432190e-02 -6.19624078e-01 -5.55230141e-01 -1.28496885e+00 -9.51510131e-01 5.35405517e-01 -1.16168726e-02 -5.42629182e-01 1.07954741e+00 -2.07248721e-02 -4.02574807e-01 8.62600803e-02 -5.93359470e-01 -7.60813773e-01 -8.65188122e-01 3.73229027e-01 4.88419026e-01 -3.58978398e-02 -2.53470361e-01 4.61693518e-02 2.94903606e-01 -9.46945488e-01 -6.23294376e-02 1.12598336e+00 9.64503586e-01 -1.40084460e-01 2.09802747e-01 -2.38848075e-01 2.95643717e-01 -4.46552306e-01 -6.07327819e-01 4.76857394e-01 -3.67380112e-01 -8.34794223e-01 1.75140217e-01 4.34168756e-01 3.27162445e-02 -1.60081565e-01 1.12573195e+00 5.62229931e-01 3.71176660e-01 1.02206051e+00 -1.25566649e+00 -1.07962584e+00 4.74344671e-01 -7.67767906e-01 1.42970076e-02 -1.81281805e-01 -3.84121060e-01 8.53378475e-01 -1.07201397e+00 5.82819700e-01 9.17658448e-01 1.01852477e+00 9.87713754e-01 -1.21923006e+00 -1.00219667e+00 2.20064148e-01 1.96547713e-02 -1.50667691e+00 -4.36697781e-01 1.00710142e+00 -4.52735811e-01 5.73100507e-01 -3.36126685e-01 -2.43469715e-01 1.35135770e+00 -2.91424394e-01 7.63149381e-01 1.17039049e+00 -8.25131118e-01 4.33117747e-01 4.57841456e-01 1.02564013e-02 3.57998163e-01 1.70714080e-01 9.67640951e-02 -2.70132363e-01 -1.89865351e-01 4.85369712e-01 -3.63491595e-01 8.09135512e-02 -8.01627874e-01 -7.79251873e-01 1.07884502e+00 2.07423314e-01 3.10050130e-01 -7.47271553e-02 -4.20791060e-01 6.42387688e-01 4.72222984e-01 8.94690096e-01 2.71092266e-01 -8.20726752e-01 7.99556747e-02 -7.98547506e-01 9.62247029e-02 6.61786914e-01 9.99509871e-01 6.73436999e-01 1.22078568e-01 1.95351690e-01 1.26450264e+00 1.58278361e-01 5.45081437e-01 7.68777609e-01 -7.50808060e-01 8.19337845e-01 2.98693657e-01 1.78998217e-01 -4.15700585e-01 -1.67342797e-01 -4.62064683e-01 -6.41060770e-01 1.72022164e-01 7.59084761e-01 -4.72353727e-01 -6.56460583e-01 2.17826581e+00 4.79750454e-01 2.29063793e-04 4.94429410e-01 5.89096963e-01 1.53151616e-01 5.50304830e-01 4.89446551e-01 -9.87981930e-02 9.68616068e-01 -9.84727740e-01 -6.49942756e-01 -4.31589216e-01 9.65874374e-01 -6.96011484e-01 1.28652847e+00 2.33526006e-01 -8.26135874e-01 -6.47220433e-01 -1.20368314e+00 -2.05108881e-01 -5.63971996e-01 1.52692333e-01 3.98325711e-01 9.98477757e-01 -8.10938597e-01 4.07424212e-01 -5.84467828e-01 -5.40259898e-01 5.20436227e-01 3.64326715e-01 -5.91668487e-01 -2.34506339e-01 -1.51203358e+00 8.92627776e-01 3.98050010e-01 -6.15687788e-01 -4.91671741e-01 -8.38091314e-01 -8.49162161e-01 -2.79473662e-01 2.11909831e-01 -4.05653894e-01 1.37046766e+00 -1.19912028e+00 -1.72771847e+00 1.22964668e+00 -1.64077822e-02 -6.95272326e-01 5.89615583e-01 -1.45285130e-01 -5.37329078e-01 -5.09734638e-02 2.52343506e-01 4.66547996e-01 7.09535062e-01 -1.17240274e+00 -6.55879676e-01 -6.20621800e-01 -2.44195580e-01 2.46389508e-01 -9.07613635e-01 -2.01842904e-01 -1.86404333e-01 -9.65364933e-01 -3.25620919e-01 -9.57007647e-01 1.86715461e-02 -2.54466664e-02 -8.41584429e-03 -4.02164191e-01 8.90474677e-01 -7.06173718e-01 9.30794299e-01 -2.36446476e+00 -1.33635346e-02 8.36974755e-02 -1.74000591e-01 4.50538397e-01 -1.10187188e-01 3.13682109e-01 -2.95106053e-01 1.43518671e-01 -7.15294123e-01 -6.91796720e-01 1.15566798e-01 3.95057917e-01 -3.53472352e-01 6.00793600e-01 1.23374425e-01 7.37238646e-01 -6.69620395e-01 -5.06184995e-01 -1.04907630e-02 3.50055903e-01 -4.68238175e-01 1.26269385e-01 -1.19786717e-01 4.67783719e-01 -4.78538901e-01 4.06257242e-01 8.19649398e-01 -5.21053001e-02 3.66568625e-01 1.73959434e-01 1.26770467e-01 2.89138019e-01 -9.19955671e-01 2.00836110e+00 -7.12919533e-01 5.13254821e-01 7.92234391e-02 -1.62010825e+00 1.25720179e+00 3.51076633e-01 5.77861667e-01 -6.17762506e-01 -3.96877341e-03 3.45642656e-01 -4.87584978e-01 -1.37926906e-01 1.59871876e-01 -7.16628134e-01 -4.95190114e-01 3.60353380e-01 1.58154875e-01 1.46882787e-01 -1.66361228e-01 -9.02327001e-02 6.17590070e-01 2.37839088e-01 5.16857088e-01 -5.78339934e-01 7.19368517e-01 1.21172711e-01 6.16766810e-01 5.89729607e-01 -6.01906717e-01 4.24159557e-01 2.62602806e-01 5.16306236e-02 -1.31819332e+00 -1.16029811e+00 -4.73816961e-01 1.47776163e+00 -6.63900301e-02 2.15859577e-01 -9.79179084e-01 -1.25662637e+00 2.08036482e-01 8.88287485e-01 -7.61255383e-01 -3.74546081e-01 -5.69293141e-01 -6.64004505e-01 7.66643345e-01 8.53799999e-01 6.36400044e-01 -5.44706702e-01 1.79706186e-01 1.25905380e-01 -3.65616888e-01 -1.29207885e+00 -7.18209743e-01 3.96355242e-01 -9.35825288e-01 -6.87909842e-01 -1.06930649e+00 -1.31944883e+00 7.21014261e-01 -5.98044135e-02 9.03319180e-01 -7.32986152e-01 3.28017533e-01 3.64831060e-01 -4.02049094e-01 -3.75259995e-01 -6.24746084e-01 3.90366763e-01 4.74854678e-01 1.50266707e-01 8.23728204e-01 -6.66039109e-01 -1.55486494e-01 4.47024107e-01 -7.99287200e-01 -3.44337702e-01 2.40359381e-01 1.08513951e+00 7.63179600e-01 2.39138871e-01 1.17143977e+00 -1.18423378e+00 5.97831666e-01 -7.07831562e-01 -6.62342131e-01 1.91035360e-01 -1.19556582e+00 3.08227271e-01 1.02609873e+00 -6.30255699e-01 -1.44489133e+00 3.38728353e-02 1.52223567e-02 -1.63678706e-01 -2.74809897e-01 1.29576176e-01 -5.65941155e-01 -1.51901513e-01 7.62111723e-01 2.02100679e-01 -1.98390305e-01 -7.91983187e-01 4.27247554e-01 9.83629704e-01 6.82152331e-01 -9.01259959e-01 1.01056385e+00 4.50473219e-01 -2.68311203e-01 -2.86958873e-01 -1.00456500e+00 -5.36085844e-01 -1.00412250e+00 3.74571979e-01 7.23349631e-01 -1.02363896e+00 -1.04408503e-01 3.58962238e-01 -8.05724919e-01 -4.94572520e-01 -4.34935570e-01 4.88921732e-01 -6.58815563e-01 4.80342060e-01 -2.96102732e-01 -6.84682250e-01 -2.19829574e-01 -6.71298683e-01 8.80709529e-01 1.47514731e-01 5.77220507e-02 -1.53628647e+00 2.61365682e-01 2.50073224e-01 1.56711891e-01 1.15321249e-01 9.29368436e-01 -1.08928931e+00 3.27971458e-01 -1.56992257e-01 -1.20548911e-01 8.73124242e-01 2.84855574e-01 -6.62711084e-01 -1.04802024e+00 -3.82769823e-01 1.96513623e-01 -3.24660569e-01 6.29608631e-01 2.59173304e-01 1.06262445e+00 -1.27239972e-01 -2.60562062e-01 7.16729760e-01 1.40780127e+00 3.48831326e-01 3.16912055e-01 6.42070055e-01 5.22994220e-01 6.87367022e-01 8.38257372e-01 2.66285539e-01 3.31606805e-01 7.23991334e-01 -1.72424659e-01 -2.49215961e-01 -8.78332630e-02 -4.91658896e-01 4.94975120e-01 6.80437088e-01 2.64906347e-01 -2.37512231e-01 -9.44615662e-01 8.36106420e-01 -1.70432866e+00 -5.81851244e-01 3.56067777e-01 2.33659792e+00 1.26275909e+00 5.19856699e-02 1.78332195e-01 5.10999374e-02 9.82266366e-01 -2.92160928e-01 -8.46773505e-01 -4.48634505e-01 -2.56821096e-01 3.69885504e-01 8.46056879e-01 5.58007419e-01 -1.41423559e+00 1.07770574e+00 5.88252258e+00 1.04450965e+00 -9.04204428e-01 4.43498105e-01 5.86765468e-01 2.24811569e-01 2.26759203e-02 -2.39100561e-01 -9.56403494e-01 4.61586267e-01 1.09665155e+00 -5.70104718e-01 3.83132666e-01 1.16008186e+00 -1.41803771e-01 5.12232840e-01 -1.12855387e+00 8.35731089e-01 2.11269304e-01 -7.31766701e-01 -2.96731949e-01 1.69599921e-01 1.09791970e+00 -6.86607957e-02 4.18765217e-01 6.22177660e-01 5.20667851e-01 -7.17644811e-01 2.57185936e-01 -1.32629991e-01 1.02749002e+00 -9.18247581e-01 6.41073525e-01 3.45485002e-01 -9.37770784e-01 1.35445017e-02 -4.42353547e-01 2.87257463e-01 4.39236546e-03 4.87271965e-01 -8.80291402e-01 5.15744925e-01 4.45846409e-01 7.17686594e-01 -3.15561146e-01 4.53047663e-01 -2.07693666e-01 8.01092207e-01 -2.98957199e-01 2.53434867e-01 3.24395001e-01 -1.98759615e-01 4.14569050e-01 1.32069767e+00 1.56370223e-01 -1.83141723e-01 1.08050078e-01 5.18602788e-01 -5.30099273e-01 3.94365758e-01 -6.35829270e-01 1.30941465e-01 6.03225052e-01 7.91331053e-01 -2.72088736e-01 -3.76402438e-01 -4.72694725e-01 1.18039250e+00 4.32784319e-01 4.70154524e-01 -7.95965314e-01 -5.31482041e-01 7.33873367e-01 -8.52345079e-02 4.97024775e-01 -1.58090904e-01 -6.97316408e-01 -1.13961649e+00 1.90086961e-01 -7.61212647e-01 6.16126478e-01 -2.86591887e-01 -1.79960549e+00 3.53904605e-01 -6.88406778e-03 -1.46259928e+00 -3.45857888e-01 -7.73447633e-01 -2.86621917e-02 1.02954948e+00 -1.85481691e+00 -1.05163813e+00 2.65428156e-01 7.86702096e-01 6.01921141e-01 -4.05884266e-01 9.82102215e-01 4.97209251e-01 -5.01530945e-01 1.24568343e+00 1.07345796e+00 3.63092124e-01 1.20588279e+00 -1.37423241e+00 5.95996022e-01 8.24840486e-01 4.80562411e-02 5.50194979e-01 3.58146995e-01 -3.50446761e-01 -8.33209217e-01 -1.19503558e+00 8.67856622e-01 -5.42036951e-01 7.61373758e-01 -5.35820961e-01 -1.04839671e+00 7.90739655e-01 2.72985678e-02 1.48293115e-02 9.51396406e-01 1.02510415e-01 -8.16886663e-01 -1.55110911e-01 -1.63295317e+00 3.58884871e-01 8.37298453e-01 -6.92442536e-01 -7.34646857e-01 3.22865218e-01 6.51701093e-01 -1.83180600e-01 -1.04578376e+00 3.51980835e-01 4.25521255e-01 -4.97534811e-01 1.04684496e+00 -9.39943194e-01 2.46962816e-01 1.25364885e-01 -3.19119215e-01 -1.46596098e+00 -4.16741408e-02 -4.38844889e-01 -3.98005396e-02 1.63534904e+00 4.23624218e-01 -1.03398120e+00 7.88361013e-01 5.92806399e-01 9.59392264e-02 -3.31901670e-01 -1.05263782e+00 -1.29909253e+00 1.17515290e+00 -3.03097606e-01 5.34416020e-01 1.36227155e+00 8.46057013e-02 3.17824811e-01 -2.24093869e-01 1.01142161e-01 8.90412033e-01 -1.14341669e-01 4.60254878e-01 -1.33469927e+00 -2.03638211e-01 -2.68346071e-01 8.24745558e-03 -1.33656263e+00 7.93025315e-01 -1.13163006e+00 -1.02708014e-02 -1.04022920e+00 -1.79809913e-01 -7.34534979e-01 -6.27402127e-01 5.78356147e-01 -1.14649072e-01 1.74225256e-01 -1.38531281e-02 2.54449874e-01 -2.63162583e-01 5.76231241e-01 9.78660583e-01 -1.38732225e-01 -9.47187915e-02 4.76169623e-02 -1.10799956e+00 6.43437505e-01 1.03595233e+00 -7.67657340e-01 -4.27342236e-01 -4.59696084e-01 -3.82110000e-01 -2.59636730e-01 1.78019628e-02 -7.18104422e-01 8.28561187e-02 -7.74973482e-02 2.97336131e-01 -1.14754150e-02 8.16532522e-02 -9.80964959e-01 -6.05493307e-01 7.81882331e-02 -6.98536694e-01 -2.44894370e-01 4.73721325e-01 7.45962501e-01 -1.95878923e-01 -3.11021715e-01 1.22614574e+00 2.75475383e-01 -8.07387650e-01 1.59591079e-01 -8.88568722e-03 5.69222391e-01 9.79461014e-01 -6.27742559e-02 -1.85408801e-01 -1.50522083e-01 -7.38402843e-01 -7.46109756e-03 4.91327018e-01 3.47621530e-01 -3.73543128e-02 -1.53174722e+00 -7.77667105e-01 2.80043751e-01 3.76262397e-01 -1.80462003e-01 -2.15622392e-02 4.49618459e-01 -3.93046178e-02 4.74388599e-01 -1.12052940e-01 -2.84632564e-01 -8.60599101e-01 6.48988366e-01 3.44176710e-01 -4.34598982e-01 -4.15466398e-01 8.55318189e-01 4.30055946e-01 -8.31128895e-01 1.98750451e-01 1.81117967e-01 3.85782197e-02 -2.03572467e-01 3.75212967e-01 3.24691117e-01 1.68428317e-01 -5.40284634e-01 -3.49399418e-01 7.72791147e-01 -2.84209430e-01 -2.53571242e-01 1.07155824e+00 -4.13839936e-01 5.86487830e-01 3.13541621e-01 1.62945402e+00 2.34282970e-01 -1.47816896e+00 -7.56922483e-01 8.56620967e-02 -1.54761299e-01 3.42164598e-02 -1.02862859e+00 -8.55411410e-01 9.75353122e-01 7.01421261e-01 -9.63620655e-03 1.29034519e+00 -1.08377665e-01 9.67896819e-01 3.48746687e-01 2.48574942e-01 -1.48569047e+00 -1.41084343e-01 4.38429743e-01 4.42646623e-01 -1.41603684e+00 -2.90559024e-01 -2.21763939e-01 -8.75094116e-01 9.14752841e-01 7.03165710e-01 -7.87331760e-02 5.93574286e-01 2.69730300e-01 2.67387331e-01 4.43135321e-01 -3.15707207e-01 -4.06803489e-02 1.37532398e-01 1.09995699e+00 2.68694848e-01 1.20861799e-01 -3.70441794e-01 9.45847452e-01 -3.14592011e-02 -6.74193054e-02 1.26918226e-01 7.35824823e-01 -1.77441493e-01 -1.64322364e+00 -4.65786129e-01 -1.27917424e-01 -5.35781682e-01 -2.16637045e-01 -4.73378986e-01 1.00021696e+00 5.74057624e-02 9.20274198e-01 -1.31885618e-01 -1.69450983e-01 4.35501873e-01 6.51846290e-01 3.04936767e-01 -4.56824452e-01 -2.33748689e-01 -1.21605664e-01 -1.54696733e-01 8.56497437e-02 -4.27073181e-01 -7.48216510e-01 -1.18410969e+00 -1.40601639e-02 -1.60695568e-01 3.85037005e-01 7.64632523e-01 9.46757853e-01 3.65980238e-01 1.68611586e-01 9.53033149e-01 -2.18547180e-01 -1.12749040e+00 -9.73832369e-01 -8.16898525e-01 6.84550941e-01 3.31213862e-01 -7.64304340e-01 -5.71473718e-01 4.13388938e-01]
[10.349250793457031, 3.191943883895874]
3c84c357-ba99-4fa4-be60-2c67b7f95723
meta-review-generation-with-checklist-guided
2305.14647
null
https://arxiv.org/abs/2305.14647v1
https://arxiv.org/pdf/2305.14647v1.pdf
Meta-review Generation with Checklist-guided Iterative Introspection
Opinions in the scientific domain can be divergent, leading to controversy or consensus among reviewers. However, current opinion summarization datasets mostly focus on product review domains, which do not account for this variability under the assumption that the input opinions are non-controversial. To address this gap, we propose the task of scientific opinion summarization, where research paper reviews are synthesized into meta-reviews. To facilitate this task, we introduce a new ORSUM dataset covering 10,989 paper meta-reviews and 40,903 paper reviews from 39 conferences. Furthermore, we propose the Checklist-guided Iterative Introspection (CGI$^2$) approach, which breaks down the task into several stages and iteratively refines the summary under the guidance of questions from a checklist. We conclude that (1) human-written summaries are not always reliable since many do not follow the guideline, and (2) the combination of task decomposition and iterative self-refinement shows promising discussion involvement ability and can be applied to other complex text generation using black-box LLM.
['Heng Ji', 'Lu Wang', 'Hou Pong Chan', 'Mankeerat Sidhu', 'Qi Zeng']
2023-05-24
null
null
null
null
['review-generation']
['natural-language-processing']
[ 4.02337313e-01 6.17888808e-01 -5.73779345e-01 -1.38197139e-01 -8.46131146e-01 -9.19060528e-01 6.00743771e-01 6.42570734e-01 -8.49663690e-02 1.12275600e+00 5.93518257e-01 -7.08634138e-01 -2.08387002e-01 -4.34357345e-01 -5.29660106e-01 -1.92316741e-01 7.98120439e-01 1.04055725e-01 -1.92107603e-01 -7.64273256e-02 9.44146693e-01 -1.83990583e-01 -1.64475095e+00 6.05358064e-01 1.75033510e+00 4.65838462e-01 8.22201222e-02 4.79684651e-01 -5.56578517e-01 6.36829436e-01 -1.17279899e+00 -6.27137601e-01 -2.44271621e-01 -6.99598908e-01 -6.74028635e-01 1.59280702e-01 4.56852645e-01 9.97701436e-02 6.31628871e-01 1.08719325e+00 4.92021739e-01 -1.44616887e-01 9.02151048e-01 -1.12933671e+00 -1.03475201e+00 9.85435843e-01 -7.75108337e-01 1.36283587e-03 6.05525911e-01 -2.17904791e-01 1.18885648e+00 -1.04256761e+00 7.64603794e-01 1.37255597e+00 4.64950204e-01 3.83085489e-01 -9.78127599e-01 -7.30923653e-01 8.40410471e-01 9.36036259e-02 -8.87689710e-01 -2.37024039e-01 7.15262115e-01 -7.19525635e-01 9.00918067e-01 2.76262254e-01 6.65928364e-01 1.03349030e+00 7.17366695e-01 4.68724519e-01 1.05783796e+00 -4.28178757e-01 7.00762093e-01 5.01271427e-01 4.75706398e-01 -3.71836271e-04 1.26512611e+00 -7.34001040e-01 -7.20407188e-01 -4.53617841e-01 -1.12809673e-01 -2.26697385e-01 -2.59654135e-01 1.62877753e-01 -1.20684695e+00 7.92131603e-01 -1.85551852e-01 1.87400401e-01 -5.53975999e-01 -4.07690674e-01 4.72687423e-01 5.16956449e-01 8.22484136e-01 9.89580572e-01 -4.57942456e-01 -8.09069052e-02 -1.09961903e+00 5.72773933e-01 1.18139660e+00 8.41429412e-01 5.59504688e-01 -8.21807534e-02 -2.89468378e-01 7.09171295e-01 3.61396939e-01 6.57074094e-01 2.79171556e-01 -1.01926243e+00 5.06131649e-01 9.08446550e-01 4.40533191e-01 -1.14155364e+00 -2.34047651e-01 -5.22795320e-01 -7.21195579e-01 9.93393064e-02 -3.23473394e-01 -7.89155722e-01 -7.49091268e-01 1.15629780e+00 -3.54700312e-02 -6.82321489e-01 1.81268051e-01 5.07200122e-01 1.63178813e+00 6.00221395e-01 1.00291021e-01 -8.12698483e-01 1.40234745e+00 -1.00343657e+00 -1.33764279e+00 -1.86305836e-01 5.30141294e-01 -8.88265669e-01 6.83518171e-01 9.64602113e-01 -1.32945299e+00 -6.29107475e-01 -1.29745924e+00 1.48384064e-01 -6.07122600e-01 4.13347632e-01 3.75124305e-01 5.31024158e-01 -1.18246758e+00 2.58221328e-01 -7.80875981e-02 -3.20881456e-01 3.41767788e-01 1.02418341e-01 -1.68650106e-01 1.12242557e-01 -1.12001741e+00 1.05295408e+00 8.16648975e-02 2.81430036e-01 -1.96654260e-01 -9.68164504e-01 -6.92889512e-01 -2.12899476e-01 6.47955894e-01 -1.13619995e+00 1.08964443e+00 -8.22437525e-01 -1.62844908e+00 4.77013677e-01 -5.07633090e-01 9.10629518e-03 1.94080949e-01 -3.92587036e-01 -3.34483892e-01 2.76173688e-02 4.22433168e-01 5.11749864e-01 6.46860898e-01 -1.43602359e+00 -7.49248981e-01 -2.29107261e-01 8.59622061e-02 2.32199639e-01 -1.12520255e-01 1.18727133e-01 -1.80238515e-01 -7.60067046e-01 -1.69585347e-01 -7.65960395e-01 -5.43486953e-01 -5.99534810e-01 -5.99637866e-01 -5.73419273e-01 4.09780711e-01 -4.25547630e-01 1.77018034e+00 -1.59701312e+00 2.55492508e-01 -1.11666575e-01 5.24674773e-01 2.46697485e-01 -1.97229818e-01 7.29069412e-01 -1.63916826e-01 7.83920288e-01 -2.52882093e-01 -2.48646989e-01 -3.74400839e-02 -2.43221983e-01 -6.17798269e-01 -3.44150467e-03 3.44376683e-01 9.36121404e-01 -1.08059442e+00 -3.81908596e-01 -1.37228565e-02 4.85222936e-02 -3.25835258e-01 -1.18958615e-01 -3.06599766e-01 2.28008911e-01 -5.60909867e-01 5.89312077e-01 7.35282600e-01 -3.30100715e-01 1.92643404e-01 -1.72699720e-01 -6.90418363e-01 6.32357717e-01 -1.04238451e+00 1.51545310e+00 -3.74914289e-01 6.84397280e-01 8.22611675e-02 -6.95293427e-01 1.14610803e+00 3.01233649e-01 2.94931531e-01 -3.28869671e-01 2.80857850e-02 3.48322272e-01 -1.39888900e-03 -4.22468513e-01 7.48333693e-01 6.43490776e-02 -1.03137247e-01 7.15973258e-01 -1.06918193e-01 -7.95047820e-01 7.57193804e-01 4.37708467e-01 1.00170124e+00 -2.12918282e-01 4.91270900e-01 -3.80004317e-01 6.39602602e-01 3.69736075e-01 5.19309819e-01 1.06156087e+00 1.57584503e-01 6.73700273e-01 9.58209515e-01 -2.63511360e-01 -5.91442883e-01 -6.20489538e-01 1.52420877e-02 5.73690116e-01 -2.02191770e-01 -1.10013354e+00 -6.25448704e-01 -7.66416967e-01 -2.36597359e-02 1.03381813e+00 -8.76252174e-01 3.25477608e-02 -4.08001430e-02 -7.36131132e-01 -9.60622430e-02 8.21218491e-02 1.03188217e-01 -1.14868605e+00 -4.96328831e-01 4.13806498e-01 -1.23612396e-01 -8.22317183e-01 -1.49790034e-01 1.17410302e-01 -7.12268353e-01 -1.08393776e+00 -1.06756163e+00 -4.05871689e-01 8.13819766e-01 3.18606317e-01 1.23915327e+00 -8.27899501e-02 2.60668635e-01 5.12065768e-01 -5.89015365e-01 -1.12507308e+00 -5.42984843e-01 4.01418179e-01 -3.87727953e-02 -4.29918468e-01 4.71180946e-01 -2.34670535e-01 -4.35061365e-01 9.29590464e-02 -7.62485564e-01 5.54035082e-02 8.43490660e-01 3.90232593e-01 3.72907817e-01 -1.69085354e-01 1.25312304e+00 -1.20803547e+00 1.57224524e+00 -4.71828997e-01 -2.54177213e-01 5.71634710e-01 -1.05403650e+00 -1.60114259e-01 4.56460238e-01 -3.75210345e-01 -1.10097289e+00 -5.68577409e-01 2.57811129e-01 3.32200795e-01 1.14368483e-01 1.19331634e+00 1.01622008e-02 3.51581365e-01 7.21520066e-01 -1.20206088e-01 5.53397238e-02 -2.09703371e-01 5.33970714e-01 7.88673401e-01 4.43234071e-02 -3.67945790e-01 5.59365988e-01 1.77754596e-01 -5.22465825e-01 -8.07723582e-01 -1.00359571e+00 -1.85447872e-01 -3.54898214e-01 -4.00266141e-01 6.77271008e-01 -8.98127735e-01 -4.39752340e-01 -3.18363495e-02 -1.68918204e+00 1.06762618e-01 -4.38439429e-01 3.34659815e-01 4.51199599e-02 4.45121139e-01 -1.05171032e-01 -9.48056102e-01 -7.91983724e-01 -1.15042412e+00 8.82589638e-01 6.47613168e-01 -9.58180547e-01 -8.19079399e-01 2.69422144e-01 2.21996054e-01 5.20773292e-01 1.76883012e-01 9.69953775e-01 -6.34870827e-01 5.01326434e-02 -9.51654837e-02 -1.48571387e-01 4.03155655e-01 5.42554557e-01 6.15432620e-01 -4.60045367e-01 8.72811824e-02 7.02914298e-02 -1.33827195e-01 8.70593548e-01 6.93296909e-01 8.30788314e-01 -4.19896096e-01 -3.11694682e-01 -3.28144938e-01 7.40056574e-01 2.71151155e-01 3.56607199e-01 3.63468885e-01 4.49929208e-01 1.21364403e+00 7.47489929e-01 5.27020812e-01 6.39602959e-01 9.04274732e-02 -2.56045610e-02 -1.22774737e-02 1.47734374e-01 1.18978042e-02 4.24983114e-01 1.45832276e+00 -4.25161310e-02 -5.14807343e-01 -8.06894541e-01 7.41053462e-01 -2.05272007e+00 -6.00327373e-01 -6.14426374e-01 1.78700483e+00 8.15677702e-01 4.10316885e-01 -1.25181571e-01 -2.35023368e-02 6.72737420e-01 3.59333247e-01 -4.73475933e-01 -1.05009782e+00 -3.15795183e-01 8.44731033e-02 1.52037129e-01 3.63625854e-01 -5.27040601e-01 5.46433151e-01 7.03938723e+00 4.56888318e-01 -8.21304619e-01 -2.06246033e-01 6.24934196e-01 5.07985540e-02 -1.16772461e+00 2.20512718e-01 -8.09200168e-01 4.33877766e-01 8.62782001e-01 -8.84240329e-01 -4.25378919e-01 5.87482631e-01 5.72399259e-01 -4.07874137e-01 -9.57103491e-01 6.98565066e-01 3.68661344e-01 -1.48824239e+00 6.21964633e-01 -2.91252322e-02 1.24232805e+00 -4.87221003e-01 2.19276827e-02 3.03261131e-01 3.91034514e-01 -9.46889102e-01 5.69199383e-01 5.22309363e-01 5.28866768e-01 -5.39281607e-01 9.53391969e-01 2.34641701e-01 -7.69979537e-01 2.13754654e-01 -4.61397856e-01 -1.78912714e-01 3.81301463e-01 1.01826560e+00 -5.12198687e-01 1.08774424e+00 4.31007892e-01 1.15334785e+00 -7.18394578e-01 8.28401625e-01 -6.06620371e-01 5.54084718e-01 2.94548571e-01 -4.63318795e-01 3.64785232e-02 -4.39278185e-01 4.57731426e-01 1.26215553e+00 2.99831599e-01 5.51067032e-02 -1.10283256e-01 9.30146575e-01 -2.99057346e-02 4.00126189e-01 -5.56362033e-01 -2.64601231e-01 2.27251083e-01 1.24874616e+00 -6.20757997e-01 -5.50070167e-01 -3.42941791e-01 3.90880913e-01 -1.78889081e-01 5.50132096e-01 -2.36459970e-01 -5.79707086e-01 2.70790398e-01 -5.14695868e-02 2.63160646e-01 1.60399079e-01 -8.83778632e-01 -1.03365386e+00 1.23114251e-01 -1.31396878e+00 1.01216853e-01 -9.35827017e-01 -1.16883707e+00 6.37435079e-01 5.81379980e-02 -1.20493388e+00 -1.56057447e-01 -4.58206028e-01 -8.14817190e-01 8.24073374e-01 -1.44674277e+00 -5.01760006e-01 -2.13277489e-01 -3.91822159e-01 9.10118818e-01 3.34869996e-02 4.55447167e-01 -1.42380834e-01 -6.70448363e-01 1.62343979e-01 -2.70035893e-01 -7.01342881e-01 1.09447634e+00 -1.31580889e+00 3.66132587e-01 5.42177141e-01 -3.99123907e-01 1.05323625e+00 8.84634197e-01 -1.18367624e+00 -1.28577709e+00 -7.89177716e-01 1.39042830e+00 -7.09947288e-01 7.92596757e-01 -9.33513045e-02 -1.01661432e+00 -4.46273200e-02 7.99884856e-01 -7.92259812e-01 9.79465425e-01 2.51110315e-01 -8.23261812e-02 -3.27255726e-02 -6.01485848e-01 7.64228582e-01 6.40151381e-01 -1.06128111e-01 -1.07860518e+00 2.63874888e-01 7.89035082e-01 -2.16650262e-01 -6.96094453e-01 5.41934371e-01 5.36231756e-01 -4.85765994e-01 3.71497750e-01 -4.34597313e-01 9.47198689e-01 -4.84803289e-01 5.10251403e-01 -1.52459347e+00 -1.86512113e-01 -7.76133120e-01 3.66542414e-02 1.31627035e+00 1.10347712e+00 -4.32893544e-01 3.70160162e-01 6.39521778e-01 -1.91370726e-01 -7.99424291e-01 -5.21441638e-01 -1.78138062e-01 1.85833052e-01 -2.81106323e-01 3.44570994e-01 7.87819505e-01 5.05250037e-01 7.39818335e-01 -1.07562445e-01 -3.64591360e-01 3.06716204e-01 1.31106317e-01 8.87053013e-01 -1.57213950e+00 2.90042758e-01 -7.69441009e-01 3.49695027e-01 -9.27699983e-01 1.69516057e-01 -3.77178878e-01 8.80764984e-03 -2.46470094e+00 3.19076508e-01 1.99835688e-01 -2.56847799e-01 1.17696889e-01 -5.66226304e-01 -3.03191334e-01 -1.25566646e-01 2.05195546e-01 -1.01206839e+00 4.85328168e-01 1.28638852e+00 -2.52587408e-01 -4.57700700e-01 1.78571083e-02 -1.76677978e+00 6.57400906e-01 5.37035465e-01 -3.94000471e-01 -5.08208394e-01 -1.47858605e-01 1.09653831e+00 -1.95199907e-01 -2.05598161e-01 -4.06996012e-01 4.76046503e-01 -3.04470628e-01 -3.40444148e-02 -1.07253063e+00 -3.96248370e-01 -3.23537707e-01 1.60041556e-01 2.28397131e-01 -5.46276391e-01 4.42817271e-01 2.64695644e-01 5.64329445e-01 -3.66691172e-01 -3.89880776e-01 -6.18574023e-02 -2.37815276e-01 5.57815246e-02 -2.26368263e-01 -8.01264644e-01 6.43772259e-02 6.35772228e-01 -2.88105369e-01 -6.96720362e-01 -5.22854924e-01 -2.17071563e-01 6.19438052e-01 3.72288078e-01 7.35663593e-01 5.26790857e-01 -8.03195298e-01 -1.16097939e+00 -4.46241409e-01 2.17685595e-01 1.70226619e-01 3.13296109e-01 8.28615606e-01 -1.24534264e-01 7.98582673e-01 2.14479849e-01 -3.89626831e-01 -8.72819424e-01 3.87525171e-01 -2.04864427e-01 -3.68739367e-01 -2.56916463e-01 4.02127296e-01 2.67173827e-01 -2.53072441e-01 2.32379213e-01 -6.93785310e-01 -9.45799649e-01 8.21334541e-01 8.97546172e-01 4.26104844e-01 3.04827362e-01 -2.58393347e-01 -5.05249321e-01 7.08717346e-01 -5.39814591e-01 -3.32852483e-01 1.34955204e+00 -2.04974189e-01 -4.40131575e-01 8.52494955e-01 5.74764252e-01 4.51349437e-01 -6.64379597e-01 8.56769457e-02 1.58015594e-01 1.74991578e-01 -7.67435879e-02 -9.85968113e-01 -5.59765637e-01 4.96753126e-01 -2.71774232e-01 5.09409249e-01 8.08341563e-01 -6.40817434e-02 1.07205495e-01 3.12602192e-01 -1.89438447e-01 -1.38695884e+00 3.79545405e-03 3.76682490e-01 1.39837384e+00 -1.21215904e+00 7.34280705e-01 -4.81430113e-01 -9.29122865e-01 1.19974792e+00 5.66104889e-01 1.34164661e-01 5.14281392e-01 -3.37165669e-02 3.06849271e-01 -4.35911804e-01 -1.35452759e+00 2.26649448e-01 6.49210632e-01 4.24256891e-01 7.54403293e-01 -2.49170318e-01 -1.41250443e+00 1.14709556e+00 -1.88528389e-01 1.03129223e-01 9.80989158e-01 9.45691884e-01 -3.45950156e-01 -1.00658011e+00 -3.85880679e-01 7.03609467e-01 -5.04702389e-01 -3.29065174e-02 -1.12284410e+00 4.48161155e-01 -2.14887798e-01 1.60595500e+00 -3.14684749e-01 -1.27708152e-01 6.75190210e-01 -2.36483157e-01 -1.11963101e-01 -1.03053904e+00 -1.02081454e+00 1.20737895e-01 2.89010346e-01 1.81384217e-02 -7.68624663e-01 -5.37667572e-01 -6.25406384e-01 -6.67033270e-02 -5.28081298e-01 7.85247505e-01 8.69302690e-01 1.03702259e+00 8.29695225e-01 9.06729341e-01 4.72555667e-01 -5.44824302e-01 -3.09316695e-01 -1.14661229e+00 -1.64136469e-01 -8.51325244e-02 2.38507152e-01 -5.50171673e-01 -7.00543523e-01 -1.99006405e-02]
[12.345035552978516, 9.56851577758789]
b49bc6f3-cf63-4d35-aae6-fc799d0f8bec
deep-markov-spatio-temporal-factorization
2003.09779
null
https://arxiv.org/abs/2003.09779v2
https://arxiv.org/pdf/2003.09779v2.pdf
Deep Markov Spatio-Temporal Factorization
We introduce deep Markov spatio-temporal factorization (DMSTF), a generative model for dynamical analysis of spatio-temporal data. Like other factor analysis methods, DMSTF approximates high dimensional data by a product between time dependent weights and spatially dependent factors. These weights and factors are in turn represented in terms of lower dimensional latents inferred using stochastic variational inference. The innovation in DMSTF is that we parameterize weights in terms of a deep Markovian prior extendable with a discrete latent, which is able to characterize nonlinear multimodal temporal dynamics, and perform multidimensional time series forecasting. DMSTF learns a low dimensional spatial latent to generatively parameterize spatial factors or their functional forms in order to accommodate high spatial dimensionality. We parameterize the corresponding variational distribution using a bidirectional recurrent network in the low-level latent representations. This results in a flexible family of hierarchical deep generative factor analysis models that can be extended to perform time series clustering or perform factor analysis in the presence of a control signal. Our experiments, which include simulated and real-world data, demonstrate that DMSTF outperforms related methodologies in terms of predictive performance for unseen data, reveals meaningful clusters in the data, and performs forecasting in a variety of domains with potentially nonlinear temporal transitions.
['J. Benjamin Hutchinson', 'Jan-Willem van de Meent', 'Eli Zachary Sennesh', 'Amirreza Farnoosh', 'Jennifer Dy', 'Behnaz Rezaei', 'Ajay Satpute', 'Zulqarnain Khan', 'Sarah Ostadabbas']
2020-03-22
null
null
null
null
['time-series-clustering']
['time-series']
[-3.07550937e-01 -2.12260067e-01 -8.85684192e-02 -1.02433950e-01 -4.15317386e-01 -7.60872841e-01 1.10371089e+00 -5.78939080e-01 7.39564300e-02 3.63681197e-01 8.09331954e-01 -2.34311000e-01 -5.36495805e-01 -6.74037695e-01 -5.77634096e-01 -1.15719759e+00 -5.12285709e-01 8.35964739e-01 1.45688325e-01 -5.32572297e-03 -1.70082182e-01 6.28119290e-01 -1.20494866e+00 3.16319704e-01 3.58339787e-01 6.31593287e-01 1.33938998e-01 7.26215661e-01 1.04839928e-01 5.14243901e-01 -2.15335369e-01 4.59742434e-02 5.79483174e-02 -2.26910800e-01 -4.19353217e-01 2.99822479e-01 -2.82209009e-01 -2.82857299e-01 -6.04427576e-01 3.68179262e-01 9.32256728e-02 4.11201119e-01 1.06770504e+00 -1.38133013e+00 -8.55823398e-01 5.46964943e-01 -3.47126037e-01 4.81367409e-01 -9.14037898e-02 1.41957536e-01 7.30896115e-01 -7.10916817e-01 6.11835063e-01 1.65396655e+00 6.97775960e-01 2.57669985e-01 -1.86203945e+00 -2.83727944e-01 4.37981665e-01 -1.63897425e-01 -1.25243533e+00 -3.64244998e-01 7.68355012e-01 -9.04370964e-01 1.10305798e+00 -1.23279847e-01 7.52283037e-01 1.61219740e+00 6.01941824e-01 6.18562162e-01 7.54030168e-01 -2.71748882e-02 3.42479020e-01 -5.37264407e-01 -6.28200620e-02 4.89851922e-01 -4.20086235e-01 1.35486543e-01 -4.41197544e-01 -6.14062250e-01 1.13081229e+00 6.17270589e-01 -6.77777827e-02 -5.63526928e-01 -1.50376189e+00 1.00434303e+00 2.47750878e-01 4.40395176e-01 -6.38458848e-01 3.94099981e-01 1.32506236e-01 2.33029172e-01 7.88693666e-01 1.08392596e-01 -2.98737884e-01 -1.23879865e-01 -1.18213010e+00 2.14144304e-01 4.89726424e-01 3.99827898e-01 7.24854529e-01 3.55483502e-01 -3.14872026e-01 4.58509833e-01 6.46508455e-01 7.78712273e-01 4.93808836e-01 -1.21020770e+00 2.21228167e-01 3.50315422e-01 3.05310339e-01 -9.68358994e-01 -5.31394541e-01 -4.66574550e-01 -1.17627585e+00 -2.46800631e-01 1.78063825e-01 -1.78413726e-02 -1.17095172e+00 2.12534308e+00 2.06164375e-01 6.40538812e-01 -5.57344221e-02 8.28782558e-01 -2.99306121e-02 1.18596601e+00 -1.18781395e-01 -4.70058918e-01 1.14534235e+00 -5.21431983e-01 -9.27189231e-01 1.43396631e-01 1.24821775e-01 -4.49118853e-01 8.50751102e-01 2.59323478e-01 -9.46597695e-01 -5.45161843e-01 -4.41533506e-01 8.16086233e-02 -3.09441864e-01 3.27281021e-02 6.66363060e-01 1.40704140e-01 -1.35831165e+00 5.12601972e-01 -1.69508398e+00 -2.36197874e-01 -1.42745546e-03 2.11621046e-01 -1.84723914e-01 1.90350085e-01 -1.34908259e+00 3.37398022e-01 7.04861954e-02 2.92030960e-01 -1.40801251e+00 -7.75039852e-01 -7.23250628e-01 1.17456056e-01 -7.87774920e-02 -1.11498749e+00 7.89535403e-01 -3.64496082e-01 -1.41166461e+00 2.46659532e-01 -6.14265859e-01 -4.21487778e-01 3.54383826e-01 5.74959926e-02 -6.03346705e-01 2.41523370e-01 1.66786849e-01 4.45034742e-01 1.42943418e+00 -1.01926255e+00 -1.10874534e-01 -3.47966760e-01 -2.79462129e-01 -6.30464703e-02 -3.08715135e-01 -3.59666437e-01 -4.71078366e-01 -9.09412563e-01 2.48622194e-01 -1.21470261e+00 -4.94822741e-01 -3.71363223e-01 -1.97251886e-01 -2.03394845e-01 1.00222838e+00 -6.52863145e-01 1.50646698e+00 -2.15162683e+00 9.60444748e-01 3.30420643e-01 5.08913755e-01 -2.61464059e-01 -4.58841547e-02 9.37177062e-01 -7.69926980e-02 8.80322009e-02 -2.27723613e-01 -6.88987494e-01 2.82252967e-01 5.46831191e-01 -9.25582767e-01 4.67856437e-01 1.03688136e-01 1.01399827e+00 -8.92319500e-01 1.99970640e-02 4.01760012e-01 9.74884093e-01 -5.53340375e-01 1.29774645e-01 -3.02476346e-01 6.63509429e-01 -5.69270432e-01 9.87003520e-02 3.28561932e-01 -5.80205202e-01 2.25529969e-01 -1.13898128e-01 -1.35103360e-01 2.56792232e-02 -1.07863450e+00 1.77508664e+00 -3.97381067e-01 5.03789902e-01 -1.12381816e-01 -8.24056149e-01 5.74668050e-01 5.12404084e-01 8.28014910e-01 -1.99071050e-01 -1.51684046e-01 -2.36329108e-01 -4.08315688e-01 -3.71615499e-01 4.42028284e-01 -2.23009989e-01 -2.71497607e-01 7.38117695e-01 3.39260638e-01 3.07690203e-01 7.05790371e-02 4.47721303e-01 1.06128645e+00 1.83036670e-01 -3.41071188e-01 -4.20079440e-01 4.65156361e-02 -3.51428926e-01 4.19049770e-01 4.22007233e-01 1.66807428e-01 3.48722637e-01 6.73445225e-01 -6.62331283e-01 -1.29496109e+00 -1.50583446e+00 -1.03133969e-01 1.02370572e+00 -3.92793149e-01 -5.54189324e-01 -3.84921610e-01 7.19901826e-03 3.52629460e-02 4.75310415e-01 -1.08527303e+00 -1.44675598e-01 -5.25591433e-01 -1.12733710e+00 3.44855011e-01 5.83460927e-01 -1.26854733e-01 -6.02053583e-01 -3.57461721e-01 3.66279304e-01 -2.93819964e-01 -9.91124034e-01 -2.80503124e-01 3.64318162e-01 -9.38175797e-01 -5.01711488e-01 -8.23700547e-01 -3.24878156e-01 5.03770888e-01 1.35062069e-01 8.80167484e-01 -6.29691422e-01 -3.39144245e-02 6.18915379e-01 -4.05873880e-02 3.92809182e-01 -4.03774559e-01 7.74408430e-02 4.93171006e-01 4.15705025e-01 1.14442535e-01 -1.19400954e+00 -7.34603465e-01 4.07960355e-01 -1.34220648e+00 4.12642583e-02 1.08483538e-01 7.72791862e-01 5.93609750e-01 7.86207467e-02 2.80218065e-01 -3.69309217e-01 7.14891016e-01 -9.47459161e-01 -5.98673761e-01 1.27146602e-01 -2.56016314e-01 5.55946887e-01 5.20422935e-01 -7.45939910e-01 -1.00482428e+00 -1.39564618e-01 1.92413360e-01 -1.20521998e+00 -7.75757208e-02 5.58972180e-01 2.10832387e-01 6.95350826e-01 4.08824652e-01 4.13322359e-01 2.19432339e-02 -5.91817737e-01 6.78974986e-01 1.33399695e-01 3.01656127e-01 -6.04843676e-01 9.06064570e-01 8.83754134e-01 2.04561025e-01 -7.19882727e-01 -5.67061067e-01 -2.65882671e-01 -9.30174291e-01 -2.62135625e-01 1.15145016e+00 -1.09455681e+00 -7.12200999e-01 3.96243989e-01 -8.94740880e-01 -6.82145238e-01 -1.92119077e-01 5.53009093e-01 -6.64638638e-01 1.29017889e-01 -9.89408195e-01 -8.60496640e-01 3.15032393e-01 -9.98936236e-01 1.56820071e+00 -3.32523614e-01 -3.29823971e-01 -1.55431604e+00 5.66155434e-01 -1.34914905e-01 3.70590985e-01 3.84391189e-01 9.87265050e-01 2.26502717e-02 -5.23551583e-01 7.37822652e-02 3.76686931e-01 -9.42225233e-02 1.70127720e-01 2.90465057e-01 -7.27030873e-01 -4.75771278e-01 4.21618447e-02 1.46549240e-01 1.24210870e+00 8.85667503e-01 6.91414058e-01 -3.53413761e-01 -5.75064838e-01 7.79879034e-01 1.02741992e+00 1.49788931e-01 3.04167718e-01 -2.25576490e-01 8.54334891e-01 4.16416466e-01 -1.41717553e-01 6.41504705e-01 5.69688261e-01 5.26316643e-01 2.65022814e-01 1.60473809e-01 2.92676747e-01 -6.02226377e-01 6.24477983e-01 1.07522774e+00 -2.88229883e-01 -4.98507172e-01 -1.04651690e+00 6.38043463e-01 -2.21397448e+00 -1.32793558e+00 1.31903347e-02 1.70414042e+00 2.65489280e-01 -1.41155541e-01 5.15132785e-01 -5.96998893e-02 6.29427791e-01 4.08020824e-01 -7.35822260e-01 -9.19775292e-03 -1.51792362e-01 -3.16582829e-01 1.10463433e-01 6.87181950e-01 -1.03149462e+00 8.95787299e-01 7.08295012e+00 5.57612479e-01 -1.11905038e+00 2.54176527e-01 3.98414940e-01 -2.90716380e-01 -7.81693697e-01 -9.90522131e-02 -6.16854012e-01 4.82257605e-01 1.37142980e+00 1.29278734e-01 7.30994940e-01 2.05964133e-01 5.63043714e-01 4.42089438e-01 -8.40805709e-01 6.30397022e-01 -3.61447364e-01 -1.47516465e+00 3.92501831e-01 4.29670066e-01 7.52523243e-01 2.09885567e-01 5.09618938e-01 1.39784753e-01 7.38699973e-01 -9.33408439e-01 7.18583226e-01 1.16824985e+00 3.38827223e-01 -6.66606367e-01 3.61134261e-02 6.25006318e-01 -1.26083708e+00 -1.66815698e-01 -3.33914191e-01 -6.55403584e-02 6.33473158e-01 5.46676278e-01 -4.64406639e-01 3.16936851e-01 7.65697896e-01 1.14134073e+00 -1.63223550e-01 3.11335802e-01 8.96095261e-02 8.14947903e-01 -5.05757689e-01 4.65858728e-01 4.55878615e-01 -4.94533658e-01 6.20722950e-01 9.58993793e-01 5.41693211e-01 3.50554623e-02 2.78078258e-01 1.04269195e+00 3.94584417e-01 -5.56270957e-01 -4.94677007e-01 -4.84038562e-01 2.46038094e-01 1.01927221e+00 -1.00351667e+00 -1.54176518e-01 -2.91958526e-02 1.06290448e+00 2.05150589e-01 9.72649455e-01 -9.29213047e-01 5.98171949e-01 8.97447050e-01 2.99116611e-01 6.72268748e-01 -9.26067650e-01 4.32260364e-01 -1.62475324e+00 -1.88979194e-01 -3.27286243e-01 4.30310011e-01 -8.69336367e-01 -1.46884120e+00 9.03206050e-01 3.52828175e-01 -1.24482238e+00 -8.79096031e-01 -3.84870589e-01 -4.20171618e-01 9.71872926e-01 -9.60341990e-01 -1.38246882e+00 2.71565169e-01 1.09608889e+00 3.14823836e-01 -1.43401414e-01 8.08068156e-01 1.61913455e-01 -5.44471383e-01 -1.07622579e-01 4.20024991e-01 -2.18220115e-01 3.16245675e-01 -1.19415307e+00 7.28726268e-01 8.55054617e-01 2.83530533e-01 9.28678691e-01 8.75904679e-01 -7.32076406e-01 -1.52865231e+00 -1.10579050e+00 6.14032388e-01 -6.96752667e-01 1.17686856e+00 -8.08219850e-01 -9.21252310e-01 1.01234472e+00 1.61534891e-01 1.29335672e-01 8.12253177e-01 1.38966352e-01 -3.72754633e-01 2.57977277e-01 -5.96372068e-01 4.57827806e-01 9.87938225e-01 -8.73663366e-01 -3.81428868e-01 2.84723222e-01 9.25046682e-01 -1.36459051e-02 -1.18507993e+00 1.65147454e-01 5.90073705e-01 -8.30588400e-01 1.02904069e+00 -8.51565778e-01 1.54591203e-01 -5.37764072e-01 -2.67279476e-01 -1.49118745e+00 -9.60417807e-01 -9.52738583e-01 -4.91196334e-01 9.84982312e-01 3.14211726e-01 -4.02871817e-01 4.03394312e-01 3.18634927e-01 -1.29412308e-01 -5.19306183e-01 -9.60839748e-01 -6.18907869e-01 1.19341865e-01 -7.23368406e-01 6.17174268e-01 1.08677292e+00 -3.06800187e-01 2.73206174e-01 -7.96813428e-01 4.33682829e-01 5.24086773e-01 3.37548852e-01 4.21261072e-01 -1.17318153e+00 -5.22262812e-01 -2.98455834e-01 -2.82820314e-01 -1.16031647e+00 3.90748203e-01 -8.03012967e-01 -3.00161690e-01 -1.41343141e+00 -3.68461828e-03 -1.60294529e-02 -3.45105618e-01 3.00249517e-01 2.34175399e-01 9.86304209e-02 5.23200184e-02 6.53786778e-01 -3.89559925e-01 9.62864459e-01 1.08145380e+00 7.20201209e-02 -2.80262291e-01 1.02362856e-01 -1.42995179e-01 5.50271451e-01 4.22576368e-01 -3.08666289e-01 -7.28534043e-01 -4.53265101e-01 5.90673566e-01 5.99234939e-01 5.65894425e-01 -6.19321287e-01 5.79116307e-02 -2.78633237e-01 5.54286182e-01 -7.06941068e-01 5.51526904e-01 -7.13601291e-01 7.12279260e-01 1.76628083e-01 -2.91920781e-01 4.95458931e-01 6.04603812e-02 9.74141240e-01 -2.09468797e-01 7.95257032e-01 3.11581880e-01 3.56318541e-02 -3.12509000e-01 5.44357002e-01 -1.03654647e+00 -3.50921243e-01 6.84258461e-01 -5.84273487e-02 -5.05362414e-02 -6.89163089e-01 -1.46228921e+00 2.07836837e-01 2.75582463e-01 6.23315096e-01 4.52218771e-01 -1.68959594e+00 -4.00703758e-01 4.21739072e-01 -1.69637188e-01 -3.85727078e-01 6.79378390e-01 9.08478796e-01 6.43039346e-02 4.80981469e-01 -1.79909825e-01 -1.02808404e+00 -3.94545019e-01 8.68363976e-01 2.76102364e-01 -5.05773842e-01 -7.15727985e-01 3.69281083e-01 4.46750730e-01 -2.99608350e-01 -1.36131540e-01 -8.19900632e-01 -2.18267683e-02 4.87717986e-01 2.95463175e-01 2.44670704e-01 -4.00058061e-01 -8.35339248e-01 -1.43759966e-01 5.23265839e-01 5.01352191e-01 -4.89586681e-01 1.65168941e+00 -4.95024383e-01 -2.09386051e-01 1.07965064e+00 1.18612146e+00 -2.86333740e-01 -1.76965356e+00 -2.98118323e-01 -1.73456535e-01 -4.10836451e-02 1.66678503e-01 -3.25805426e-01 -9.95189011e-01 1.00230491e+00 3.77775133e-01 6.45741820e-01 1.08026910e+00 1.87358722e-01 6.64414048e-01 1.10716410e-01 9.92194191e-02 -6.29125178e-01 1.00512147e-01 6.42187655e-01 8.17740142e-01 -7.05420375e-01 -4.02767241e-01 1.86241180e-01 -5.10187507e-01 1.09863234e+00 -1.10123664e-01 -4.10739094e-01 1.18676043e+00 4.88326818e-01 -2.30926424e-01 -2.94269204e-01 -1.20836663e+00 1.65562816e-02 5.93779802e-01 3.70703518e-01 1.20452151e-01 2.50269800e-01 3.52103502e-01 6.41548634e-01 -1.78622350e-01 -3.20633382e-01 1.31530046e-01 5.16961277e-01 -1.41246781e-01 -9.25278127e-01 -3.71411473e-01 1.16204157e-01 -1.75450891e-01 7.79121667e-02 3.71599644e-02 5.65270185e-01 -2.25296747e-02 6.04774594e-01 5.44681728e-01 -3.93040895e-01 -1.74208328e-01 4.23121959e-01 1.50453225e-01 -2.27089673e-01 -2.02075660e-01 8.68224740e-01 -4.69486624e-01 -7.65323222e-01 -6.33986235e-01 -9.45822418e-01 -9.02357817e-01 -3.21599931e-01 2.40264356e-01 2.57036150e-01 3.53753656e-01 1.08620322e+00 6.47172272e-01 6.97311103e-01 4.79714453e-01 -1.19100296e+00 -2.22022850e-02 -9.19384778e-01 -6.15715981e-01 2.26762593e-01 7.71722615e-01 -7.92668998e-01 -5.29195070e-01 3.38158965e-01]
[7.020599842071533, 3.444119930267334]
65f20a09-8fbf-41bc-81ab-94bcf9dc19ca
going-beyond-research-datasets-novel-intent
2305.05474
null
https://arxiv.org/abs/2305.05474v1
https://arxiv.org/pdf/2305.05474v1.pdf
Going beyond research datasets: Novel intent discovery in the industry setting
Novel intent discovery automates the process of grouping similar messages (questions) to identify previously unknown intents. However, current research focuses on publicly available datasets which have only the question field and significantly differ from real-life datasets. This paper proposes methods to improve the intent discovery pipeline deployed in a large e-commerce platform. We show the benefit of pre-training language models on in-domain data: both self-supervised and with weak supervision. We also devise the best method to utilize the conversational structure (i.e., question and answer) of real-life datasets during fine-tuning for clustering tasks, which we call Conv. All our methods combined to fully utilize real-life datasets give up to 33pp performance boost over state-of-the-art Constrained Deep Adaptive Clustering (CDAC) model for question only. By comparison CDAC model for the question data only gives only up to 13pp performance boost over the naive baseline.
['Piotr Rybak', 'Robert Mroczkowski', 'Dariusz Kajtoch', 'Tsimur Hadeliya', 'Aleksandra Chrabrowa']
2023-05-09
null
null
null
null
['intent-discovery']
['natural-language-processing']
[-7.80233666e-02 2.17483714e-02 4.82829101e-02 -9.23935890e-01 -1.02636731e+00 -7.21992075e-01 8.07571590e-01 4.63635437e-02 -4.10241365e-01 1.51607603e-01 7.29662180e-01 -2.41997913e-01 -3.36414203e-02 -3.18854958e-01 -5.18192708e-01 -2.68614888e-01 -1.73635527e-01 9.39409256e-01 2.30310373e-02 -8.96220654e-02 3.38835955e-01 -2.62504607e-01 -1.29466510e+00 9.92759466e-01 6.20172262e-01 8.21367383e-01 1.76608115e-01 7.78426051e-01 -4.69827235e-01 1.14631546e+00 -3.96819353e-01 -4.95881855e-01 2.21407309e-01 -1.87772885e-01 -1.27875459e+00 2.96701729e-01 6.79488361e-01 -4.70171779e-01 -1.17583871e-01 4.05324221e-01 4.41416323e-01 2.12085068e-01 1.90463856e-01 -1.34062994e+00 -6.87458158e-01 9.08511996e-01 -2.89589137e-01 1.90368965e-01 4.14118826e-01 2.51766723e-02 1.34429276e+00 -6.50818467e-01 7.27388144e-01 1.03036237e+00 8.91514242e-01 4.25822884e-01 -1.17545855e+00 -6.22982383e-01 4.54463691e-01 3.40876549e-01 -1.08045483e+00 -6.39533520e-01 8.41827452e-01 -2.32831106e-01 1.42111659e+00 2.99315929e-01 5.39934225e-02 1.23623645e+00 -1.39337897e-01 1.14298308e+00 8.21461678e-01 -1.48548245e-01 3.79197419e-01 3.18392605e-01 8.59067619e-01 5.60879290e-01 -2.36521214e-01 -6.09698057e-01 -5.27876854e-01 -4.05840814e-01 -1.39291435e-01 4.35674563e-03 9.13618878e-02 -5.30694872e-02 -1.08681655e+00 1.13294268e+00 1.14968024e-01 4.90250170e-01 -4.50302452e-01 -3.99265550e-02 4.37255353e-01 4.07633424e-01 6.49176776e-01 7.21258879e-01 -1.06037247e+00 -2.92642981e-01 -8.86603892e-01 1.58112198e-01 1.56510282e+00 1.18959665e+00 9.57759559e-01 -6.31726682e-01 -2.27546766e-01 9.40805972e-01 3.71270835e-01 -5.41342199e-02 5.56452513e-01 -1.27280116e+00 5.60504138e-01 8.11855495e-01 -5.07976413e-02 -7.70171165e-01 -7.92839289e-01 -4.90774542e-01 -6.27796173e-01 -6.64549232e-01 3.68129492e-01 -5.15173554e-01 -6.28088117e-01 1.64598334e+00 4.30663586e-01 2.78077900e-01 -2.51932051e-02 6.25076413e-01 9.91349518e-01 6.59583390e-01 -4.72470187e-03 -8.17436799e-02 1.49049985e+00 -1.44038093e+00 -5.72719038e-01 -3.72295111e-01 8.50578785e-01 -9.37472463e-01 1.26284802e+00 2.97613472e-01 -5.67287564e-01 -5.62090278e-01 -6.54302955e-01 -3.13767910e-01 -3.12605113e-01 5.49302362e-02 1.04374897e+00 6.62121773e-01 -1.25595534e+00 4.51814607e-02 -7.92638242e-01 -7.13399529e-01 1.51266009e-01 5.08744359e-01 -1.55706599e-01 -4.47148867e-02 -9.88989770e-01 3.18334013e-01 4.68868539e-02 -4.00173932e-01 -6.70343161e-01 -1.04115486e+00 -6.39981329e-01 2.25565672e-01 6.09251976e-01 -6.66598082e-01 1.58827591e+00 -8.78844261e-01 -1.45577848e+00 8.89881670e-01 -3.86609495e-01 -8.06034982e-01 7.97636509e-02 -4.17471498e-01 -3.23494941e-01 1.99407652e-01 2.50831366e-01 9.50372934e-01 4.73770648e-01 -1.23188937e+00 -7.19553828e-01 -3.22164446e-01 3.57228398e-01 8.43883306e-02 -7.02545702e-01 2.03238726e-01 -7.53403485e-01 -2.95722067e-01 -2.12545648e-01 -1.17210424e+00 -2.13115707e-01 -4.50024396e-01 -2.68265575e-01 -6.08262002e-01 1.11603606e+00 -7.12382853e-01 1.12514663e+00 -2.01207638e+00 -5.81371844e-01 -2.18972042e-01 2.19146132e-01 1.46087840e-01 -3.57778251e-01 6.13750875e-01 8.53439569e-02 1.61919862e-01 7.42238574e-03 -8.32784534e-01 3.82491291e-01 2.91822642e-01 -2.79852003e-01 2.16298550e-01 1.69505123e-02 9.06632543e-01 -4.83968019e-01 -4.78371203e-01 6.33492991e-02 2.61162013e-01 -1.09197044e+00 3.90014917e-01 -7.34736800e-01 2.08543554e-01 -3.52295786e-01 1.83522984e-01 8.23827982e-01 -8.54227901e-01 6.40953839e-01 -2.39517748e-01 1.17248051e-01 9.43902731e-01 -8.42912674e-01 1.94384956e+00 -8.19419086e-01 6.56900465e-01 5.45458317e-01 -9.27796185e-01 6.07639492e-01 2.68644929e-01 6.82072043e-01 -5.67805290e-01 -1.55922294e-01 -2.83608407e-01 -1.27040800e-02 -6.89630866e-01 5.01553953e-01 1.85131937e-01 -1.56396329e-01 9.82602119e-01 2.59990364e-01 4.12023604e-01 1.52279198e-01 6.45445287e-01 1.42400467e+00 -3.64318758e-01 7.02990666e-02 -5.13915539e-01 4.98254597e-01 2.35626951e-01 4.16785061e-01 8.36113870e-01 -1.93831608e-01 4.64253813e-01 5.02197087e-01 -3.88187855e-01 -6.37066245e-01 -5.86282194e-01 8.91018435e-02 1.83117294e+00 5.95345955e-05 -6.38223767e-01 -5.08784175e-01 -1.28685129e+00 4.62708287e-02 9.94975150e-01 -5.00254214e-01 3.79794419e-01 -7.70541966e-01 -8.76835704e-01 4.46232319e-01 3.16302657e-01 8.24639857e-01 -7.35548198e-01 -3.08066234e-02 2.78810352e-01 -6.16383672e-01 -1.54115689e+00 -7.43303597e-01 1.90223932e-01 -5.72066247e-01 -9.79373753e-01 -9.47588235e-02 -9.04675841e-01 2.53931701e-01 4.60881710e-01 1.61353374e+00 1.46747410e-01 1.62275940e-01 7.01145113e-01 -4.88596350e-01 -3.99566591e-02 -3.58091801e-01 4.80704635e-01 -2.18798921e-01 1.74786434e-01 9.93750274e-01 -7.61410356e-01 -7.96426356e-01 4.27450657e-01 -6.05859041e-01 -1.51699498e-01 5.38587093e-01 5.54305434e-01 -1.05163731e-01 1.38823032e-01 7.62778699e-01 -1.29277563e+00 7.23679841e-01 -8.35061908e-01 -2.31941953e-01 1.77236736e-01 -8.16253960e-01 -9.85105056e-04 6.28794670e-01 -4.66964722e-01 -1.44466817e+00 8.13350528e-02 -5.15294671e-01 6.94188774e-02 -6.73572838e-01 3.99160087e-01 -1.46895096e-01 4.03606772e-01 5.53781331e-01 -9.78260413e-02 -1.73495978e-01 -7.86153913e-01 5.19578516e-01 1.08629525e+00 4.01984036e-01 -4.20159698e-01 2.33116508e-01 7.17031181e-01 -7.80325115e-01 -7.79900491e-01 -9.57134426e-01 -1.32595789e+00 -5.86018145e-01 1.88936993e-01 9.02031839e-01 -1.12394559e+00 -9.62009966e-01 2.80262321e-01 -1.14019465e+00 -3.98478359e-01 1.79854751e-01 3.44097733e-01 -2.33909473e-01 5.45880318e-01 -1.03777516e+00 -6.20519757e-01 -7.17247903e-01 -7.85466611e-01 1.05417752e+00 -8.69579539e-02 -5.20649910e-01 -1.11722207e+00 6.84356019e-02 1.17278004e+00 5.80724120e-01 -4.45684612e-01 9.53096628e-01 -1.60188043e+00 -6.47172451e-01 9.42354128e-02 -2.24387899e-01 2.04148397e-01 7.72930384e-02 -4.82927382e-01 -1.09969628e+00 -1.67784274e-01 3.79544526e-01 -3.44747335e-01 9.38580930e-01 1.93697400e-02 9.04736519e-01 -6.32352591e-01 -4.43402320e-01 3.25795293e-01 1.07649207e+00 -3.79176624e-02 2.53901601e-01 1.65836647e-01 4.71268922e-01 8.02499533e-01 2.64667422e-01 3.96737993e-01 1.14341557e+00 7.19316125e-01 8.93875360e-02 1.30012184e-02 -1.52909935e-01 -8.66707116e-02 3.10144484e-01 9.97360885e-01 8.31419349e-01 -3.29697877e-01 -1.01818156e+00 7.50980735e-01 -2.01457930e+00 -9.41149235e-01 -4.09665853e-01 1.57195055e+00 1.05028582e+00 8.27031881e-02 9.86643061e-02 -4.75738049e-01 6.32705510e-01 3.92263234e-02 -4.53671336e-01 -2.33867019e-01 3.46156955e-01 -3.44987512e-02 1.98320985e-01 6.02917552e-01 -1.35377800e+00 7.41426706e-01 6.07151556e+00 6.71242237e-01 -8.19574714e-01 4.73280668e-01 6.53324604e-01 -4.20308523e-02 -3.55853826e-01 1.17493659e-01 -9.26261008e-01 5.98216653e-01 1.31601512e+00 2.33889490e-01 3.89684141e-01 1.03251910e+00 6.08374886e-02 -1.01076186e-01 -1.45682514e+00 7.76476204e-01 4.13033485e-01 -1.45463216e+00 -2.81718940e-01 -5.16426153e-02 8.91062021e-01 4.29606557e-01 -1.63478062e-01 9.47064281e-01 6.93787396e-01 -4.70174193e-01 -4.43877988e-02 1.74005643e-01 1.35218665e-01 -4.66491938e-01 9.20875549e-01 7.84648538e-01 -8.14735651e-01 -2.29187921e-01 -4.69041988e-02 -1.11248389e-01 2.12306857e-01 6.62806571e-01 -1.28892529e+00 2.02081174e-01 8.07984412e-01 6.48981214e-01 -6.99144363e-01 6.31389499e-01 2.49764860e-01 1.25000143e+00 -5.50271630e-01 -8.09999853e-02 3.81650090e-01 -1.55772138e-02 2.05835581e-01 1.57505703e+00 -2.51586139e-01 -9.42260697e-02 3.50343227e-01 7.08858848e-01 -4.61407691e-01 -3.55186290e-03 -1.26826808e-01 -2.91478168e-02 3.19849074e-01 1.60465670e+00 -6.92584157e-01 -4.46262598e-01 -8.08747828e-01 1.06700289e+00 4.00364041e-01 1.23613209e-01 -8.06365728e-01 -6.65681064e-02 8.30795109e-01 1.40181139e-01 6.04778826e-01 -2.91271985e-01 -3.91156137e-01 -1.22777820e+00 -1.46659166e-01 -1.06226873e+00 7.01083779e-01 -3.47291648e-01 -1.86842132e+00 2.68307924e-01 -5.35762236e-02 -6.20604277e-01 -5.06515741e-01 -2.73989648e-01 -7.60456681e-01 4.40596759e-01 -1.29721546e+00 -1.31654572e+00 -3.15454632e-01 6.03723347e-01 9.51697588e-01 -7.95485973e-02 8.19011748e-01 5.88892817e-01 -3.56327057e-01 6.69420481e-01 2.28364184e-01 3.48130971e-01 1.03598177e+00 -1.21981728e+00 8.46141398e-01 4.80772674e-01 3.25576752e-01 1.03948057e+00 4.03182983e-01 -4.83038127e-01 -1.47697365e+00 -1.23510933e+00 1.20108795e+00 -7.96502948e-01 7.64140546e-01 -1.04526997e+00 -9.38000679e-01 9.37489390e-01 8.36363018e-01 -4.79883254e-01 1.13377285e+00 8.47823799e-01 -4.17721182e-01 -1.31777644e-01 -1.12123728e+00 3.89330298e-01 9.80293512e-01 -6.18670225e-01 -7.78472900e-01 7.31830537e-01 1.32134604e+00 -1.89547632e-02 -7.71519899e-01 1.84490129e-01 3.05156231e-01 -1.00022554e+00 7.90086865e-01 -6.57487690e-01 2.80080348e-01 1.37989903e-02 -1.62594348e-01 -7.49320924e-01 -4.33898270e-01 -8.00056517e-01 -2.00695753e-01 1.63914573e+00 6.35089219e-01 -4.43752497e-01 1.04487252e+00 8.45655918e-01 -1.13673732e-01 -4.84549224e-01 -6.16211653e-01 -3.40931773e-01 -2.57479784e-04 -7.15124846e-01 4.76146042e-01 1.28718746e+00 8.27555284e-02 1.11184812e+00 -6.66731298e-02 1.73535690e-01 4.74426985e-01 6.20945096e-01 9.97132003e-01 -1.01346922e+00 -4.92631465e-01 -1.21919364e-01 1.37655273e-01 -1.63257337e+00 4.82532501e-01 -9.31348979e-01 -2.13919766e-02 -1.59659159e+00 4.61155206e-01 -4.31652576e-01 4.99577783e-02 3.69012803e-01 -2.77619123e-01 -8.54277518e-03 1.73312843e-01 2.42369995e-01 -1.33748949e+00 3.22055876e-01 3.58000189e-01 -1.70304701e-01 -3.29822302e-01 7.18258172e-02 -1.18769908e+00 6.90548480e-01 9.58717763e-01 -4.88687575e-01 -5.89733899e-01 -5.11408091e-01 1.27683908e-01 -2.62856483e-01 3.90077196e-02 -8.60592663e-01 6.46830738e-01 2.60554671e-01 1.28072426e-01 -9.13937271e-01 2.31037408e-01 -9.11053777e-01 -2.19454736e-01 6.38832375e-02 -8.63100171e-01 -1.11732543e-01 1.15654059e-01 8.19951534e-01 -5.00366762e-02 -2.19922230e-01 3.20491761e-01 -2.19246849e-01 -8.05877209e-01 6.10531354e-03 -5.65346718e-01 5.18157065e-01 7.22540140e-01 4.63224500e-01 -5.60361862e-01 -9.52369452e-01 -5.40363252e-01 6.72330558e-01 6.59710765e-02 6.50960743e-01 1.23569511e-01 -9.01063263e-01 -7.69446790e-01 -5.36671355e-02 -2.77020652e-02 -2.81612247e-01 5.04223585e-01 8.20211411e-01 -1.46468744e-01 8.72725248e-01 4.19715017e-01 -6.88497186e-01 -1.39662969e+00 7.78730869e-01 2.05725562e-02 -7.48936534e-01 -3.18208158e-01 9.63689268e-01 1.48229063e-01 -1.20748639e+00 3.77000362e-01 -3.04962426e-01 -1.28571481e-01 5.75045571e-02 3.60609859e-01 3.35747004e-01 2.67270416e-01 -1.95097506e-01 -6.50393128e-01 -5.24710566e-02 -6.50210500e-01 -6.49427772e-02 1.35697675e+00 -4.79130059e-01 -1.69701561e-01 1.47154048e-01 1.75377691e+00 3.13407518e-02 -9.59904850e-01 -5.35801291e-01 2.81796932e-01 -1.02454454e-01 8.49235803e-02 -1.07615054e+00 -6.51190639e-01 5.67558348e-01 3.06202710e-01 4.45763737e-01 9.32948291e-01 3.64639074e-01 1.10155737e+00 9.08355534e-01 8.70703459e-02 -1.35720575e+00 3.90732139e-01 6.81411684e-01 8.10268462e-01 -1.63414800e+00 -7.55538717e-02 -3.98343980e-01 -9.79404867e-01 7.53743410e-01 7.43008316e-01 5.59092052e-02 1.02402449e+00 4.21533197e-01 3.02387893e-01 -2.79838771e-01 -1.40375388e+00 -3.46820951e-01 1.09113142e-01 5.02904415e-01 6.99453950e-01 -1.86685696e-01 -2.21930206e-01 9.32000279e-01 -2.22680554e-01 -1.51299283e-01 2.55765498e-01 7.95055926e-01 -1.98535994e-01 -9.93901432e-01 -2.25597508e-02 6.88331842e-01 -6.26890421e-01 -4.22767639e-01 -5.39178371e-01 5.54244578e-01 1.40848592e-01 1.82715690e+00 3.13428611e-01 -5.95308900e-01 -3.86167243e-02 3.00553501e-01 -3.02350312e-01 -7.85367370e-01 -1.16022706e+00 3.61669771e-02 6.14190042e-01 -7.26696849e-01 -5.23647726e-01 -6.89079702e-01 -1.30662501e+00 -4.53637183e-01 -3.07837337e-01 4.35463339e-01 6.67443573e-01 9.79702055e-01 1.08951485e+00 2.13345543e-01 6.81459546e-01 -3.00152451e-01 -6.68379962e-01 -1.20625913e+00 -1.07986629e-01 5.15558362e-01 2.42076278e-01 -4.05696519e-02 -5.57959199e-01 1.93184406e-01]
[12.393815994262695, 7.491898059844971]
5eba7a83-f3c5-4494-8e82-018c15dd075b
person-search-via-a-mask-guided-two-stream
1807.08107
null
http://arxiv.org/abs/1807.08107v1
http://arxiv.org/pdf/1807.08107v1.pdf
Person Search via A Mask-Guided Two-Stream CNN Model
In this work, we tackle the problem of person search, which is a challenging task consisted of pedestrian detection and person re-identification~(re-ID). Instead of sharing representations in a single joint model, we find that separating detector and re-ID feature extraction yields better performance. In order to extract more representative features for each identity, we segment out the foreground person from the original image patch. We propose a simple yet effective re-ID method, which models foreground person and original image patches individually, and obtains enriched representations from two separate CNN streams. From the experiments on two standard person search benchmarks of CUHK-SYSU and PRW, we achieve mAP of $83.0\%$ and $32.6\%$ respectively, surpassing the state of the art by a large margin (more than 5pp).
['Wanli Ouyang', 'Shanshan Zhang', 'Jian Yang', 'Di Chen', 'Ying Tai']
2018-07-21
person-search-via-a-mask-guided-two-stream-1
http://openaccess.thecvf.com/content_ECCV_2018/html/Di_Chen_Person_Search_via_ECCV_2018_paper.html
http://openaccess.thecvf.com/content_ECCV_2018/papers/Di_Chen_Person_Search_via_ECCV_2018_paper.pdf
eccv-2018-9
['person-search']
['computer-vision']
[ 4.93664257e-02 -3.94844800e-01 1.75726175e-01 -3.37800056e-01 -8.39368701e-01 -3.75578821e-01 4.73836899e-01 -1.50434151e-01 -1.07010877e+00 7.77993619e-01 -5.07158563e-02 1.21558525e-01 3.45147550e-01 -7.12584078e-01 -7.50047445e-01 -6.66581154e-01 1.60777867e-01 4.01262611e-01 4.65590864e-01 2.07003251e-01 -2.26899281e-01 3.18148971e-01 -1.74476576e+00 1.61299556e-01 6.41688764e-01 1.06228232e+00 9.69463065e-02 6.56066179e-01 7.94385895e-02 4.73517537e-01 -7.35214293e-01 -9.57387030e-01 3.55241030e-01 -1.96067899e-01 -8.78973305e-01 1.05547845e-01 8.25013459e-01 -5.20861685e-01 -7.74347365e-01 1.31391358e+00 9.46153700e-01 2.08383173e-01 5.78635931e-01 -1.08317113e+00 -7.39641726e-01 3.36072624e-01 -9.97471511e-01 6.33956194e-01 3.11501920e-01 3.46843272e-01 7.24073708e-01 -9.59731996e-01 3.58400017e-01 1.47982383e+00 7.16390789e-01 8.17568541e-01 -1.36393583e+00 -1.17785418e+00 5.00170112e-01 4.38786596e-01 -1.92277396e+00 -5.15232325e-01 2.29045913e-01 -3.30294758e-01 7.21232414e-01 2.54965484e-01 5.33560336e-01 1.14767027e+00 -4.70146149e-01 1.19047296e+00 9.78577077e-01 -2.37588391e-01 -2.74791837e-01 3.13927770e-01 5.15347779e-01 7.28040040e-01 5.51929891e-01 -3.12247742e-02 -3.86721343e-01 -1.63952664e-01 6.83144569e-01 1.26560062e-01 -3.86496961e-01 1.09280042e-01 -1.02687597e+00 5.82697451e-01 5.73537469e-01 -9.47053060e-02 -1.67781770e-01 3.50685298e-01 3.48925263e-01 -1.72640115e-01 2.36818150e-01 -2.06414700e-01 -6.18778840e-02 7.19509646e-02 -6.90902948e-01 5.96490204e-01 3.58907789e-01 1.25458705e+00 7.14187145e-01 -2.31995910e-01 -6.30488217e-01 9.29331720e-01 2.69561291e-01 8.74261856e-01 2.70432979e-01 -5.35573423e-01 4.75297421e-01 6.27412021e-01 4.22874659e-01 -8.17882001e-01 -1.34093566e-02 -6.44860744e-01 -9.48415577e-01 -1.81355760e-01 6.01367831e-01 -2.29476243e-01 -1.15032804e+00 1.66222227e+00 3.24075878e-01 4.18397665e-01 -1.13184772e-01 1.00550652e+00 1.12351263e+00 6.26168549e-01 4.28871512e-01 2.01389536e-01 1.92560983e+00 -1.19507027e+00 -3.52989852e-01 -4.41382796e-01 1.18260115e-01 -5.85903108e-01 5.38448811e-01 2.64749136e-02 -1.11450922e+00 -9.20320570e-01 -9.17578399e-01 -1.68011621e-01 -2.45438457e-01 4.97651279e-01 2.52861023e-01 7.39124775e-01 -9.10709023e-01 2.73607165e-01 -3.66825998e-01 -5.69931030e-01 7.34391928e-01 5.76851249e-01 -4.87918913e-01 -2.61824936e-01 -9.50280011e-01 6.46254063e-01 2.53420085e-01 1.12165019e-01 -8.27034116e-01 -5.72186053e-01 -6.24889910e-01 1.13610484e-01 2.75665879e-01 -7.94504702e-01 1.13771796e+00 -6.43547893e-01 -1.11271656e+00 1.14888716e+00 -6.32669091e-01 -5.91267645e-01 6.92768037e-01 -4.08857584e-01 -4.89491493e-01 1.79672949e-02 3.23979676e-01 6.61508977e-01 8.06362450e-01 -1.24109161e+00 -1.33990633e+00 -4.81797814e-01 -1.46596044e-01 3.28407623e-02 -1.48467109e-01 5.61591744e-01 -1.24560261e+00 -6.06997192e-01 -2.88059711e-01 -8.93912613e-01 -3.04721981e-01 -5.94203454e-03 -4.11454141e-01 -3.62879068e-01 3.71275008e-01 -9.70427573e-01 1.10886908e+00 -2.01377201e+00 3.35197449e-02 7.96947554e-02 4.42577958e-01 5.19828975e-01 -2.82350443e-02 -2.77234972e-01 1.09423921e-01 -4.91017513e-02 -1.76430661e-02 -7.60841846e-01 1.00537203e-01 -1.76047921e-01 3.47514562e-02 5.99060714e-01 1.32530749e-01 9.67728555e-01 -7.05936491e-01 -6.36481822e-01 -9.59966928e-02 3.46067131e-01 -3.03269953e-01 2.22028747e-01 3.98423791e-01 1.47987813e-01 -2.39873067e-01 7.67889380e-01 9.40690696e-01 -3.02622437e-01 -7.20675886e-02 -1.89937875e-01 -2.24198420e-02 -1.63111806e-01 -1.63639140e+00 1.33356345e+00 7.86374956e-02 5.27429223e-01 2.54537798e-02 -7.98558474e-01 7.66497493e-01 8.61719847e-02 2.04451233e-01 -6.96238816e-01 1.44865021e-01 -2.16126051e-02 -1.95899829e-01 -1.56128600e-01 4.93141651e-01 3.76398295e-01 -2.32792273e-01 2.68499851e-01 3.07104468e-01 9.10538793e-01 8.63550976e-02 -6.60006283e-03 9.61057365e-01 -9.43837613e-02 1.84023857e-01 -3.43110979e-01 8.45527411e-01 -2.08538651e-01 7.04408050e-01 1.06543124e+00 -7.05300987e-01 6.11057401e-01 3.27219162e-03 -5.75067163e-01 -9.82486963e-01 -1.14296448e+00 -8.77698809e-02 1.14747679e+00 5.54017186e-01 -2.31484026e-01 -8.67195070e-01 -7.05909312e-01 2.01731473e-01 1.61540538e-01 -4.97150004e-01 -1.05431546e-02 -9.06770825e-01 -1.24924445e+00 8.33524346e-01 7.97293663e-01 1.04156601e+00 -6.78105175e-01 -3.40014935e-01 1.55834898e-01 -3.26517284e-01 -1.32549214e+00 -7.50460923e-01 -1.60971418e-01 -1.09831288e-01 -9.55815136e-01 -1.30461490e+00 -1.07180989e+00 6.95634961e-01 6.14136219e-01 1.04955208e+00 8.17861259e-02 -5.83519399e-01 1.48508549e-01 -1.16167873e-01 -3.09304029e-01 2.73194075e-01 1.07653052e-01 2.10542113e-01 2.42111906e-01 6.46011591e-01 -4.92059439e-02 -1.13785887e+00 4.54509020e-01 -1.20857760e-01 -1.93220496e-01 4.92270678e-01 8.50532234e-01 6.05219483e-01 -2.58754436e-02 3.30099881e-01 -4.55676138e-01 2.92995244e-01 -2.27553621e-01 -4.62521851e-01 3.85202229e-01 -3.93156558e-01 -2.35624418e-01 3.19419861e-01 -4.78571206e-01 -1.13687193e+00 1.53327405e-01 -2.22256213e-01 -2.25496024e-01 -4.13873404e-01 -4.64987427e-01 -3.25954646e-01 -2.74585247e-01 5.52591681e-01 6.09582365e-01 -3.28501493e-01 -7.73360252e-01 3.28620166e-01 7.46040285e-01 9.06268775e-01 -6.00014150e-01 1.04067981e+00 5.81032157e-01 -4.75877702e-01 -5.00525594e-01 -6.07473969e-01 -9.01776254e-01 -5.36018610e-01 1.95561107e-02 9.99820054e-01 -1.34427643e+00 -1.03768539e+00 6.37967348e-01 -1.21736264e+00 1.42720953e-01 -9.34422240e-02 2.56316006e-01 4.82257903e-02 6.73091471e-01 -7.17393994e-01 -9.77178156e-01 -6.92544758e-01 -1.21684539e+00 1.11554170e+00 7.40925014e-01 1.36008859e-01 -1.93945035e-01 -2.50798315e-01 3.76126051e-01 1.25293478e-01 6.88798307e-03 3.80342305e-01 -6.55152678e-01 -8.11233878e-01 -5.57818353e-01 -9.80428517e-01 1.71942055e-01 -1.60743341e-01 -5.06592572e-01 -1.18901634e+00 -5.26650488e-01 -5.40991962e-01 2.37626806e-02 1.24859571e+00 2.14080382e-02 1.11120713e+00 -2.06305668e-01 -7.41035759e-01 7.53770888e-01 1.41534913e+00 2.21871391e-01 5.34022689e-01 3.78244996e-01 7.01260626e-01 3.68875235e-01 3.37834537e-01 4.03439701e-01 6.10586703e-01 9.00142491e-01 7.52533153e-02 -2.42489651e-01 -3.98333699e-01 -1.45504147e-01 1.82883725e-01 1.10183116e-02 -5.30413210e-01 -2.34447196e-01 -8.10223162e-01 7.52272964e-01 -2.13757133e+00 -1.20653093e+00 4.63219732e-03 2.22128391e+00 6.33947253e-01 -2.88484688e-03 6.82448030e-01 -2.03045666e-01 1.12236083e+00 9.52271186e-03 -2.80385762e-01 3.28892976e-01 -1.48668349e-01 3.24897677e-01 9.46037233e-01 3.21182609e-01 -1.50301802e+00 1.11167216e+00 6.10067463e+00 9.85919416e-01 -5.03946483e-01 2.43337467e-01 7.11398065e-01 -5.90171926e-02 4.11218464e-01 -3.97163004e-01 -1.68877959e+00 7.70593882e-01 5.99426210e-01 -1.73723772e-01 3.56286407e-01 9.66414630e-01 -3.94091249e-01 -1.18686944e-01 -1.02417660e+00 1.53835142e+00 2.63895124e-01 -1.19470024e+00 2.10009459e-02 -2.65245279e-03 4.49644357e-01 -1.53488949e-01 -3.74167226e-02 5.08635044e-01 4.40297961e-01 -9.52445030e-01 6.84263527e-01 4.87133652e-01 6.52471602e-01 -8.98883879e-01 8.90254021e-01 3.10036689e-01 -1.66729522e+00 -1.76741406e-01 -6.08350098e-01 2.57692516e-01 1.23934530e-01 1.01029746e-01 -4.89278615e-01 6.14683449e-01 1.40049529e+00 3.19699317e-01 -8.69708657e-01 1.45116091e+00 1.43689021e-01 1.66488156e-01 -3.18514138e-01 7.46022817e-03 -9.32511911e-02 1.82775542e-01 3.51505816e-01 1.65664303e+00 2.33180627e-01 1.73364997e-01 3.37834537e-01 8.24986279e-01 -1.73913941e-01 -7.84160048e-02 -1.01113744e-01 5.33932745e-01 2.59368807e-01 1.15634513e+00 -4.77163225e-01 -5.26742578e-01 -6.18097246e-01 1.68966866e+00 3.91095281e-01 4.54595983e-01 -1.24140465e+00 -3.12150836e-01 1.01847506e+00 6.22838512e-02 6.22091234e-01 -5.36212772e-02 1.74527004e-01 -1.32669151e+00 1.45210668e-01 -6.63544059e-01 4.93507951e-01 -1.60963029e-01 -1.50620973e+00 7.66595185e-01 4.78523001e-02 -9.45318341e-01 -1.54989315e-02 -6.30966246e-01 -4.40834790e-01 1.32447267e+00 -1.66188061e+00 -1.42913604e+00 -5.84725440e-01 7.35182941e-01 5.47797501e-01 -3.01872730e-01 6.25451684e-01 7.54020929e-01 -1.17629504e+00 1.15883946e+00 -9.03568342e-02 6.21023834e-01 7.31187940e-01 -1.08396018e+00 9.15329039e-01 1.14710915e+00 -1.60003439e-01 6.17963731e-01 3.33695352e-01 -5.43102086e-01 -1.08617568e+00 -1.17971396e+00 9.52867091e-01 -5.46612620e-01 2.39267692e-01 -5.52217662e-01 -6.99068308e-01 5.31873167e-01 9.48054492e-02 3.01075757e-01 5.93272209e-01 3.17503256e-03 -3.55597466e-01 -2.52519280e-01 -1.16310990e+00 7.09826708e-01 1.51551068e+00 -3.59531939e-01 -3.47806841e-01 7.49395564e-02 4.70334798e-01 -2.36054659e-01 -6.17684186e-01 3.39367628e-01 5.64995944e-01 -6.76786065e-01 1.70565176e+00 -6.03496194e-01 -2.11266160e-01 -5.40831089e-01 -2.68775731e-01 -7.18309283e-01 -8.29619288e-01 -3.89083862e-01 -1.79898173e-01 1.44579768e+00 1.24619089e-01 -6.04091287e-01 8.90677571e-01 7.55027652e-01 3.55774343e-01 -4.53298658e-01 -1.08512115e+00 -8.36455226e-01 -1.44585654e-01 -5.43816797e-02 7.76342571e-01 3.33349258e-01 -5.01674652e-01 2.01184139e-01 -6.95523143e-01 5.59906244e-01 1.19735026e+00 2.95343902e-02 1.07661152e+00 -1.09885955e+00 -4.71867740e-01 -4.91527766e-01 -5.78796923e-01 -1.38637578e+00 1.06343970e-01 -5.89411139e-01 1.16194330e-01 -1.36196780e+00 8.42092693e-01 -3.28881055e-01 -4.93926734e-01 2.84634322e-01 -7.32989252e-01 5.63492417e-01 4.44337606e-01 4.63823497e-01 -7.83620059e-01 2.81759202e-01 7.61337340e-01 -4.15651828e-01 1.37633830e-01 2.21786961e-01 -8.19779217e-01 7.15413809e-01 5.55277824e-01 -2.32430652e-01 1.25461176e-01 -5.75640798e-01 -6.17398739e-01 -4.40231770e-01 8.96312296e-01 -1.25013804e+00 4.17330533e-01 2.33773991e-01 1.02209234e+00 -6.92826390e-01 5.33154905e-01 -5.57115436e-01 6.22737706e-02 4.46498096e-01 3.38049717e-02 3.40892486e-02 1.91450045e-01 8.26827586e-01 -5.71680218e-02 -1.58591166e-01 9.27559733e-01 -3.59013438e-01 -1.23120844e+00 5.88568389e-01 2.15945626e-03 -2.27155566e-01 1.08295333e+00 -3.45339447e-01 -5.80463052e-01 3.15369368e-02 -6.12378478e-01 3.82448673e-01 2.45260537e-01 2.39082247e-01 5.94626904e-01 -1.38578367e+00 -7.72322714e-01 6.05164282e-02 2.03449860e-01 -1.30376920e-01 4.65640247e-01 4.68174279e-01 -2.70753026e-01 3.58378857e-01 -9.33576450e-02 -4.72813517e-01 -1.47936618e+00 5.63007534e-01 5.07011712e-01 -7.68233314e-02 -7.72540152e-01 1.29532015e+00 4.89814639e-01 -2.38165334e-02 5.89064956e-01 2.44505972e-01 -3.17046106e-01 -1.36018902e-01 8.96677196e-01 5.05894899e-01 -3.67984116e-01 -9.79473472e-01 -7.12259650e-01 6.29988194e-01 -3.09279203e-01 5.41284606e-02 8.63596320e-01 -2.71726876e-01 1.60467848e-01 -2.67032683e-01 1.08555961e+00 -1.82617798e-01 -1.29669178e+00 -6.34920359e-01 -7.16647208e-02 -7.07145572e-01 -3.39712471e-01 -5.71821630e-01 -8.58884573e-01 5.89491844e-01 1.19104815e+00 -2.36945093e-01 8.90996218e-01 2.54620641e-01 1.08455133e+00 2.52197236e-01 4.90444571e-01 -1.06809998e+00 -5.25860861e-02 1.30098149e-01 6.18608952e-01 -1.38764977e+00 6.83124289e-02 -5.61519027e-01 -4.55551803e-01 5.75761318e-01 9.57978547e-01 -2.13505298e-01 4.82384801e-01 -6.18534023e-03 -2.80046672e-01 1.90283492e-01 -2.46795848e-01 -8.33315790e-01 4.84006971e-01 7.22361028e-01 6.49612322e-02 1.73750132e-01 1.21953316e-01 1.11424470e+00 1.78593799e-01 -1.18422821e-01 -2.07840115e-01 7.41053760e-01 -5.11530638e-01 -1.08880937e+00 -5.79573154e-01 3.37987363e-01 -4.85972494e-01 -6.22121319e-02 -1.58989817e-01 7.48553514e-01 6.31289899e-01 7.94978321e-01 1.03072591e-01 -4.91760701e-01 5.75682521e-01 -2.08614375e-02 3.77894372e-01 -4.04559284e-01 -6.42603934e-01 2.82441340e-02 1.94413289e-01 -4.51309025e-01 -2.64443547e-01 -6.77976608e-01 -7.56294549e-01 -5.77004433e-01 -2.70364255e-01 -1.17105164e-01 1.15208380e-01 6.71685636e-01 2.99920410e-01 3.56118411e-01 2.57662207e-01 -7.75485098e-01 -4.39001352e-01 -7.00964928e-01 -4.09105867e-01 5.19552350e-01 9.77434069e-02 -6.95975482e-01 1.02385422e-02 3.95305678e-02]
[14.79239273071289, 0.8354737162590027]
85f5e8d1-93b3-4dba-912b-3a925a184625
information-based-disentangled-representation
2103.13283
null
https://arxiv.org/abs/2103.13283v1
https://arxiv.org/pdf/2103.13283v1.pdf
Information-based Disentangled Representation Learning for Unsupervised MR Harmonization
Accuracy and consistency are two key factors in computer-assisted magnetic resonance (MR) image analysis. However, contrast variation from site to site caused by lack of standardization in MR acquisition impedes consistent measurements. In recent years, image harmonization approaches have been proposed to compensate for contrast variation in MR images. Current harmonization approaches either require cross-site traveling subjects for supervised training or heavily rely on site-specific harmonization models to encourage harmonization accuracy. These requirements potentially limit the application of current harmonization methods in large-scale multi-site studies. In this work, we propose an unsupervised MR harmonization framework, CALAMITI (Contrast Anatomy Learning and Analysis for MR Intensity Translation and Integration), based on information bottleneck theory. CALAMITI learns a disentangled latent space using a unified structure for multi-site harmonization without the need for traveling subjects. Our model is also able to adapt itself to harmonize MR images from a new site with fine tuning solely on images from the new site. Both qualitative and quantitative results show that the proposed method achieves superior performance compared with other unsupervised harmonization approaches.
['Jerry L. Prince', 'Peter A. Calabresi', 'Yufan He', 'Yihao Liu', 'Aaron Carass', 'Blake E. Dewey', 'Lianrui Zuo']
2021-03-24
null
null
null
null
['image-harmonization']
['computer-vision']
[ 2.27010950e-01 4.03362699e-02 -2.32720375e-01 -4.64083880e-01 -1.21436024e+00 -1.82064980e-01 2.24934191e-01 2.19003752e-01 -6.58237755e-01 7.53165364e-01 3.13039839e-01 1.24369517e-01 -7.08041370e-01 -4.03329819e-01 -4.04725760e-01 -8.54343116e-01 -3.94796804e-02 6.65316701e-01 1.46602824e-01 -1.45013496e-01 2.04768464e-01 2.83671051e-01 -9.21822190e-01 1.80547744e-01 1.23086810e+00 1.99897125e-01 6.80935562e-01 5.01351833e-01 1.92995518e-01 5.07583618e-01 -3.65619987e-01 -5.21604382e-02 2.98616469e-01 -6.31633461e-01 -9.45747435e-01 5.76500967e-02 2.60164410e-01 9.40372329e-03 -9.53489766e-02 9.54788089e-01 1.05414414e+00 2.12923765e-01 6.15377545e-01 -8.52749169e-01 -4.36986029e-01 9.31911886e-01 -9.26337540e-01 3.58202219e-01 -2.27991983e-01 4.03437577e-02 7.16199577e-01 -5.94580352e-01 1.01639211e+00 5.47136128e-01 6.34777009e-01 2.92033404e-01 -1.66108429e+00 -5.91155231e-01 -1.96228266e-01 4.07407612e-01 -1.33437514e+00 -2.80464351e-01 1.00535655e+00 -5.85583985e-01 8.46772492e-01 2.82113731e-01 6.70561731e-01 5.36623418e-01 6.83023870e-01 4.69086021e-01 1.43540549e+00 -4.43845689e-01 1.32949740e-01 2.24285899e-03 -1.45570949e-01 4.97945309e-01 1.16346352e-01 -1.19472757e-01 -5.24959981e-01 -9.56176519e-02 7.62824357e-01 -3.68538082e-01 -3.45939159e-01 -1.03884852e+00 -1.50303364e+00 9.31070030e-01 5.32456577e-01 7.58088648e-01 -5.92355430e-01 -2.53450722e-01 5.55401623e-01 2.56688982e-01 3.61386031e-01 8.24969292e-01 6.74598888e-02 2.87845761e-01 -1.45820618e+00 -3.48165371e-02 1.40471920e-01 5.23367465e-01 5.32074749e-01 -1.67330615e-02 -1.76761702e-01 1.04413092e+00 2.64686556e-03 4.15537149e-01 9.22919273e-01 -9.97939587e-01 3.73104870e-01 1.62149504e-01 -3.22421789e-01 -1.01188242e+00 -8.95268738e-01 -8.57539713e-01 -1.01835644e+00 1.98084638e-01 1.06681928e-01 1.14903159e-01 -7.28780627e-01 1.87187266e+00 2.86840439e-01 -1.43517017e-01 -2.54298002e-01 1.20104349e+00 4.16690409e-01 2.54550397e-01 9.31568593e-02 -6.10088229e-01 1.13131225e+00 -9.37649846e-01 -8.75298619e-01 1.29032075e-01 6.27695858e-01 -1.04527211e+00 1.10111082e+00 3.32055897e-01 -1.36732757e+00 -4.31937188e-01 -1.25331914e+00 2.06405982e-01 -7.56260976e-02 -5.45783937e-02 4.55749929e-01 6.89388275e-01 -1.10563886e+00 5.10881305e-01 -1.08246481e+00 -5.85470423e-02 1.44938797e-01 6.09167933e-01 -6.53824449e-01 1.69147044e-01 -1.25622153e+00 1.22859728e+00 2.43539780e-01 2.84889460e-01 -4.05807197e-01 -1.01452601e+00 -7.08939016e-01 -3.49344730e-01 2.76051581e-01 -8.65815580e-01 1.02087700e+00 -6.88870788e-01 -1.64194071e+00 9.95126069e-01 2.25500196e-01 -4.08543527e-01 5.05771220e-01 1.04055643e-01 -3.27385008e-01 5.03103137e-01 4.50307667e-01 7.66837597e-01 7.83256650e-01 -1.13832855e+00 -1.26665726e-03 -6.15822494e-01 -6.08535409e-01 4.28869724e-01 -2.00020894e-03 -1.15275465e-01 -2.12025002e-01 -8.48288000e-01 4.46071565e-01 -9.40059960e-01 -4.82908338e-01 -3.78573149e-01 -3.22246641e-01 3.99185658e-01 3.05918545e-01 -8.47476304e-01 1.16493058e+00 -1.82459152e+00 5.48986971e-01 4.58250016e-01 6.88942909e-01 -2.09737942e-01 -1.81798294e-01 -4.74371277e-02 -4.43524987e-01 -1.45264968e-01 -4.36520815e-01 -2.83498079e-01 -2.60450602e-01 1.36306167e-01 2.53661841e-01 8.01119506e-01 -3.57819825e-01 7.95155108e-01 -8.41294587e-01 -9.02674139e-01 2.62995154e-01 2.65583843e-01 -8.40583682e-01 1.20713301e-01 3.70976925e-01 1.00525057e+00 -3.43402535e-01 3.84504981e-02 6.00134790e-01 -1.72989681e-01 5.57399869e-01 -7.22533643e-01 -2.26417989e-01 -1.14818737e-01 -9.63237464e-01 2.32211041e+00 -7.01610506e-01 2.52368301e-01 7.09584057e-02 -1.18948638e+00 6.31598353e-01 4.68143165e-01 1.15617299e+00 -1.15283132e+00 7.17064962e-02 4.27613467e-01 3.89739275e-01 -3.19092065e-01 4.35746431e-01 -3.69277269e-01 -5.74441962e-02 6.77537739e-01 -1.67939682e-02 -5.31509519e-01 1.41897053e-01 2.12335005e-01 6.35297835e-01 -1.93550691e-01 2.78395474e-01 -6.30284727e-01 4.09093410e-01 -1.22819375e-02 6.62063062e-01 9.21648979e-01 -3.86131585e-01 8.12154651e-01 1.24280892e-01 -3.22860926e-01 -1.38713539e+00 -1.29131424e+00 -4.39241201e-01 5.56356370e-01 2.48745959e-02 -1.70373097e-01 -8.18884432e-01 -3.20391029e-01 -4.80541557e-01 3.25784266e-01 -5.93539774e-01 -1.31148949e-01 -8.66167903e-01 -9.60480332e-01 1.91427484e-01 2.18149081e-01 5.21433055e-01 -8.55541825e-01 -7.78573871e-01 5.03970623e-01 -7.43140519e-01 -8.53561938e-01 -6.63385987e-01 1.30365118e-01 -1.17758381e+00 -7.62239754e-01 -1.07665658e+00 -5.98105311e-01 5.19000709e-01 1.60742924e-01 1.01855528e+00 -2.45964393e-01 -6.93634987e-01 3.97568464e-01 -2.05423146e-01 2.55153812e-02 -3.39681685e-01 5.37419558e-01 1.10818081e-01 -2.52290487e-01 -3.19245398e-01 -8.05763125e-01 -9.41438973e-01 4.83052999e-01 -1.04697526e+00 3.98529500e-01 8.13584626e-01 1.11891544e+00 9.26351607e-01 1.00580498e-03 5.35738170e-01 -9.36043859e-01 6.15832984e-01 -1.58646926e-01 -4.32582110e-01 5.37050426e-01 -8.79641831e-01 3.70049685e-01 1.25246778e-01 -3.16541940e-01 -1.13152349e+00 -5.58130182e-02 1.49672881e-01 -8.55037123e-02 2.00665906e-01 6.29669607e-01 2.28403211e-01 -2.69602925e-01 8.29250216e-01 1.09093681e-01 3.14417303e-01 -9.10204425e-02 4.16705042e-01 3.12818944e-01 6.73076451e-01 -5.84083676e-01 4.29429978e-01 4.91769254e-01 -8.85922834e-02 -7.24337280e-01 -4.22502309e-01 -3.24348450e-01 -1.08291280e+00 -4.63113844e-01 1.20086575e+00 -6.62271440e-01 -2.68854707e-01 2.44444817e-01 -7.77494729e-01 -4.00713116e-01 -3.00756216e-01 1.10802448e+00 -8.41533720e-01 5.21907270e-01 -6.41560674e-01 -1.80614647e-02 -6.71884120e-01 -1.72030187e+00 8.03419232e-01 9.88948867e-02 -4.27869409e-01 -1.03821623e+00 5.09665132e-01 6.82336390e-01 7.89490402e-01 1.70629263e-01 1.02397990e+00 -2.58866251e-02 -4.26992685e-01 1.42322972e-01 9.68260691e-02 1.88527419e-03 3.28846991e-01 -4.84333724e-01 -3.74429971e-01 -3.00782770e-01 3.61285865e-01 -2.07775921e-01 4.52030897e-01 9.87527072e-01 9.11594689e-01 1.87951937e-01 1.33563519e-01 5.52094936e-01 1.18723309e+00 -6.98410273e-02 6.72643721e-01 6.82400823e-01 4.76391762e-01 7.49228179e-01 3.44388664e-01 3.23263228e-01 3.91666681e-01 1.10964262e+00 -2.15706751e-01 -5.33626437e-01 -3.41824323e-01 7.94528350e-02 -1.76225215e-01 1.47018993e+00 -2.08532270e-02 3.04071426e-01 -9.56119239e-01 4.35347795e-01 -1.72594190e+00 -8.17111492e-01 -1.43929143e-02 2.13732481e+00 1.02161300e+00 -1.69458151e-01 3.31029035e-02 -1.64278224e-01 7.83342898e-01 1.04675286e-01 -4.08210963e-01 -9.81915221e-02 -2.23688483e-01 2.88550228e-01 6.29655302e-01 6.57128751e-01 -8.40251803e-01 7.46340692e-01 6.72344255e+00 7.68775821e-01 -1.29077148e+00 7.03299522e-01 5.84929168e-01 -9.88514423e-02 -3.99198323e-01 -1.02364935e-01 6.83474019e-02 4.04867865e-02 5.83324432e-01 -2.04510003e-01 3.36168557e-01 2.26189956e-01 4.78112906e-01 -3.17724138e-01 -6.17209613e-01 1.10598898e+00 2.09516451e-01 -1.24693215e+00 -8.15462545e-02 -1.57304808e-01 9.17138934e-01 -4.54419069e-02 4.78587568e-01 -1.33862704e-01 -1.99453041e-01 -8.88051391e-01 3.68961811e-01 6.19618952e-01 7.06954658e-01 -6.27535403e-01 7.39521325e-01 -3.58505659e-02 -8.97612035e-01 4.87352520e-01 -2.25084886e-01 5.50485551e-01 4.29301590e-01 5.95483124e-01 -6.41444147e-01 8.58954310e-01 4.95775282e-01 1.66324109e-01 -5.37612557e-01 9.23057914e-01 -1.30185470e-01 2.67964721e-01 -1.35734424e-01 6.57417417e-01 4.75335754e-02 -3.58970463e-01 5.83837032e-01 8.06361616e-01 1.13596492e-01 1.56255569e-02 1.68288946e-01 7.17050731e-01 4.27887917e-01 6.17780507e-01 -3.02685708e-01 2.84145504e-01 3.27289738e-02 1.33336449e+00 -1.02309704e+00 -2.78726190e-01 -2.51193464e-01 9.57295120e-01 2.16857508e-01 2.33937338e-01 -7.89337337e-01 -4.73565720e-02 -7.01432824e-02 1.85093716e-01 -9.69563052e-02 -4.02810335e-01 -4.85301942e-01 -1.17314506e+00 -1.92132741e-01 -9.37113762e-01 6.81050122e-01 -7.83510685e-01 -1.20917487e+00 7.32342601e-01 3.81226718e-01 -1.01860988e+00 -2.75046557e-01 -3.38200033e-02 -3.28725398e-01 7.53263652e-01 -1.15208220e+00 -1.06366563e+00 -6.36469275e-02 7.67238975e-01 2.23119408e-01 -1.52560875e-01 6.75107777e-01 6.51172936e-01 -4.12906170e-01 6.71727657e-01 8.39532092e-02 -2.25505561e-01 1.17248571e+00 -1.20760882e+00 -3.57405931e-01 7.53559768e-01 5.26524223e-02 6.97183192e-01 8.45428050e-01 -7.08539844e-01 -8.78320992e-01 -4.53658730e-01 7.52430618e-01 2.62847263e-02 6.23590469e-01 -7.26712793e-02 -9.08739209e-01 3.87651235e-01 3.91203105e-01 -2.52657622e-01 9.58547175e-01 9.02670994e-02 -9.55100209e-02 -2.57163823e-01 -1.05255616e+00 6.38910949e-01 7.71840751e-01 -5.46925426e-01 -6.78416669e-01 3.72647613e-01 3.58405739e-01 -4.92898643e-01 -1.35875976e+00 4.11414117e-01 5.62254548e-01 -9.58888948e-01 9.01570201e-01 -1.85742110e-01 2.24595100e-01 -3.25774729e-01 2.17268810e-01 -1.41510832e+00 -4.77375656e-01 -6.32846534e-01 6.25728905e-01 5.85420251e-01 4.33587343e-01 -5.47456503e-01 6.75533772e-01 7.69712627e-01 -2.56799370e-01 -3.15492243e-01 -1.12199104e+00 -5.49040496e-01 3.01375926e-01 -2.85557479e-01 2.31900215e-01 1.12931454e+00 1.01977177e-01 2.61455476e-01 -5.85755050e-01 1.69472694e-01 1.08870924e+00 3.68551403e-01 5.30042946e-01 -7.50041842e-01 -6.11959338e-01 -5.79119265e-01 -4.86821800e-01 -3.38108957e-01 2.26106755e-02 -1.15614951e+00 -8.23989511e-02 -1.30824649e+00 6.67411029e-01 -5.35159290e-01 -4.36728150e-01 3.16877782e-01 -1.45247439e-02 5.43192208e-01 1.32903263e-01 4.69167382e-01 -4.54588473e-01 6.33665979e-01 1.57711184e+00 -1.63186640e-01 -4.31402355e-01 -3.05943489e-01 -5.00131667e-01 2.61119664e-01 9.00059581e-01 -6.76132500e-01 -6.16610348e-01 -4.96870071e-01 1.85233086e-01 4.19837058e-01 5.60548492e-02 -9.75322604e-01 3.65977377e-01 1.38524130e-01 2.32092112e-01 -3.13823283e-01 -1.28464788e-01 -5.28674483e-01 5.12491167e-01 5.94180763e-01 -4.26695794e-01 3.51640970e-01 1.01556415e-02 7.18187243e-02 -4.35692161e-01 -1.24171868e-01 1.06452858e+00 -4.37704362e-02 -2.98021168e-01 3.67036253e-01 -5.01565456e-01 -3.78038585e-02 7.76098728e-01 -9.79673415e-02 1.59791231e-01 -3.20735395e-01 -1.30366611e+00 1.41257271e-01 1.42963409e-01 2.67616898e-01 5.56691051e-01 -1.17246652e+00 -8.60534847e-01 -7.07363570e-03 4.51405048e-02 -3.81393641e-01 1.10585558e+00 1.65926611e+00 -5.56463778e-01 4.32654142e-01 -6.35127902e-01 -9.01739776e-01 -1.16208768e+00 3.26136589e-01 5.74095190e-01 -8.65493953e-01 -7.05357015e-01 2.59976476e-01 3.97972912e-01 -5.89420319e-01 -2.35273197e-01 3.53407636e-02 -9.08436626e-02 -2.57433538e-04 1.32080629e-01 1.39727980e-01 2.49443695e-01 -7.18061984e-01 -2.58791059e-01 9.25442398e-01 -3.79817992e-01 -5.70004702e-01 1.26931918e+00 -4.51581538e-01 -1.64659157e-01 3.21722299e-01 1.18768489e+00 -3.63244340e-02 -7.15110183e-01 -2.88919300e-01 -1.35027155e-01 -2.42645547e-01 4.28581059e-01 -7.66409993e-01 -1.18707395e+00 6.54875576e-01 1.11526513e+00 -5.86232722e-01 1.22288549e+00 -1.16500802e-01 6.30960882e-01 6.47486225e-02 4.56809729e-01 -1.32565236e+00 1.89046517e-01 -6.03941157e-02 9.72786725e-01 -1.08024967e+00 2.32760087e-01 -1.70693710e-01 -9.05082047e-01 9.20718193e-01 3.25388372e-01 8.72667357e-02 5.47523022e-01 1.97957680e-01 2.85595357e-01 -3.95651162e-01 -1.87002748e-01 1.16587579e-01 4.80849445e-01 5.07260561e-01 5.88706970e-01 2.24783391e-01 -8.84333074e-01 2.16862917e-01 -2.48892546e-01 -2.60073960e-01 3.46166432e-01 7.59282589e-01 -1.27089173e-02 -1.47048342e+00 -4.17665690e-01 1.66599825e-01 -2.91294903e-01 -1.87149141e-02 1.78002238e-01 9.16735530e-01 1.33159058e-02 5.30170083e-01 -1.31567866e-01 3.63714173e-02 2.68499643e-01 -2.19126672e-01 9.09198284e-01 -4.20683146e-01 -6.17622077e-01 4.83443320e-01 -4.26980257e-01 -5.75094819e-01 -7.76935458e-01 -6.73046768e-01 -1.26539719e+00 -9.01560262e-02 -2.05219716e-01 3.94152492e-01 7.52484500e-01 8.00216556e-01 2.73501366e-01 6.87424481e-01 5.86810112e-01 -6.69891775e-01 -2.87462026e-01 -8.28731835e-01 -7.79742897e-01 4.73057508e-01 -2.46214420e-01 -6.86437726e-01 1.64304432e-02 -9.14562121e-02]
[13.782447814941406, -2.344944715499878]
8f6af934-d66d-489c-96aa-d8d6686888c1
one-shot-learning-from-a-demonstration-with-1
2203.04806
null
https://arxiv.org/abs/2203.04806v1
https://arxiv.org/pdf/2203.04806v1.pdf
One-Shot Learning from a Demonstration with Hierarchical Latent Language
Humans have the capability, aided by the expressive compositionality of their language, to learn quickly by demonstration. They are able to describe unseen task-performing procedures and generalize their execution to other contexts. In this work, we introduce DescribeWorld, an environment designed to test this sort of generalization skill in grounded agents, where tasks are linguistically and procedurally composed of elementary concepts. The agent observes a single task demonstration in a Minecraft-like grid world, and is then asked to carry out the same task in a new map. To enable such a level of generalization, we propose a neural agent infused with hierarchical latent language--both at the level of task inference and subtask planning. Our agent first generates a textual description of the demonstrated unseen task, then leverages this description to replicate it. Through multiple evaluation scenarios and a suite of generalization tests, we find that agents that perform text-based inference are better equipped for the challenge under a random split of tasks.
['Benjamin Van Durme', 'Harm van Seijen', 'Ida Momennejad', 'Romain Laroche', 'Matthew Hausknecht', 'Marc-Alexandre Côté', 'Xingdi Yuan', 'Nathaniel Weir']
2022-03-09
null
null
null
null
['one-shot-learning']
['methodology']
[ 2.70478636e-01 3.89717400e-01 2.78468311e-01 -4.08629358e-01 -5.19646049e-01 -1.03762150e+00 1.26000607e+00 7.31236339e-02 -2.88684994e-01 7.09962487e-01 3.95463526e-01 -3.70943397e-01 8.25802013e-02 -7.07371891e-01 -6.80970311e-01 -3.81115377e-01 -3.12965214e-01 9.99978602e-01 6.97411075e-02 -3.84873182e-01 1.92597359e-01 5.14059305e-01 -1.68767846e+00 5.38458467e-01 8.13442945e-01 3.16119045e-01 7.76237726e-01 6.64481103e-01 2.24224791e-01 1.50477648e+00 -4.78863031e-01 4.78312932e-02 2.90508300e-01 -4.16388035e-01 -1.09746349e+00 2.35116091e-02 2.31955945e-01 -5.83855033e-01 -3.54757369e-01 7.11138189e-01 2.70640161e-02 6.81198597e-01 8.19958568e-01 -1.35210955e+00 -5.12189388e-01 9.01595116e-01 3.57843906e-01 -6.11511469e-02 8.17968726e-01 7.82608628e-01 8.94389093e-01 -6.29429638e-01 7.25199938e-01 1.31499314e+00 4.54465628e-01 8.22161138e-01 -1.37545574e+00 -4.93542761e-01 2.49356955e-01 -2.17172638e-01 -1.13870394e+00 -3.84919047e-01 4.22982812e-01 -5.93632042e-01 1.46604478e+00 -1.46092892e-01 6.16792321e-01 1.45026958e+00 1.58109248e-01 6.26843750e-01 1.34118795e+00 -2.22293869e-01 7.00370431e-01 9.03309509e-02 -2.68638462e-01 1.12532520e+00 1.43971562e-01 7.12409616e-01 -8.92086625e-01 -8.16592500e-02 7.77227640e-01 7.05715492e-02 -9.23926607e-02 -4.92494911e-01 -1.53690410e+00 5.08534610e-01 4.97162968e-01 2.93010682e-01 -6.09951496e-01 3.78654450e-01 5.12860179e-01 3.09696794e-01 -8.27725157e-02 1.25219774e+00 -3.36169302e-01 -1.28784582e-01 -8.94431829e-01 8.81535292e-01 1.20075905e+00 1.31951070e+00 7.16282010e-01 1.90468490e-01 -3.65618110e-01 6.35559857e-02 -5.92015684e-02 3.54549587e-01 6.96689963e-01 -1.37942910e+00 4.38605785e-01 5.07247388e-01 3.36712271e-01 -4.51828033e-01 -5.55851698e-01 -2.25178942e-01 -4.01158124e-01 6.57752454e-01 1.94503710e-01 -1.92947298e-01 -8.12728405e-01 2.03278422e+00 1.03243992e-01 3.22802514e-02 5.58647633e-01 8.71415317e-01 3.02892178e-01 6.43421531e-01 4.59347457e-01 3.58443148e-02 1.40163934e+00 -1.13108253e+00 -1.64829060e-01 -7.37757623e-01 8.86562228e-01 3.47791433e-01 1.53918886e+00 3.53128225e-01 -1.24442470e+00 -7.42793739e-01 -1.25100553e+00 -2.52784222e-01 -5.29738784e-01 -4.33778554e-01 8.83631945e-01 5.68735637e-02 -1.24609208e+00 6.24601960e-01 -1.04565370e+00 -5.66842496e-01 3.58879626e-01 -5.79906702e-02 -4.59681660e-01 -8.66824538e-02 -8.17229509e-01 1.32865584e+00 1.01034701e+00 -1.57871872e-01 -2.03473663e+00 -5.80705047e-01 -1.26200092e+00 3.37238044e-01 5.39249241e-01 -1.24048162e+00 1.83380222e+00 -4.73174661e-01 -1.45726740e+00 9.15046990e-01 1.01053655e-01 -5.28785467e-01 4.66968775e-01 4.69118468e-02 1.49290264e-01 -4.61003743e-02 4.43891615e-01 8.96109760e-01 6.52807832e-01 -1.41123450e+00 -7.50859737e-01 -4.31269854e-01 6.80837750e-01 5.07239938e-01 3.90946180e-01 -3.74893546e-01 2.94555515e-01 -3.48948032e-01 7.68390968e-02 -1.02489114e+00 -1.98798403e-01 -4.46785450e-01 -2.96740949e-01 -2.59603322e-01 2.99014151e-01 -3.89787376e-01 4.32788342e-01 -2.10785556e+00 4.81314272e-01 -1.44543633e-01 2.93086618e-01 -3.35868090e-01 -1.19400220e-02 7.28050768e-01 2.35157564e-01 9.16328058e-02 -2.83224612e-01 -5.51373005e-01 7.48456776e-01 2.25595474e-01 -6.19002044e-01 1.44951344e-02 -4.58197296e-02 1.25230670e+00 -1.32525229e+00 -1.77797556e-01 5.14370203e-02 -6.40709698e-02 -6.45438850e-01 5.01765251e-01 -8.93807232e-01 7.09471703e-01 -5.43710291e-01 3.71489733e-01 -8.60312656e-02 -1.29419699e-01 3.08354467e-01 2.31533140e-01 -1.65683195e-01 4.93160725e-01 -7.48783529e-01 2.41298628e+00 -9.09134388e-01 5.29211044e-01 1.43917259e-02 -5.72492540e-01 5.16861677e-01 3.07702065e-01 -4.53339010e-01 -4.05227661e-01 -1.50542080e-01 1.12155795e-01 1.84603527e-01 -7.16861606e-01 5.57286561e-01 -4.46232587e-01 -4.98715639e-01 8.34251702e-01 1.58848017e-01 -9.24790025e-01 2.35874712e-01 3.75124127e-01 1.28842628e+00 6.70991838e-01 4.60338831e-01 -4.33169574e-01 2.42067665e-01 5.40627539e-01 -6.93690553e-02 1.33017921e+00 -6.55117780e-02 2.44520474e-02 2.50352681e-01 -7.70196021e-01 -1.22524142e+00 -1.33230925e+00 4.60221499e-01 1.61379886e+00 -1.62106678e-01 -3.08461964e-01 -7.00233459e-01 -5.79300880e-01 2.42330208e-02 1.47738802e+00 -8.09121311e-01 -1.91812649e-01 -5.76819777e-01 1.31888077e-01 7.53852308e-01 6.77017033e-01 7.34479010e-01 -1.81903505e+00 -1.42538178e+00 1.52350470e-01 1.17861487e-01 -1.03404677e+00 -1.63636953e-01 4.41129476e-01 -7.97067344e-01 -6.99996352e-01 -2.25235686e-01 -1.02629387e+00 5.09338856e-01 -2.20339730e-01 1.32981241e+00 2.59467680e-03 -9.19516385e-02 6.79941595e-01 -1.73882246e-01 -3.53276521e-01 -8.42602134e-01 2.07012117e-01 2.05424607e-01 -6.91788316e-01 6.86520636e-02 -8.96926284e-01 -2.11018637e-01 -1.15613244e-01 -9.05801237e-01 5.06537259e-01 6.44731939e-01 7.59993494e-01 4.58271913e-02 8.16121697e-02 3.77590686e-01 -7.84950614e-01 1.18633652e+00 -4.40809399e-01 -7.84634471e-01 3.25931400e-01 -3.17787349e-01 5.62443733e-01 7.80790627e-01 -3.53088111e-01 -1.23164868e+00 -4.34274375e-02 5.48005760e-01 2.59319171e-02 -5.96648693e-01 6.68641448e-01 -3.91298644e-02 4.11502570e-01 1.10715032e+00 6.31362617e-01 -2.01231673e-01 -8.03859308e-02 7.12737560e-01 1.43212438e-01 1.00223100e+00 -1.35948563e+00 8.23886514e-01 3.13423663e-01 -3.53996232e-02 -4.96396601e-01 -8.15467775e-01 8.39449763e-02 -5.43802619e-01 9.65406895e-02 1.05368304e+00 -8.07446122e-01 -1.18208468e+00 1.49036080e-01 -1.10477507e+00 -1.44790649e+00 -5.49375832e-01 3.02816033e-01 -1.25370347e+00 -2.50769615e-01 -5.59493363e-01 -6.45857453e-01 1.05223529e-01 -1.34870386e+00 1.15116870e+00 1.71979312e-02 -5.36505759e-01 -9.96873319e-01 2.41929263e-01 -8.05118401e-03 5.32891452e-01 2.03996837e-01 1.27982926e+00 -9.89627421e-01 -6.84212267e-01 2.86436111e-01 -2.13179253e-02 -9.96438041e-02 -8.72670412e-02 -8.07264924e-01 -1.03847528e+00 -4.01144475e-01 2.69651473e-01 -9.81062770e-01 3.72100681e-01 -1.70355365e-01 9.43149745e-01 -4.13121223e-01 -4.14180875e-01 6.02572799e-01 1.04855347e+00 1.99462976e-02 2.72561610e-01 5.16833425e-01 1.86021894e-01 6.79129064e-01 3.17655057e-01 2.13907957e-01 6.28008604e-01 5.38873911e-01 2.22139537e-01 4.11532998e-01 5.79912923e-02 -7.42254257e-01 4.45628554e-01 1.96895793e-01 -7.34656677e-02 3.04966774e-02 -1.35726976e+00 3.96419168e-01 -1.73649323e+00 -1.13300526e+00 8.48574400e-01 1.69715869e+00 1.02126110e+00 1.98769882e-01 3.98792848e-02 -5.08320451e-01 1.54320762e-01 1.18533194e-01 -7.92581320e-01 -5.01617908e-01 3.04525405e-01 1.18277811e-01 -1.15145132e-01 8.84990513e-01 -6.30750477e-01 1.30881011e+00 6.51093197e+00 1.79058671e-01 -5.55457115e-01 -1.08917855e-01 2.35625014e-01 2.30449857e-03 -1.65197507e-01 7.79299065e-03 -4.43752110e-01 3.08467019e-02 1.00178599e+00 -3.43519032e-01 1.05130875e+00 8.52011621e-01 -6.71528429e-02 -2.36465722e-01 -2.10950089e+00 5.37826240e-01 9.82966051e-02 -1.32346070e+00 2.43084460e-01 -2.16493279e-01 6.33746207e-01 5.27017489e-02 2.56576508e-01 1.11082685e+00 1.01122844e+00 -1.57249987e+00 8.91412735e-01 3.81998301e-01 6.57393336e-01 -1.42844498e-01 3.61999422e-01 1.14330721e+00 -8.09095144e-01 -3.49769473e-01 -7.06057400e-02 -6.45837843e-01 1.40062705e-01 -5.44060528e-01 -1.31185377e+00 1.75796166e-01 3.47141534e-01 1.75483644e-01 -3.88402611e-01 3.67512554e-01 -6.80339336e-01 -3.33931670e-02 -8.76616240e-02 -1.61324620e-01 4.07739729e-01 -8.15377757e-02 3.97763461e-01 1.15077841e+00 3.03806424e-01 4.74920630e-01 6.80708766e-01 1.41639221e+00 1.23292468e-01 -4.70669329e-01 -1.05893075e+00 -4.92722727e-02 6.15014672e-01 8.91205966e-01 -3.95983964e-01 -6.77666545e-01 -3.82134132e-02 1.06397474e+00 7.07079947e-01 7.61678755e-01 -5.63954532e-01 9.72044375e-03 2.85158724e-01 -1.40767977e-01 -9.08458754e-02 -6.79001927e-01 -1.37785301e-01 -1.03569245e+00 -8.11344311e-02 -1.20528698e+00 5.82595207e-02 -1.54495978e+00 -1.00978899e+00 7.26461172e-01 4.70781744e-01 -5.14228761e-01 -9.30124223e-01 -6.15516901e-01 -5.14908969e-01 9.95848060e-01 -1.04681802e+00 -1.26991415e+00 -6.68078780e-01 5.59845448e-01 8.93022120e-01 -4.98655796e-01 1.21749270e+00 -6.29305184e-01 2.06311885e-03 -2.95875538e-02 -6.37691915e-01 -9.58686881e-03 2.02949896e-01 -1.37407959e+00 6.99679136e-01 4.52996194e-01 1.64687559e-01 1.00361586e+00 1.00223446e+00 -7.37639785e-01 -1.19683611e+00 -9.19969141e-01 5.29563069e-01 -1.00251043e+00 8.24165702e-01 -8.70574474e-01 -8.28597546e-01 1.43819857e+00 3.13175112e-01 -2.68998742e-01 3.20145041e-01 5.98840266e-02 -5.69853902e-01 2.75230169e-01 -1.11150169e+00 1.00552642e+00 1.48866165e+00 -1.06836462e+00 -1.59890962e+00 6.37962878e-01 9.39175069e-01 -5.96289277e-01 -5.26680946e-01 4.28600162e-02 5.44286191e-01 -9.08890724e-01 7.80850649e-01 -1.16000009e+00 5.42888761e-01 -3.30236495e-01 -4.04307425e-01 -1.67247391e+00 -4.80260700e-01 -7.15247869e-01 7.32460693e-02 6.08613014e-01 5.91222107e-01 -7.67135799e-01 6.03481174e-01 6.69167697e-01 -4.46648300e-01 -2.86205471e-01 -5.23827553e-01 -9.45406437e-01 2.34421849e-01 -3.08750629e-01 7.67942309e-01 6.87346697e-01 6.45724356e-01 4.86384153e-01 3.76649529e-01 2.99082905e-01 5.30051649e-01 1.66997865e-01 9.16561842e-01 -1.04317486e+00 -5.69623947e-01 -4.14556324e-01 3.42420526e-02 -1.09813976e+00 7.10636497e-01 -1.29574251e+00 4.89101052e-01 -1.52234995e+00 2.50752091e-01 -3.96820307e-01 1.08306631e-01 6.54325843e-01 2.11731002e-01 -3.49049181e-01 3.39373082e-01 3.15842479e-01 -7.89361954e-01 3.33740115e-01 1.14434743e+00 -8.53281990e-02 -3.74349773e-01 -3.50851744e-01 -5.77627718e-01 9.13881719e-01 8.89376581e-01 -3.35179776e-01 -8.33497643e-01 -6.68794572e-01 4.71952885e-01 1.87218100e-01 7.30461597e-01 -1.32852530e+00 7.61574090e-01 -3.27591747e-01 3.41256946e-01 -5.65649122e-02 4.12037551e-01 -8.18264186e-01 7.18524754e-02 5.99008262e-01 -9.83313203e-01 5.61841309e-01 4.10313934e-01 4.34973270e-01 3.14684026e-02 -2.19874829e-01 2.21499890e-01 -7.73887873e-01 -1.02196622e+00 1.65570173e-02 -5.96329033e-01 2.71504998e-01 1.11792290e+00 -1.60589308e-01 -5.86090505e-01 -3.57628882e-01 -1.08813143e+00 3.25507045e-01 8.70087385e-01 8.15951228e-02 5.15108109e-01 -9.50563967e-01 -8.18172395e-01 2.66260982e-01 3.85057271e-01 2.41587237e-01 -2.32677814e-02 4.07202363e-01 -6.20173514e-01 4.61164802e-01 -4.71582055e-01 -4.18070078e-01 -3.35870296e-01 8.90913248e-01 6.05695665e-01 -2.36219883e-01 -8.21212292e-01 7.60219276e-01 6.71278536e-01 -7.09854424e-01 1.61521867e-01 -7.05363572e-01 1.97972223e-01 -4.71355945e-01 4.65354025e-01 -6.96004704e-02 -1.33514330e-01 -3.85971256e-02 -1.33040011e-01 7.46665942e-03 -6.39186874e-02 -6.64177179e-01 1.11259675e+00 2.15050444e-01 -8.37733373e-02 6.10611379e-01 5.80781400e-01 -2.41302669e-01 -1.61622822e+00 -1.07372999e-01 1.21366546e-01 -1.08008549e-01 -4.37182367e-01 -1.22684729e+00 7.78499171e-02 7.68004417e-01 -1.83124274e-01 3.54169428e-01 6.45745635e-01 2.84642220e-01 1.22178704e-01 1.08319604e+00 8.49021375e-01 -8.68741453e-01 4.08101708e-01 7.87056327e-01 1.21942329e+00 -9.80881512e-01 -3.08537092e-02 1.60697937e-01 -8.08484733e-01 8.57390404e-01 6.14220381e-01 -2.06122369e-01 -1.94202617e-01 3.87117475e-01 -3.62829357e-01 -5.33016026e-01 -1.16928542e+00 -3.87250036e-02 -5.73611408e-02 1.03007877e+00 -1.71410926e-02 1.06362894e-01 7.23969102e-01 3.69733632e-01 -7.85283387e-01 6.66422099e-02 5.28420448e-01 1.03651595e+00 -5.95200121e-01 -4.12780106e-01 3.13896663e-03 9.42577198e-02 3.19838732e-01 -1.53369650e-01 -3.03748727e-01 9.94940042e-01 -1.64572150e-02 6.51408315e-01 -1.83958479e-03 2.09649168e-02 3.95805538e-01 5.59712648e-01 8.26003253e-01 -1.21260083e+00 -6.89752162e-01 -8.17872941e-01 1.61685809e-01 -8.79077554e-01 -3.57824154e-02 -6.90856695e-01 -1.46812141e+00 -1.79395303e-01 5.78013897e-01 2.47241780e-01 5.76925874e-01 9.97417331e-01 1.22959375e-01 6.70726717e-01 4.47303951e-02 -1.21973908e+00 -9.68841791e-01 -1.03840566e+00 -4.71832365e-01 6.73747301e-01 5.00088215e-01 -5.66385746e-01 -4.25460696e-01 1.90568522e-01]
[4.298157215118408, 0.9836122989654541]
8ae4381a-354b-42c5-9e2f-ed5460df74b7
improving-speaker-verification-with-self
2305.10517
null
https://arxiv.org/abs/2305.10517v1
https://arxiv.org/pdf/2305.10517v1.pdf
Improving Speaker Verification with Self-Pretrained Transformer Models
Recently, fine-tuning large pre-trained Transformer models using downstream datasets has received a rising interest. Despite their success, it is still challenging to disentangle the benefits of large-scale datasets and Transformer structures from the limitations of the pre-training. In this paper, we introduce a hierarchical training approach, named self-pretraining, in which Transformer models are pretrained and finetuned on the same dataset. Three pre-trained models including HuBERT, Conformer and WavLM are evaluated on four different speaker verification datasets with varying sizes. Our experiments show that these self-pretrained models achieve competitive performance on downstream speaker verification tasks with only one-third of the data compared to Librispeech pretraining, such as VoxCeleb1 and CNCeleb1. Furthermore, when pre-training only on the VoxCeleb2-dev, the Conformer model outperforms the one pre-trained on 94k hours of data using the same fine-tuning settings.
['Jan Černocký', 'Lukáš Burget', 'Ladislav Mošner', 'Themos Stafylakis', 'Oldřich Plchot', 'Junyi Peng']
2023-05-17
null
null
null
null
['speaker-verification']
['speech']
[-6.21139295e-02 7.52702877e-02 1.46081829e-02 -6.79248273e-01 -1.30599308e+00 -7.14729905e-01 5.93051851e-01 -2.69625545e-01 -4.13034528e-01 6.35680974e-01 4.16020662e-01 -4.53398347e-01 7.41490200e-02 -3.45064849e-01 -6.86136603e-01 -5.22006989e-01 2.24493146e-01 7.42774189e-01 1.93558991e-01 -2.64902651e-01 -2.23412082e-01 1.31739289e-01 -1.40869081e+00 4.43444669e-01 8.09742868e-01 1.09416318e+00 -1.87789366e-01 6.93696082e-01 2.99687803e-01 3.99925321e-01 -6.07702017e-01 -8.89375687e-01 2.27114141e-01 -8.44062492e-02 -8.96764219e-01 -1.97487012e-01 9.15023327e-01 -2.87607789e-01 -2.92856187e-01 7.18638480e-01 8.13116789e-01 -3.67215611e-02 2.86569059e-01 -1.22193098e+00 -7.50564754e-01 1.27361310e+00 -1.29317001e-01 3.67901087e-01 5.84897324e-02 2.32876316e-01 1.17749238e+00 -9.36699986e-01 3.69538546e-01 1.30516529e+00 1.05015004e+00 7.06913888e-01 -1.22526550e+00 -1.07650912e+00 6.54219687e-02 2.69732982e-01 -1.67677784e+00 -1.13540924e+00 4.39650267e-01 -2.56232917e-01 1.20993078e+00 1.79003835e-01 -7.53109460e-04 1.46629274e+00 -1.89912856e-01 6.84833825e-01 1.14703250e+00 -2.90249705e-01 -8.98782611e-02 3.37588906e-01 2.54103065e-01 5.48521638e-01 -2.78268725e-01 4.49141771e-01 -8.47958624e-01 -1.27567008e-01 4.13058937e-01 -4.93320316e-01 -4.51762915e-01 -4.38521020e-02 -1.24485028e+00 8.72862697e-01 2.71826446e-01 3.99485946e-01 8.37157443e-02 -2.12552458e-01 7.32610404e-01 6.69956684e-01 4.78041291e-01 1.70273244e-01 -9.52490687e-01 -3.49698246e-01 -1.16236353e+00 -1.04303226e-01 6.61393046e-01 1.19542813e+00 5.98971009e-01 2.31844217e-01 -3.64803016e-01 1.12076986e+00 2.56825268e-01 4.19155270e-01 8.89608920e-01 -4.33160186e-01 8.92843306e-01 2.92343557e-01 -4.09751981e-01 7.73227811e-02 -7.77547136e-02 -5.70936501e-01 -8.77578914e-01 -2.03047007e-01 4.44176704e-01 -2.58942485e-01 -1.15455294e+00 1.86598253e+00 3.03610444e-01 2.79831141e-01 2.62025714e-01 5.77988386e-01 1.15850782e+00 6.88822687e-01 1.87672004e-01 6.34718016e-02 1.25164080e+00 -1.27981436e+00 -4.66071725e-01 -7.76685029e-02 5.21759331e-01 -8.34040225e-01 1.15363801e+00 3.05475771e-01 -8.90516281e-01 -8.63085091e-01 -1.06387663e+00 -1.60118535e-01 -4.53435510e-01 1.77586332e-01 2.21235201e-01 1.12027025e+00 -1.40883374e+00 4.57121581e-01 -5.36926091e-01 -4.88410532e-01 3.56550634e-01 4.78494376e-01 -6.58045948e-01 1.18205681e-01 -1.23585820e+00 9.41169918e-01 1.98311508e-01 6.86242580e-02 -1.48922968e+00 -1.18256712e+00 -7.45693684e-01 3.05905342e-01 -7.82086551e-02 -5.13373554e-01 1.61519730e+00 -6.62931442e-01 -1.94908512e+00 1.05267787e+00 -1.73906133e-01 -6.49795055e-01 6.64960623e-01 -6.59926310e-02 -7.16777623e-01 -4.70797271e-01 -9.07400921e-02 6.08305454e-01 1.06674671e+00 -9.18255746e-01 -6.09413505e-01 -3.40627283e-01 -1.63015038e-01 -1.14339545e-01 -3.97569388e-01 1.88723609e-01 -3.47465485e-01 -4.01584446e-01 -4.44006145e-01 -7.43741691e-01 1.51134089e-01 -6.98251188e-01 -6.96946561e-01 -4.95922923e-01 1.08235061e+00 -6.98268116e-01 1.07745755e+00 -2.37937307e+00 3.93861383e-02 1.59671992e-01 1.06692463e-01 5.13587594e-01 -5.90608239e-01 3.00154001e-01 -2.24089831e-01 1.00859568e-01 -9.90226194e-02 -7.13489890e-01 2.48339608e-01 -2.65137549e-03 -4.61259305e-01 4.69373852e-01 -1.16495773e-01 8.31784725e-01 -3.60849380e-01 -3.78802419e-01 7.17949867e-02 5.89199901e-01 -6.06274843e-01 4.41279799e-01 6.46304861e-02 5.23185074e-01 -1.00911081e-01 6.02749944e-01 7.78009176e-01 5.92825413e-02 -2.95162089e-02 -1.79051548e-01 -5.37432469e-02 6.28610373e-01 -8.44762743e-01 1.64461458e+00 -5.73495686e-01 6.25857711e-01 3.35041344e-01 -8.04042280e-01 8.05786908e-01 7.23093092e-01 7.72810448e-03 -7.84698009e-01 1.92187577e-01 1.66409835e-01 -8.80237296e-03 -1.41947255e-01 3.91797245e-01 -6.05786264e-01 -2.62208223e-01 4.93649840e-01 5.78957617e-01 -2.50389390e-02 -1.48723889e-02 7.31091797e-02 9.18064713e-01 -2.36781582e-01 -1.14220520e-02 -1.71084672e-01 9.11732793e-01 -4.42773283e-01 6.32081747e-01 5.98614097e-01 -5.46527565e-01 5.45538783e-01 8.85633528e-02 -7.05201179e-02 -8.29669058e-01 -1.13341451e+00 -3.47295612e-01 1.54173064e+00 -4.32375669e-01 -7.56838024e-01 -1.04389000e+00 -8.50121319e-01 6.11096807e-02 6.53600454e-01 -6.59601212e-01 -8.76362026e-02 -5.95343471e-01 -4.92017925e-01 1.29609132e+00 4.88491565e-01 7.16326475e-01 -8.68499458e-01 4.58045974e-02 6.38854515e-04 -2.75491536e-01 -1.31528997e+00 -8.87341857e-01 2.33908847e-01 -7.76771069e-01 -7.29110897e-01 -4.68696445e-01 -8.92064035e-01 1.62996426e-01 -1.65055662e-01 1.29723728e+00 3.41513865e-02 5.04129753e-02 1.61449417e-01 -2.26343647e-01 -2.28528589e-01 -7.06698835e-01 6.73138499e-01 1.65448114e-01 2.83946116e-02 4.18538809e-01 -5.36974967e-01 -7.83268362e-02 5.67087173e-01 -3.86627495e-01 -3.02805662e-01 5.46614587e-01 1.08087170e+00 3.47165465e-01 -2.39978746e-01 8.14648449e-01 -8.92839193e-01 4.82920915e-01 -7.14978054e-02 -7.34951735e-01 6.02862060e-01 -6.60658777e-01 2.29826510e-01 5.05658507e-01 -3.43460053e-01 -1.26701152e+00 -1.18431956e-01 -5.49914241e-01 -5.90113401e-01 -1.78263962e-01 1.79184154e-01 -5.45404136e-01 -3.76148224e-02 6.10279977e-01 1.79103568e-01 -2.56590456e-01 -7.27238774e-01 4.84670728e-01 8.11309516e-01 5.84559619e-01 -5.63226879e-01 1.01309586e+00 -5.82579337e-02 -7.92456686e-01 -8.14863205e-01 -7.97071517e-01 -3.46440345e-01 -6.85795605e-01 1.73670009e-01 7.39065647e-01 -1.09648681e+00 -7.18808115e-01 5.92136621e-01 -9.11691844e-01 -5.95383406e-01 -4.44310158e-01 3.16535234e-01 -2.38195539e-01 1.78751752e-01 -8.72665703e-01 -4.50239480e-01 -7.38520384e-01 -1.24025261e+00 1.18563569e+00 -9.49038491e-02 -5.59417047e-02 -9.28558648e-01 3.47599417e-01 9.31973577e-01 6.98199332e-01 -5.14714420e-01 7.01332808e-01 -9.13081050e-01 -4.50828165e-01 -5.51061071e-02 -7.77128413e-02 5.21120131e-01 1.17061414e-01 -6.59888685e-02 -1.73483539e+00 -6.12288415e-01 -1.77626148e-01 -6.23709738e-01 7.87895381e-01 1.82983190e-01 9.95452583e-01 -2.94804037e-01 -3.34702998e-01 8.65741134e-01 9.24523115e-01 -2.56236047e-01 5.36722839e-01 2.31961712e-01 4.68141556e-01 2.82120138e-01 2.82964259e-01 1.24171466e-01 5.42760611e-01 7.79953897e-01 6.37207776e-02 4.28771228e-02 -3.09025168e-01 -5.90393662e-01 5.97106457e-01 9.77330387e-01 -4.21120450e-02 -1.26896575e-01 -6.88799202e-01 5.15501738e-01 -1.35450470e+00 -1.10598230e+00 1.49870470e-01 2.29579163e+00 1.01231813e+00 2.60862429e-02 2.33753696e-01 2.23573118e-01 7.71629333e-01 7.49133751e-02 -4.74275291e-01 -4.30981278e-01 -3.45542401e-01 4.91905808e-01 2.32337564e-01 7.60707855e-01 -1.11889338e+00 1.36936462e+00 6.87216473e+00 8.17907274e-01 -1.36464262e+00 6.13792062e-01 5.60258389e-01 -3.60378176e-02 -9.94851142e-02 -9.92259085e-02 -1.32772756e+00 2.49802053e-01 1.61990571e+00 -1.34367287e-01 3.61638248e-01 8.77735913e-01 5.53878536e-03 6.19032919e-01 -1.41595423e+00 9.43026423e-01 1.66075051e-01 -1.17422378e+00 -6.95853233e-02 -1.06034195e-02 6.12716675e-01 5.19168794e-01 2.43698850e-01 1.09227228e+00 5.73202133e-01 -1.10475469e+00 9.76875842e-01 -2.62169987e-01 1.06430101e+00 -5.46169639e-01 7.78730452e-01 2.70172805e-01 -1.21920276e+00 -1.51724830e-01 -3.53644401e-01 3.78315061e-01 3.09729367e-01 4.23979253e-01 -1.25865924e+00 4.72579032e-01 9.70505297e-01 4.35570002e-01 -7.48696506e-01 9.09815073e-01 -3.35552990e-01 9.27499950e-01 -2.91361809e-01 4.52656060e-01 2.36892588e-02 2.50449181e-01 2.17121542e-01 1.33772063e+00 2.73921788e-01 -3.39324653e-01 -2.78981537e-01 4.92560089e-01 -4.76074129e-01 -5.65480627e-02 -1.59491763e-01 1.08936438e-02 5.96456468e-01 1.32792604e+00 8.17778483e-02 -5.06257415e-01 -3.76094997e-01 8.37986827e-01 7.08433986e-01 3.29019785e-01 -1.02289069e+00 -8.85549262e-02 8.41686726e-01 1.28947683e-02 6.32837772e-01 1.90675586e-01 -1.05186023e-01 -1.32977414e+00 -1.91808581e-01 -1.28853548e+00 7.75676906e-01 -5.27863801e-01 -1.36919582e+00 1.19814777e+00 -1.89728618e-01 -9.70212221e-01 -2.90170997e-01 -4.59381878e-01 -7.56183326e-01 9.73209381e-01 -1.65412378e+00 -1.56012416e+00 -2.28255942e-01 1.09123278e+00 6.28062665e-01 -4.53600734e-01 9.38587725e-01 6.31076217e-01 -7.82108307e-01 1.44533062e+00 1.83000583e-02 2.72886723e-01 1.11348414e+00 -1.10146618e+00 4.95535284e-01 7.78634191e-01 2.92715371e-01 6.31399214e-01 4.91592944e-01 -2.12989047e-01 -1.12355745e+00 -1.11594987e+00 1.28342855e+00 -7.39999294e-01 6.23971581e-01 -7.86308825e-01 -9.18703079e-01 1.17100537e+00 6.57538176e-01 8.38617608e-02 9.32226896e-01 6.36129200e-01 -9.59438622e-01 -5.36209881e-01 -1.32689726e+00 8.03365111e-02 1.08870351e+00 -9.55834746e-01 -8.02166045e-01 2.29716256e-01 7.21993804e-01 -4.54606414e-01 -1.12959874e+00 3.84578586e-01 4.42304611e-01 -9.88060594e-01 9.22944188e-01 -7.13290393e-01 -3.64149362e-02 2.04565804e-02 -2.55794525e-01 -1.67366576e+00 -3.10808212e-01 -7.42924631e-01 2.57844627e-01 1.85994172e+00 9.21400011e-01 -7.53262162e-01 7.08133876e-01 3.40911031e-01 -3.35978657e-01 -2.35119477e-01 -1.38924563e+00 -8.57509792e-01 4.50060159e-01 -3.90964895e-01 1.02954590e+00 1.06327164e+00 7.74077177e-02 8.58465910e-01 -2.96949685e-01 2.18029976e-01 5.87820530e-01 1.33359715e-01 1.02293432e+00 -1.15265751e+00 -2.79938728e-01 -4.00101691e-01 -2.80045241e-01 -1.04551733e+00 4.36811447e-01 -1.07632232e+00 1.45886064e-01 -1.01210177e+00 2.57124484e-01 -6.20271623e-01 -2.74605989e-01 7.65079498e-01 -1.43467918e-01 2.43830264e-01 -4.68232762e-03 -3.18733579e-03 -3.14614475e-01 7.94370294e-01 8.78981411e-01 -4.60613102e-01 -6.76851720e-02 -9.16491076e-03 -6.60730004e-01 1.69740006e-01 6.16735637e-01 -3.73705745e-01 -4.39842999e-01 -6.11272991e-01 -5.09762883e-01 -1.23837300e-01 7.29641467e-02 -8.84652019e-01 2.83579469e-01 4.46013480e-01 2.51241952e-01 -5.49223483e-01 4.78990406e-01 -5.58640420e-01 1.13746949e-01 1.12244323e-01 -3.29034239e-01 1.79346517e-01 5.04974663e-01 2.21466601e-01 -4.73389775e-01 1.60376981e-01 9.25054014e-01 2.31969789e-01 -4.97108608e-01 3.89188379e-01 -2.53725052e-01 2.74029493e-01 7.32456923e-01 -4.00558859e-02 -5.09041250e-01 -3.22599441e-01 -7.70764828e-01 1.83126718e-01 -5.70655568e-03 6.87989354e-01 4.27736133e-01 -1.20895934e+00 -9.53585446e-01 5.92097342e-01 2.37567514e-01 -2.95671016e-01 6.07981741e-01 7.77326882e-01 6.98807985e-02 7.79472947e-01 -6.83919340e-02 -7.99229562e-01 -1.61182308e+00 5.89768350e-01 4.63338464e-01 -5.92550933e-01 -3.84503931e-01 1.29231036e+00 3.58679801e-01 -1.15051794e+00 4.02547002e-01 -5.00264704e-01 8.25060531e-02 -4.64380495e-02 4.44223791e-01 1.35457397e-01 5.26543021e-01 -9.85808492e-01 -5.76972961e-01 4.50272977e-01 -3.59792948e-01 -2.55119026e-01 1.32314801e+00 7.81476572e-02 3.06958348e-01 6.64677396e-02 1.24689484e+00 2.46492952e-01 -1.11140084e+00 -4.45894688e-01 -4.04441245e-02 -1.97465479e-01 9.92805362e-02 -9.50616300e-01 -1.30526304e+00 9.91728544e-01 6.40839517e-01 -8.38745758e-02 9.38837111e-01 2.08414987e-01 8.13741088e-01 2.36530185e-01 2.98942417e-01 -1.01354456e+00 -1.86434835e-01 5.71264207e-01 1.06975484e+00 -1.24995840e+00 -4.25034761e-01 -3.49496812e-01 -7.33839631e-01 7.12818623e-01 8.57246161e-01 4.21441734e-01 8.50685835e-01 3.44729006e-01 4.20629799e-01 6.89387321e-02 -9.19230938e-01 3.12939659e-02 2.84818739e-01 5.59331119e-01 6.37171030e-01 1.72502398e-01 4.09266770e-01 7.63236463e-01 -8.91500413e-01 -1.26489580e-01 -1.81073751e-02 5.54022908e-01 2.06135456e-02 -1.31960142e+00 -3.80564302e-01 1.96351662e-01 -3.65798950e-01 -3.11879873e-01 -4.77422416e-01 8.29425097e-01 9.86689609e-03 1.11867571e+00 -2.26138264e-01 -5.94847202e-01 5.79734325e-01 3.22097749e-01 4.24367636e-01 -5.16774893e-01 -1.15787446e+00 -2.68274210e-02 3.92852336e-01 -4.03153509e-01 -1.02318175e-01 -7.41025448e-01 -7.22371519e-01 -5.00018299e-01 -5.13869166e-01 5.76438189e-01 4.54996020e-01 9.77081358e-01 4.48297381e-01 4.58324254e-01 4.48191136e-01 -5.55889249e-01 -1.03201997e+00 -1.47651827e+00 -4.10143822e-01 1.56244412e-01 5.46441793e-01 -6.18946135e-01 -5.96698403e-01 6.10986128e-02]
[14.226432800292969, 6.408614635467529]
6d31b5d3-b456-4c65-afb6-44d48f6e17b0
learning-deep-representations-for-scene
1706.02493
null
http://arxiv.org/abs/1706.02493v2
http://arxiv.org/pdf/1706.02493v2.pdf
Learning Deep Representations for Scene Labeling with Semantic Context Guided Supervision
Scene labeling is a challenging classification problem where each input image requires a pixel-level prediction map. Recently, deep-learning-based methods have shown their effectiveness on solving this problem. However, we argue that the large intra-class variation provides ambiguous training information and hinders the deep models' ability to learn more discriminative deep feature representations. Unlike existing methods that mainly utilize semantic context for regularizing or smoothing the prediction map, we design novel supervisions from semantic context for learning better deep feature representations. Two types of semantic context, scene names of images and label map statistics of image patches, are exploited to create label hierarchies between the original classes and newly created subclasses as the learning supervisions. Such subclasses show lower intra-class variation, and help CNN detect more meaningful visual patterns and learn more effective deep features. Novel training strategies and network structure that take advantages of such label hierarchies are introduced. Our proposed method is evaluated extensively on four popular datasets, Stanford Background (8 classes), SIFTFlow (33 classes), Barcelona (170 classes) and LM+Sun datasets (232 classes) with 3 different networks structures, and show state-of-the-art performance. The experiments show that our proposed method makes deep models learn more discriminative feature representations without increasing model size or complexity.
['Zhe Wang', 'Hongsheng Li', 'Wanli Ouyang', 'Xiaogang Wang']
2017-06-08
null
null
null
null
['scene-labeling']
['computer-vision']
[ 2.28937984e-01 -2.37094373e-01 -4.86495435e-01 -9.94242668e-01 -2.28713900e-01 -3.86768937e-01 4.42646742e-01 1.25413343e-01 -5.33067167e-01 5.49627364e-01 1.53057510e-02 -4.66006286e-02 -6.91233426e-02 -8.55628252e-01 -7.24993289e-01 -7.53563583e-01 -9.30304676e-02 6.06248565e-02 7.27648377e-01 1.01173170e-01 3.32539558e-01 6.21061742e-01 -1.82755542e+00 6.23687983e-01 8.30378115e-01 1.44588017e+00 4.61743802e-01 1.25738844e-01 -5.47599852e-01 1.06389260e+00 -4.66300905e-01 -4.72614281e-02 4.45496291e-01 -8.86180475e-02 -8.66132677e-01 2.27214515e-01 8.99589181e-01 -1.12172700e-01 -4.11580950e-01 1.25576234e+00 2.91036576e-01 1.96994290e-01 7.18837559e-01 -1.19498456e+00 -8.04639518e-01 4.30522293e-01 -6.19115114e-01 3.06274593e-01 -2.98672080e-01 2.08652571e-01 1.00817466e+00 -7.20329523e-01 5.58528841e-01 1.26342344e+00 8.05327177e-01 4.67243195e-01 -1.09022844e+00 -7.89074481e-01 6.66186094e-01 6.47969842e-01 -1.40820408e+00 -1.60140514e-01 9.42700028e-01 -5.41178346e-01 7.04608798e-01 2.07537949e-01 5.38171709e-01 1.05219424e+00 -6.15796261e-03 1.04308963e+00 1.23860753e+00 -2.06898764e-01 1.33812323e-01 4.07995999e-01 4.99448687e-01 8.10667276e-01 2.13601351e-01 -8.11049417e-02 -2.90660352e-01 2.53607392e-01 7.37529814e-01 4.44395036e-01 -1.56072661e-01 -5.21451354e-01 -8.32727134e-01 9.44624901e-01 1.00175655e+00 3.08773488e-01 -7.52282143e-02 1.34141162e-01 4.94904965e-01 -7.28337020e-02 3.13325763e-01 1.69248834e-01 -7.50283062e-01 3.74093622e-01 -6.61662638e-01 -5.92392944e-02 3.35371137e-01 1.09109628e+00 1.41635597e+00 5.03815413e-02 -3.77202034e-01 1.03038931e+00 2.85858274e-01 3.95295203e-01 6.67900085e-01 -7.84294307e-01 3.02715868e-01 9.52275455e-01 -2.93794960e-01 -1.37706614e+00 -5.59565127e-01 -6.58862829e-01 -8.26449871e-01 8.24560300e-02 3.19043249e-01 9.32343081e-02 -1.30612051e+00 1.67701554e+00 1.75102651e-01 2.83363551e-01 -9.84827429e-03 8.50742459e-01 1.22701252e+00 6.32485986e-01 3.01029295e-01 2.15641230e-01 1.23114467e+00 -1.35484266e+00 -4.97877091e-01 -3.95084798e-01 7.79197276e-01 -6.61146581e-01 1.15672886e+00 1.95446402e-01 -4.50247407e-01 -1.07466435e+00 -9.59296465e-01 -1.59713387e-01 -6.21323049e-01 3.88287514e-01 9.38620031e-01 4.73554820e-01 -9.39061105e-01 5.54887354e-01 -5.70920944e-01 -3.47590029e-01 9.72508132e-01 2.96961069e-01 -2.54282445e-01 -3.42874289e-01 -9.71093476e-01 5.62820137e-01 5.37773192e-01 1.38270751e-01 -1.03953588e+00 -6.41313136e-01 -9.78051007e-01 1.61909416e-01 2.94818789e-01 -3.23263228e-01 8.83691370e-01 -1.49308562e+00 -1.13533115e+00 1.12392998e+00 -6.53041154e-02 -4.14817572e-01 2.02820599e-01 -1.45898059e-01 -3.12160045e-01 1.62698373e-01 3.27808022e-01 1.05505455e+00 7.75336266e-01 -1.34550238e+00 -9.38062072e-01 -2.89323002e-01 1.61269575e-01 2.10538153e-02 -5.54285049e-01 -2.38958180e-01 -4.47956264e-01 -6.32420421e-01 1.98108032e-01 -7.92369843e-01 -3.71767789e-01 1.05360471e-01 -5.45654535e-01 -4.91331369e-01 1.06415486e+00 -3.12398553e-01 9.43627179e-01 -2.23104262e+00 -3.77895355e-01 1.87981844e-01 1.22364901e-01 3.99329275e-01 -3.69019896e-01 -3.13049480e-02 -3.12845334e-02 -3.63971330e-02 -7.61440769e-02 -1.99392125e-01 -1.69645324e-01 4.75353688e-01 -1.56245291e-01 3.92157912e-01 3.90909523e-01 7.43663669e-01 -7.67117381e-01 -5.76965272e-01 3.86830568e-01 3.46172571e-01 -6.01415336e-01 1.54747799e-01 -2.79289305e-01 2.95672923e-01 -5.79398513e-01 7.80669451e-01 8.45879734e-01 -4.65427518e-01 1.14758536e-01 -5.24924695e-01 -9.47697647e-03 1.74311981e-01 -1.30592871e+00 1.65215993e+00 -5.20680070e-01 7.50663280e-01 -3.52349728e-01 -1.50899863e+00 1.21353555e+00 -4.54024762e-01 3.23900074e-01 -8.24677885e-01 1.23487152e-01 -1.78243052e-02 -8.91026333e-02 -6.70722961e-01 2.53862470e-01 2.29370505e-01 9.13156196e-02 -5.56067899e-02 2.85647452e-01 2.82353580e-01 2.23822743e-01 -6.85616881e-02 7.66070724e-01 1.21467307e-01 4.15384397e-02 -5.76617479e-01 6.83973789e-01 4.61064652e-02 1.09957612e+00 8.61183465e-01 -3.60413671e-01 4.55753863e-01 3.44467640e-01 -9.64210212e-01 -6.58000648e-01 -8.64208698e-01 -4.95786220e-01 1.31803346e+00 5.78927875e-01 -4.55969900e-01 -5.90899587e-01 -1.08525634e+00 1.16644077e-01 5.33683181e-01 -8.20412934e-01 -2.32661396e-01 -4.63051736e-01 -7.74555206e-01 1.99294627e-01 7.75224864e-01 9.63266253e-01 -1.25337481e+00 -5.06112695e-01 -9.66435894e-02 -4.97979438e-03 -1.19262862e+00 -1.73585758e-01 4.68981892e-01 -9.66122270e-01 -1.24828720e+00 -3.29103321e-01 -1.12367833e+00 8.91373754e-01 5.39086640e-01 1.15037227e+00 1.76078677e-01 -5.88207841e-01 2.12582678e-01 -3.65000874e-01 -2.47946605e-01 8.63344222e-02 1.46227973e-02 -1.90169811e-01 2.23001212e-01 6.01861238e-01 -4.33166027e-01 -7.73005903e-01 4.74782854e-01 -8.50071311e-01 1.97800666e-01 7.42846847e-01 1.00904262e+00 8.18759620e-01 1.48476884e-01 5.03121614e-01 -1.13114095e+00 -1.18322141e-01 -3.83958668e-01 -6.32990181e-01 1.83395267e-01 -4.54879165e-01 1.53233916e-01 7.17777312e-01 -3.02741885e-01 -1.17633343e+00 3.44992429e-01 -1.59250014e-02 -3.14142048e-01 -5.92517614e-01 2.31268629e-01 -3.38470936e-01 -2.06900164e-01 6.50240064e-01 2.08145604e-01 -2.61772454e-01 -6.41758859e-01 3.83722246e-01 4.98400241e-01 4.08903331e-01 -6.08746767e-01 6.08823240e-01 6.33367896e-01 -4.44224067e-02 -7.05089748e-01 -1.28886449e+00 -6.62068605e-01 -8.72120142e-01 4.91178632e-02 9.72201586e-01 -1.04357302e+00 -3.70950907e-01 5.00228941e-01 -9.75556612e-01 -3.00146520e-01 -3.14831734e-01 2.86875635e-01 -3.71534735e-01 1.84559196e-01 -5.10013878e-01 -3.26611727e-01 2.29243357e-02 -1.36911893e+00 1.05616295e+00 7.04922616e-01 2.67407805e-01 -1.03740728e+00 -3.22365731e-01 1.55715123e-01 3.64625484e-01 2.24304184e-01 9.54798639e-01 -6.75186574e-01 -7.97500312e-01 -3.50889340e-02 -6.79877698e-01 7.07621992e-01 3.74433070e-01 3.72387283e-02 -1.35808146e+00 -1.15275376e-01 -2.83101588e-01 -4.24871743e-01 1.25967109e+00 4.59828407e-01 1.85889971e+00 -1.91810429e-01 -5.76618910e-01 1.00694430e+00 1.56349361e+00 1.37789935e-01 5.83032429e-01 5.99551201e-01 9.72756505e-01 6.72747016e-01 6.13010108e-01 3.35358888e-01 2.93538749e-01 6.11224532e-01 6.57586873e-01 -3.93184543e-01 -3.32851052e-01 -1.78223386e-01 1.71100631e-01 6.09208524e-01 7.78105482e-02 2.00498790e-01 -7.61268735e-01 4.47811306e-01 -1.85408068e+00 -7.68117905e-01 -1.24567129e-01 1.91524494e+00 7.90995955e-01 1.94486335e-01 -6.70744106e-02 -1.86519578e-01 7.37310648e-01 2.42940158e-01 -7.09784210e-01 -9.18776020e-02 -3.21663678e-01 1.08283684e-01 6.36313796e-01 1.29224747e-01 -1.66520190e+00 1.27997065e+00 5.63317347e+00 1.07011473e+00 -1.20780730e+00 -7.28143156e-02 9.89511132e-01 1.98688462e-01 6.83706552e-02 -1.45189226e-01 -1.24703968e+00 4.44933712e-01 4.83014256e-01 3.95944774e-01 -1.23985611e-01 1.34696341e+00 -7.29625225e-02 -1.16044499e-01 -1.15384090e+00 1.17954230e+00 2.34513711e-02 -1.64329517e+00 2.56065011e-01 -2.63462782e-01 9.69390392e-01 1.40414938e-01 1.37416065e-01 5.29123247e-01 3.29610616e-01 -9.64400232e-01 6.52853608e-01 3.68008196e-01 5.87140143e-01 -6.54343784e-01 8.62915695e-01 6.68473393e-02 -1.37351823e+00 -4.36843544e-01 -9.36118364e-01 -7.37238377e-02 -2.87900418e-01 6.85488999e-01 -6.08970702e-01 4.01672482e-01 1.16720164e+00 1.35413563e+00 -1.12238395e+00 1.11309111e+00 -3.25484157e-01 6.12876415e-01 -1.49789089e-02 -1.23845795e-02 5.33579826e-01 -7.92522430e-02 -5.68717830e-02 1.35099137e+00 -9.28777754e-02 -1.99982926e-01 6.75758660e-01 8.12177479e-01 -1.23889185e-01 6.06829226e-02 -5.62330782e-01 2.93944627e-01 2.31214374e-01 1.43298542e+00 -1.09082651e+00 -4.23145086e-01 -5.98887920e-01 8.06744158e-01 4.27753866e-01 4.83124912e-01 -8.47854435e-01 -3.57678920e-01 9.70919073e-01 -1.09264022e-02 3.88318688e-01 -5.41866897e-03 -3.83457035e-01 -1.10346019e+00 -1.24263659e-01 -5.99325120e-01 5.71574986e-01 -4.63445306e-01 -1.50829816e+00 5.24237216e-01 -8.97095874e-02 -1.26582408e+00 1.57732457e-01 -9.77588654e-01 -5.44957221e-01 5.92855096e-01 -1.87280655e+00 -1.21706784e+00 -6.25986457e-01 7.05599844e-01 7.66699910e-01 -3.06642324e-01 6.52658463e-01 4.52930152e-01 -6.11028850e-01 5.90721250e-01 2.70136952e-01 4.71653521e-01 7.00666010e-01 -1.10863221e+00 -1.15224034e-01 6.50165081e-01 4.06404465e-01 3.71109784e-01 2.06726328e-01 -2.90218323e-01 -1.00189137e+00 -1.52179980e+00 4.17528659e-01 -2.58978784e-01 4.89843249e-01 -4.19266522e-01 -7.88094997e-01 6.33892000e-01 5.51621504e-02 5.97805798e-01 6.90686643e-01 2.20165297e-01 -5.80341578e-01 -5.62520504e-01 -1.00011826e+00 3.13948810e-01 1.31172442e+00 -3.54760200e-01 -3.50799948e-01 5.41208148e-01 7.68513083e-01 -9.82118919e-02 -3.81961823e-01 6.52470529e-01 2.50425905e-01 -1.12733400e+00 1.07444406e+00 -6.93368554e-01 3.68822634e-01 -4.54682797e-01 -2.89727747e-01 -1.02780676e+00 -5.02339005e-01 4.79805022e-02 3.07467192e-01 1.21575081e+00 1.77576542e-01 -4.44172442e-01 9.36719477e-01 2.42956102e-01 -3.82058352e-01 -8.95268917e-01 -5.90956748e-01 -8.34819734e-01 -8.58711898e-02 -3.47769260e-01 5.28392553e-01 1.21384311e+00 -7.72071838e-01 2.33886600e-01 -1.21971406e-01 1.89646512e-01 4.90124851e-01 4.60244268e-01 8.16481233e-01 -1.40050042e+00 5.01449872e-03 -5.72334945e-01 -8.55359375e-01 -1.11406219e+00 3.36754471e-01 -1.02339649e+00 -5.19530773e-02 -1.41930425e+00 4.00998950e-01 -9.17041659e-01 -7.06898749e-01 8.47933471e-01 -1.89758003e-01 3.03228915e-01 2.09194839e-01 9.46628675e-02 -9.34677780e-01 6.80954516e-01 1.20454812e+00 -5.24442017e-01 4.07112390e-02 -2.47695800e-02 -6.29180849e-01 1.04565680e+00 6.76627159e-01 -4.80238467e-01 -5.34016848e-01 -5.18876493e-01 -2.37363458e-01 -8.08357477e-01 5.01153409e-01 -1.28596497e+00 6.27273098e-02 -3.28675985e-01 8.59888852e-01 -6.13672256e-01 3.75191495e-02 -9.59357321e-01 -2.49352798e-01 4.28091735e-01 -4.59906399e-01 -2.98280299e-01 3.45929384e-01 5.47358394e-01 -4.33432728e-01 -3.77901763e-01 1.02670240e+00 -1.72589093e-01 -1.50487661e+00 5.27080178e-01 -1.79342041e-03 5.76212443e-02 9.44006026e-01 -3.68999004e-01 -3.57453734e-01 3.66354026e-02 -7.94144034e-01 2.07751587e-01 2.06980884e-01 5.85222423e-01 5.74476421e-01 -1.46422505e+00 -3.19480568e-01 2.94359207e-01 2.65410185e-01 2.70448834e-01 3.73364449e-01 4.80809450e-01 -4.96640116e-01 2.98049629e-01 -4.84486639e-01 -1.06917059e+00 -1.12215722e+00 6.04196310e-01 3.26711416e-01 -1.50697455e-01 -7.06497729e-01 1.16919744e+00 9.06609237e-01 -5.43113708e-01 4.43501770e-01 -5.22814274e-01 -4.97010618e-01 -9.77115054e-03 5.10706007e-01 8.90191570e-02 -1.50105461e-01 -6.32485211e-01 -3.37085694e-01 7.69646585e-01 -3.14024627e-01 7.79357791e-01 1.40848219e+00 -1.28612727e-01 -4.09199037e-02 2.56660312e-01 1.45899451e+00 -3.66979837e-01 -1.50183547e+00 -5.40051520e-01 2.24747121e-01 -6.52550340e-01 2.99067376e-03 -7.03546584e-01 -1.57061243e+00 1.16471815e+00 9.73270416e-01 1.38338208e-02 1.28125691e+00 1.94798224e-02 6.84419036e-01 5.26207864e-01 2.92460740e-01 -1.16934049e+00 4.30565149e-01 6.56981230e-01 5.59074402e-01 -1.51349092e+00 -1.02651983e-01 -5.96228242e-01 -5.05396783e-01 1.22506273e+00 1.12864196e+00 -2.57295519e-01 7.16946959e-01 -2.64298245e-02 1.94393918e-01 -1.73794836e-01 -3.21382493e-01 -4.64626372e-01 4.03059512e-01 7.60993600e-01 2.45684326e-01 4.94771376e-02 1.34489164e-01 7.27485359e-01 1.50745943e-01 -3.07744145e-01 1.57689363e-01 8.39470208e-01 -7.16211498e-01 -9.31925297e-01 -1.07977442e-01 4.43052500e-01 -2.30004877e-01 -1.25671268e-01 -1.50055930e-01 8.59958231e-01 6.00738049e-01 6.42814279e-01 2.67000139e-01 -3.98898393e-01 1.80040792e-01 -1.44803017e-01 2.59454787e-01 -7.76190341e-01 -3.58351380e-01 -1.83914542e-01 -1.56267956e-01 -8.11606109e-01 -6.96392894e-01 -3.70225281e-01 -1.25373733e+00 6.55945614e-02 -2.20311582e-01 -4.00657840e-02 4.83454853e-01 8.64556730e-01 3.56949151e-01 6.64418221e-01 7.39904225e-01 -8.73351812e-01 -3.57308507e-01 -8.79846454e-01 -5.78924596e-01 7.21642852e-01 8.85178372e-02 -1.08228886e+00 -2.37868875e-01 3.15935075e-01]
[9.594000816345215, 1.9131540060043335]
c2673258-dbb8-4c48-a9b1-422ca3fd9e4b
natgen-generative-pre-training-by
2206.07585
null
https://arxiv.org/abs/2206.07585v2
https://arxiv.org/pdf/2206.07585v2.pdf
NatGen: Generative pre-training by "Naturalizing" source code
Pre-trained Generative Language models (e.g. PLBART, CodeT5, SPT-Code) for source code yielded strong results on several tasks in the past few years, including code generation and translation. These models have adopted varying pre-training objectives to learn statistics of code construction from very large-scale corpora in a self-supervised fashion; the success of pre-trained models largely hinges on these pre-training objectives. This paper proposes a new pre-training objective, "Naturalizing" of source code, exploiting code's bimodal, dual-channel (formal & natural channels) nature. Unlike natural language, code's bimodal, dual-channel nature allows us to generate semantically equivalent code at scale. We introduce six classes of semantic preserving transformations to introduce un-natural forms of code, and then force our model to produce more natural original programs written by developers. Learning to generate equivalent, but more natural code, at scale, over large corpora of open-source code, without explicit manual supervision, helps the model learn to both ingest & generate code. We fine-tune our model in three generative Software Engineering tasks: code generation, code translation, and code refinement with limited human-curated labeled data and achieve state-of-the-art performance rivaling CodeT5. We show that our pre-trained model is especially competitive at zero-shot and few-shot learning, and better at learning code properties (e.g., syntax, data flow).
['Baishakhi Ray', 'Premkumar Devanbu', 'Yangruibo Ding', 'Toufique Ahmed', 'Saikat Chakraborty']
2022-06-15
null
null
null
null
['code-translation']
['computer-code']
[ 2.55862564e-01 3.78438354e-01 -1.48392215e-01 -2.13901550e-01 -1.11058998e+00 -7.30050862e-01 5.88269770e-01 -3.26005854e-02 1.38854429e-01 3.57810885e-01 3.27538460e-01 -5.96017599e-01 3.19812775e-01 -8.55733812e-01 -1.04028976e+00 -7.64080361e-02 -6.28078356e-02 2.50394344e-01 1.19473971e-01 -4.96823817e-01 4.06450868e-01 -2.83264548e-01 -1.59640586e+00 6.52496517e-01 1.10727763e+00 1.02386087e-01 5.61086118e-01 1.05019546e+00 -4.93311018e-01 1.15818381e+00 -4.00393754e-01 -5.28022528e-01 1.89243883e-01 -8.67599964e-01 -1.00718689e+00 -1.25771716e-01 -6.69611897e-03 4.01703008e-02 1.00259557e-01 1.19864821e+00 2.34029859e-01 -3.67821574e-01 6.07025802e-01 -1.25338900e+00 -1.39175570e+00 1.19415820e+00 -3.83450478e-01 -1.20413862e-01 4.83841479e-01 4.33809876e-01 1.28427458e+00 -1.00571680e+00 8.42894018e-01 1.02956223e+00 9.77755070e-01 9.87696230e-01 -1.82698691e+00 -5.98299503e-01 -4.79187399e-01 -4.75013793e-01 -1.24498951e+00 -4.37994957e-01 5.77454984e-01 -9.85437155e-01 1.29183745e+00 -1.20763704e-02 4.62400258e-01 1.26392221e+00 5.06382227e-01 5.55711210e-01 8.07245731e-01 -5.82193136e-01 2.91460812e-01 2.80099005e-01 -2.22928733e-01 9.82310355e-01 5.05064242e-02 1.13650255e-01 -2.02596098e-01 -5.47338367e-01 6.03670180e-01 -1.09379724e-01 -7.05701560e-02 -6.61576390e-01 -1.19992924e+00 1.14954114e+00 2.66619653e-01 4.35314745e-01 1.76183552e-01 4.80473638e-01 5.03495932e-01 5.49867213e-01 2.05186203e-01 8.78590703e-01 -5.85293233e-01 -6.67336702e-01 -1.06795990e+00 3.16137820e-01 8.69271100e-01 1.71174169e+00 1.23171914e+00 3.80978703e-01 -2.91668206e-01 8.64770234e-01 3.00913244e-01 4.88085628e-01 8.24017704e-01 -7.77542651e-01 6.50450528e-01 6.88228488e-01 -3.59609216e-01 -4.91933197e-01 -2.13295054e-02 -2.99766302e-01 -5.93609154e-01 3.48576039e-01 3.69538777e-02 -3.29461902e-01 -9.60875690e-01 1.75990903e+00 -4.64223713e-01 -3.17252517e-01 3.11944455e-01 1.36724889e-01 5.40310442e-01 6.80128872e-01 -1.18925929e-01 1.56234264e-01 9.09092903e-01 -1.03534734e+00 -4.98422384e-02 -4.78899330e-01 1.04817331e+00 -9.04908001e-01 1.52502310e+00 -3.01335985e-03 -1.00399196e+00 -7.31507719e-01 -7.79695570e-01 -4.55233306e-02 -1.74359843e-01 4.68322448e-03 7.38593638e-01 7.16879666e-01 -1.42862761e+00 4.93642569e-01 -8.19458425e-01 -2.71551222e-01 5.70100784e-01 6.08464703e-02 -1.41624331e-01 1.13759830e-03 -7.65779078e-01 6.34140253e-01 2.47744262e-01 -7.54653752e-01 -1.37461436e+00 -1.01873338e+00 -1.29392695e+00 1.44218385e-01 4.88508940e-02 -7.75135517e-01 1.56037259e+00 -1.28044045e+00 -1.33638155e+00 9.35523510e-01 -3.41173797e-03 -4.85430807e-01 2.20406100e-01 9.94140729e-02 -2.37425566e-01 -5.50977170e-01 4.98501360e-01 7.52753019e-01 8.72947454e-01 -1.45020664e+00 -3.84764075e-01 2.07818314e-01 2.01075286e-01 -4.45009917e-01 -3.64043295e-01 1.57879844e-01 -2.53262576e-02 -8.42160285e-01 -6.06525421e-01 -1.00910246e+00 -2.86385119e-01 -3.46494526e-01 -2.67346591e-01 -1.25419572e-01 5.38449883e-01 -5.01806200e-01 1.22498918e+00 -2.22988319e+00 3.28862071e-01 -1.74467231e-03 3.54058921e-01 -1.51385143e-02 -4.85528380e-01 5.78532100e-01 -1.95585191e-01 4.73951042e-01 -7.34176517e-01 -1.43199638e-01 1.06390908e-01 2.44734719e-01 -3.54778022e-01 -2.53872983e-02 5.68684459e-01 1.33313227e+00 -1.12886477e+00 -4.79356110e-01 -2.70378053e-01 1.94272876e-01 -1.33943129e+00 4.04491395e-01 -5.67383289e-01 1.37539208e-01 -2.99986541e-01 4.98675883e-01 7.52458349e-02 -3.88130277e-01 -9.09978673e-02 4.96856749e-01 -1.85061485e-01 1.06161207e-01 -5.48580587e-01 2.28657079e+00 -1.21792865e+00 8.49964142e-01 -3.73021305e-01 -7.12921023e-01 1.05398178e+00 2.36312717e-01 4.81612757e-02 -5.62400281e-01 -6.49856627e-02 3.54221791e-01 2.35701710e-01 -7.80609429e-01 3.97496432e-01 -8.15880597e-02 -4.97653127e-01 7.54732311e-01 5.00231802e-01 -6.34281874e-01 3.81239712e-01 4.70075220e-01 1.42987502e+00 4.58913177e-01 3.45174015e-01 -5.04434705e-01 1.17659800e-01 2.72202849e-01 2.33062789e-01 8.48434329e-01 3.55641663e-01 8.02865148e-01 7.93754756e-01 -2.23947957e-01 -1.50477052e+00 -1.05618322e+00 8.22517052e-02 1.47538126e+00 -3.37305248e-01 -8.25347483e-01 -1.05076587e+00 -7.44645119e-01 -1.10416114e-01 1.01908946e+00 -7.92805851e-01 -5.05875409e-01 -5.20596147e-01 -6.35992527e-01 7.74645567e-01 6.51481450e-01 7.73766115e-02 -1.16700220e+00 -6.62340522e-01 3.57918024e-01 4.42718416e-02 -5.58499992e-01 -7.47007966e-01 4.68099952e-01 -7.18534887e-01 -8.94570827e-01 -6.12167358e-01 -1.00268030e+00 8.22899818e-01 -1.37826055e-01 1.64130592e+00 3.20953906e-01 -4.43137318e-01 2.15478584e-01 -6.72420800e-01 -2.68585652e-01 -1.43652642e+00 2.91821599e-01 -5.16622961e-01 -5.15231550e-01 1.50884762e-01 -7.52572238e-01 3.78391407e-02 -9.78564024e-02 -1.13683617e+00 2.43627012e-01 8.77269387e-01 1.14804733e+00 1.17433192e-02 -2.57375211e-01 3.25800419e-01 -1.31134176e+00 8.34174156e-01 -6.94689572e-01 -5.17804086e-01 2.26009935e-01 -7.56205440e-01 6.85976386e-01 1.00641358e+00 -4.46704328e-01 -1.05439270e+00 2.86939479e-02 -2.92283803e-01 -1.92835584e-01 1.35911023e-02 5.98494351e-01 3.00397038e-01 -1.11518770e-01 1.44496310e+00 4.51870888e-01 -9.92609337e-02 -2.81827748e-01 6.53514206e-01 5.87406456e-01 5.10605276e-01 -1.07450354e+00 1.30170059e+00 -3.79980877e-02 -5.90442777e-01 -3.11424851e-01 -3.20567191e-01 7.07940906e-02 -5.54544806e-01 3.35159600e-01 9.48148251e-01 -8.98791552e-01 1.10312469e-01 1.97048753e-01 -1.29205608e+00 -6.90630317e-01 -5.29410779e-01 -9.75263044e-02 -7.89029717e-01 1.52432751e-02 -5.94322562e-01 -4.12618309e-01 -2.43981659e-01 -1.37405312e+00 1.20672882e+00 -9.04484168e-02 -5.98651350e-01 -1.05024230e+00 5.35069704e-01 -8.54671225e-02 8.79614532e-01 1.63631245e-01 1.53683138e+00 -5.31635642e-01 -6.11709893e-01 2.17571989e-01 -1.34240791e-01 4.49962050e-01 2.97378361e-01 2.70446181e-01 -8.78345668e-01 -2.31271625e-01 -2.05076873e-01 -5.63850105e-01 6.50501311e-01 -9.76295546e-02 9.65838373e-01 -4.14820701e-01 -1.09538749e-01 8.64392877e-01 1.58633184e+00 2.38739505e-01 6.41643763e-01 1.89912111e-01 7.82646716e-01 2.16747656e-01 2.49833949e-02 5.14066756e-01 4.64749902e-01 3.42304617e-01 3.06667775e-01 -6.80032670e-02 -2.42699370e-01 -4.83604193e-01 6.57850146e-01 1.04237497e+00 1.13008851e-02 1.43832311e-01 -1.34127629e+00 8.03030670e-01 -1.58807337e+00 -1.00887680e+00 -1.60948746e-03 1.78211391e+00 1.41015661e+00 2.16018975e-01 4.13621143e-02 -3.27368706e-01 4.28467780e-01 -1.83352530e-01 -3.88827711e-01 -5.30660748e-01 2.11002395e-01 6.05787694e-01 3.43815833e-01 3.02281111e-01 -6.93881691e-01 9.85859394e-01 6.50441599e+00 8.32408905e-01 -9.39455211e-01 3.72466564e-01 4.53694254e-01 3.66927117e-01 -9.70064640e-01 4.06834424e-01 -4.49853033e-01 5.64517677e-01 1.08236718e+00 -6.59032941e-01 8.91561866e-01 1.43135440e+00 -2.48543039e-01 2.79795736e-01 -1.53554416e+00 7.70461142e-01 1.61858350e-01 -1.65150142e+00 4.22359481e-02 -1.11307129e-01 1.31002688e+00 2.96102077e-01 -2.69485638e-02 8.87891769e-01 1.06249118e+00 -1.12765586e+00 1.13078082e+00 2.59296894e-01 1.11877775e+00 -6.03643775e-01 5.43272853e-01 3.56920332e-01 -1.25303018e+00 -2.85822541e-01 -3.31966490e-01 -9.76497233e-02 -2.40856722e-01 4.25879836e-01 -7.62941182e-01 2.74098307e-01 5.03903985e-01 8.31457555e-01 -1.09866655e+00 5.56246161e-01 -2.07321897e-01 4.73762155e-01 3.69357824e-01 -1.33336410e-01 1.66932791e-01 3.10688585e-01 1.90888539e-01 1.54298389e+00 6.05491459e-01 -3.43622655e-01 2.12552935e-01 1.81872475e+00 -3.09878141e-01 -7.34760836e-02 -9.24452841e-01 -4.30819333e-01 2.08677888e-01 1.11966932e+00 -5.67767441e-01 -4.94581223e-01 -7.50762641e-01 8.55807245e-01 3.32146913e-01 3.41237009e-01 -9.35155213e-01 -7.90571690e-01 5.54325998e-01 6.93781748e-02 2.11062580e-01 -1.53483301e-01 -4.14200485e-01 -1.30185235e+00 -3.97200100e-02 -1.33135211e+00 -2.05403790e-01 -7.33586907e-01 -1.08029699e+00 9.10779953e-01 -7.75139034e-02 -1.20055556e+00 -7.87679136e-01 -3.83986533e-01 -8.31335664e-01 8.54274571e-01 -1.12437391e+00 -1.08391118e+00 -1.98061496e-01 3.34259123e-01 8.68385255e-01 -6.61937714e-01 9.32603002e-01 1.18801825e-01 -7.12906569e-02 7.06397831e-01 6.55978099e-02 4.43891048e-01 4.68044043e-01 -1.52583098e+00 1.09457684e+00 1.03190279e+00 2.82113373e-01 1.01006544e+00 6.88112617e-01 -6.50152802e-01 -1.50556397e+00 -1.40378630e+00 5.64698219e-01 -8.77564371e-01 9.28181589e-01 -7.45874822e-01 -7.82829225e-01 7.91739106e-01 3.59341711e-01 7.21790420e-04 5.47906280e-01 -2.21183170e-02 -8.45361352e-01 3.40059727e-01 -8.78043354e-01 5.28892279e-01 1.21624053e+00 -9.50757802e-01 -7.40513384e-01 2.66835868e-01 1.03998017e+00 -3.28867733e-01 -7.82030165e-01 3.01501229e-02 2.83886671e-01 -1.01988375e+00 7.31896996e-01 -6.58172190e-01 1.30220246e+00 -1.17534511e-01 -2.92118415e-02 -1.62932265e+00 -4.74001914e-01 -8.93887579e-01 1.32653534e-01 1.35404074e+00 8.12356114e-01 -3.33494663e-01 5.32480359e-01 3.60385925e-01 -4.90014583e-01 -5.39116681e-01 -2.49787122e-01 -8.91610146e-01 3.84258658e-01 -4.20611799e-01 6.37657881e-01 9.87522721e-01 2.76362270e-01 3.99249732e-01 -1.74971968e-01 -4.17636633e-01 3.10016364e-01 1.49288058e-01 1.05682075e+00 -1.03025913e+00 -8.57159376e-01 -6.48665428e-01 -2.59087205e-01 -7.50641942e-01 3.81177455e-01 -1.45239961e+00 4.09954607e-01 -1.26936507e+00 6.23937070e-01 -3.29704434e-01 2.99960256e-01 7.56929278e-01 -1.27700984e-01 1.25115225e-02 4.76522371e-02 2.24119097e-01 -3.16251367e-01 4.05560046e-01 9.48867679e-01 -3.09666723e-01 -2.38302067e-01 -1.71640053e-01 -9.72010970e-01 4.40991372e-01 5.70521712e-01 -8.67808819e-01 -5.77479601e-01 -7.84659088e-01 5.95687151e-01 1.03136050e-02 3.07645261e-01 -9.98151422e-01 -4.32722867e-02 -1.34635895e-01 -3.42357308e-02 3.01535308e-01 -4.15676087e-01 -4.05966401e-01 1.82762563e-01 5.89623332e-01 -5.90932131e-01 3.46662521e-01 2.55003214e-01 3.31479728e-01 -1.35278493e-01 -6.22188330e-01 9.23838377e-01 -5.67421317e-01 -7.33206749e-01 3.70131806e-02 -4.27873194e-01 7.15751886e-01 8.41508865e-01 -1.09895445e-01 -4.11789894e-01 -1.01159655e-01 -3.23105514e-01 -1.21722102e-01 9.02036905e-01 9.01547313e-01 4.00495082e-01 -1.25936615e+00 -7.76840508e-01 4.25745487e-01 6.62562490e-01 -1.32398486e-01 -2.26004601e-01 4.32632744e-01 -6.51180267e-01 1.00692116e-01 -2.79027492e-01 -6.33046806e-01 -7.89738059e-01 6.67671561e-01 9.45149437e-02 -2.24039197e-01 -4.60637748e-01 8.85865033e-01 2.91784436e-01 -6.83185875e-01 -3.07024479e-01 -7.05314517e-01 4.00583208e-01 -4.16619807e-01 3.33934367e-01 -5.10168187e-02 1.97603973e-03 -2.30403200e-01 -2.70413943e-02 4.93519217e-01 -4.62050177e-03 1.04922324e-01 1.57171357e+00 4.04022664e-01 -3.65610808e-01 3.90304714e-01 1.42116976e+00 1.04869366e-01 -1.14749300e+00 -1.26326695e-01 2.97943860e-01 -4.52238202e-01 -3.07790428e-01 -5.86008430e-01 -8.48369002e-01 8.82084668e-01 1.84251472e-01 3.62282366e-01 7.66137481e-01 3.10119689e-01 6.20597482e-01 4.63549137e-01 7.88996696e-01 -8.38904202e-01 4.48682755e-01 7.51759529e-01 6.72545433e-01 -1.24398327e+00 -6.06289625e-01 -3.98571193e-02 -5.76557159e-01 1.27853096e+00 7.27670491e-01 -1.49549350e-01 3.41603458e-01 8.58993232e-01 -8.82263333e-02 -1.14061259e-01 -9.07253683e-01 -1.45556899e-02 1.13610663e-01 7.93952525e-01 8.50142360e-01 -1.74373195e-01 1.83681920e-01 4.47332442e-01 -4.83386457e-01 -2.23536510e-02 8.56036901e-01 1.12322736e+00 -4.77390707e-01 -1.31232035e+00 -2.16438472e-01 6.28487587e-01 -1.23524383e-01 -7.41162300e-01 -2.56464481e-01 6.78014100e-01 3.62487108e-01 6.13926530e-01 -1.84089169e-01 -4.76646364e-01 1.02305084e-01 2.71815956e-01 3.31215143e-01 -1.49352086e+00 -7.41824985e-01 -2.54802018e-01 -1.95364758e-01 -4.08820122e-01 -1.66371837e-01 -5.23498356e-01 -1.19918120e+00 -1.90902188e-01 -1.17209792e-01 2.41037950e-01 3.55375707e-01 7.75596857e-01 5.26147783e-01 8.59306395e-01 5.32366872e-01 -7.84859776e-01 -5.88389277e-01 -7.99629629e-01 -1.54678464e-01 5.40531337e-01 3.89197856e-01 -2.76850939e-01 -2.58094400e-01 8.64590704e-01]
[7.767595291137695, 7.859583854675293]
53418f85-4e7b-42c1-afdd-a2e62064683a
robust-target-localization-in-2d-a-value-at
2307.00548
null
https://arxiv.org/abs/2307.00548v2
https://arxiv.org/pdf/2307.00548v2.pdf
Robust Target Localization in 2D: A Value-at-Risk Approach
This paper consider considers the problem of locating a two dimensional target from range-measurements containing outliers. Assuming that the number of outlier is known, we formulate the problem of minimizing inlier losses while ignoring outliers. This leads to a combinatorial, non-convex, non-smooth problem involving the percentile function. Using the framework of risk analysis from Rockafellar et al., we start by interpreting this formulation as a Value-at-risk (VaR) problem from portfolio optimization. To the best of our knowledge, this is the first time that a localization problem was formulated using risk analysis theory. To study the VaR formulation, we start by designing a majorizer set that contains any solution of a general percentile problem. This set is useful because, when applied to a localization scenario in 2D, it allows to majorize the solution set in terms of singletons, circumferences, ellipses and hyperbolas. Using know parametrization of these curves, we propose a grid method for the original non-convex problem. So we reduce the task of optimizing the VaR objective to that of efficiently sampling the proposed majorizer set. We compare our algorithm with four benchmarks in target localization. Numerical simulations show that our method is fast while, on average, improving the accuracy of the best benchmarks by at least 100m in a 1 Km$^2$ area.
['João Xavier', 'João Domingos']
2023-07-02
null
null
null
null
['portfolio-optimization']
['time-series']
[-1.87292382e-01 2.01344609e-01 2.56132901e-01 4.59827147e-02 -1.32454967e+00 -7.66386509e-01 3.22708756e-01 3.88544470e-01 -3.52077991e-01 6.73507750e-01 -1.20302297e-01 -3.38261485e-01 -5.12156427e-01 -6.59507513e-01 -9.78706837e-01 -1.00674725e+00 -4.43588823e-01 5.32351077e-01 -3.49990204e-02 6.57080561e-02 4.47326124e-01 9.83369708e-01 -1.10058808e+00 -6.81749582e-01 8.88728678e-01 1.35298479e+00 -2.44396880e-01 4.55721527e-01 3.71394277e-01 -8.75105038e-02 -9.67271924e-01 -3.52525890e-01 7.69103289e-01 -1.08692758e-01 -1.82744592e-01 4.46741916e-02 4.40960705e-01 8.74956623e-02 2.99810678e-01 1.34978652e+00 8.14442337e-01 1.34681866e-01 9.20336962e-01 -1.29438043e+00 9.57823396e-02 2.16635033e-01 -9.53983963e-01 9.82782245e-02 3.32778603e-01 -4.00015652e-01 6.33987188e-01 -1.07806623e+00 1.59072906e-01 7.58842349e-01 1.14189482e+00 -1.81334242e-01 -1.21348548e+00 -3.20098370e-01 -1.26290575e-01 -4.19950187e-01 -1.93172562e+00 -2.88906664e-01 5.05834818e-01 -6.29224539e-01 3.18863153e-01 4.05196756e-01 4.21043634e-01 5.90281427e-01 2.63957858e-01 1.98068663e-01 8.87759805e-01 -4.93505597e-01 3.99293393e-01 6.76054135e-02 1.82273462e-02 4.80372548e-01 9.52528358e-01 1.48539349e-01 1.85848698e-02 -6.20786190e-01 3.80481482e-01 -9.65239182e-02 -3.96200567e-01 -7.64250815e-01 -1.03315401e+00 9.22422886e-01 2.95371115e-01 1.66012287e-01 -3.96663725e-01 9.88543332e-02 -3.65511067e-02 1.49453104e-01 7.86872864e-01 5.33033371e-01 -5.03033698e-02 5.61810918e-02 -1.13562751e+00 2.13873342e-01 1.13487041e+00 1.03478944e+00 3.90937865e-01 1.13141917e-01 1.42422784e-02 3.00207585e-01 2.94740498e-01 8.99567127e-01 -1.65928587e-01 -5.91560364e-01 6.03071868e-01 1.95347428e-01 5.08485794e-01 -1.20812869e+00 -6.47502899e-01 -7.56773114e-01 -7.36355662e-01 3.37097347e-01 8.66264880e-01 -4.51654792e-01 -4.06328559e-01 1.60190713e+00 3.66599321e-01 4.45499361e-01 1.32757276e-01 7.54685700e-01 -1.09255001e-01 4.55019355e-01 -4.96523291e-01 -6.09316349e-01 1.05932450e+00 -2.85002112e-01 -2.80137300e-01 2.77868897e-01 5.79358697e-01 -8.02929163e-01 4.49139714e-01 7.14620709e-01 -8.37705970e-01 -2.22953167e-02 -1.08962953e+00 7.95357168e-01 -2.51314044e-01 1.49511218e-01 3.97263542e-02 9.53298569e-01 -8.69422734e-01 3.04251671e-01 -6.13882422e-01 -4.42520142e-01 1.81588396e-01 1.67666078e-01 -1.94106400e-01 1.35220706e-01 -7.79207408e-01 9.17851508e-01 2.18033835e-01 1.30485654e-01 -7.19869316e-01 -7.78411508e-01 -7.48224735e-01 -1.11400068e-01 6.50294781e-01 -3.27308625e-01 8.25892329e-01 -5.35478890e-01 -1.22358370e+00 4.06064868e-01 -5.79104908e-02 -6.12372458e-01 7.60278642e-01 -2.39443913e-01 -1.04199566e-01 -3.66276689e-03 2.03495622e-01 -2.76016444e-01 8.88455689e-01 -1.36427188e+00 -4.84614640e-01 -4.51945692e-01 -1.87457919e-01 -2.17722468e-02 8.62138569e-02 -2.01217979e-01 -2.90642470e-01 -7.70899296e-01 2.73945689e-01 -1.02154529e+00 -4.78450716e-01 -2.75977135e-01 -6.61373854e-01 9.47534442e-02 4.46021765e-01 -5.51635683e-01 1.13909781e+00 -2.30345321e+00 1.06700599e-01 9.92645144e-01 1.09816119e-01 -1.29578993e-01 3.15636575e-01 4.07577187e-01 -6.03654943e-02 1.80723697e-01 -4.06845331e-01 -3.35049450e-01 1.05103038e-01 -1.22395739e-01 -5.49016654e-01 1.35137916e+00 -2.46337473e-01 3.82277220e-01 -7.35750973e-01 -2.16376007e-01 1.38346106e-01 2.56882906e-01 -4.15273488e-01 -3.83643014e-03 2.01147720e-02 3.14316183e-01 -4.58395034e-01 6.90733373e-01 1.03417122e+00 2.59638011e-01 -3.11949342e-01 2.42034402e-02 -3.60654861e-01 -4.54953462e-01 -1.66834128e+00 1.31242120e+00 -4.59474325e-01 4.59000587e-01 2.36736715e-01 -1.13698959e+00 1.16248798e+00 6.24551065e-02 9.81682062e-01 -1.10259421e-01 1.70511633e-01 4.69084948e-01 -3.04529458e-01 -4.43714932e-02 3.31022978e-01 -1.01944946e-01 -3.70777696e-01 2.65482038e-01 -2.83875197e-01 -1.62890345e-01 -1.55243590e-01 -1.47598237e-01 9.70254540e-01 -3.52017283e-01 4.41835344e-01 -5.59673965e-01 2.72353142e-01 -1.00792669e-01 4.58529830e-01 7.85755396e-01 4.76606265e-02 6.78315759e-01 7.22804546e-01 -9.72419381e-02 -7.98280835e-01 -1.23845410e+00 -4.65948671e-01 2.32712150e-01 9.37824324e-02 -1.27594218e-01 -6.42845631e-01 -4.47691083e-01 3.78538638e-01 7.74342716e-01 -4.64938879e-01 1.40536472e-01 -4.43251610e-01 -7.98492074e-01 6.27982616e-01 -4.14866060e-02 2.15376362e-01 -8.97189304e-02 -7.24397659e-01 6.66698739e-02 9.74811912e-02 -9.29123759e-01 -4.76737738e-01 2.76944399e-01 -7.41567135e-01 -1.19636571e+00 -8.97476673e-01 -3.53146821e-01 8.99969101e-01 1.48079112e-01 9.84971881e-01 -3.70608240e-01 -1.79643407e-01 6.48962677e-01 -2.58486629e-01 -6.81256592e-01 1.66305378e-01 -2.04529062e-01 3.31996650e-01 2.61561483e-01 -1.61017552e-01 -3.44230980e-01 -2.70790368e-01 5.50722122e-01 -6.30413413e-01 -8.70158792e-01 4.55092520e-01 5.56056678e-01 9.25115287e-01 3.01484108e-01 4.24792051e-01 -5.01557648e-01 5.77439785e-01 -7.92618573e-01 -1.42339599e+00 2.42925987e-01 -5.28228521e-01 1.08986437e-01 5.99097550e-01 -2.31782377e-01 -6.11313999e-01 2.99073756e-01 3.22117090e-01 -4.54949349e-01 1.84418932e-01 4.96712446e-01 -3.92510533e-01 -7.24639416e-01 5.97156048e-01 -5.07013686e-03 -7.00599030e-02 -3.35224390e-01 3.69079083e-01 4.04662251e-01 4.46285933e-01 -6.13115072e-01 1.13980985e+00 5.08251607e-01 7.85001516e-01 -1.00707734e+00 -7.80379832e-01 -7.42617786e-01 -4.19553936e-01 -2.76011199e-01 4.52385396e-01 -6.25182211e-01 -8.30889344e-01 5.14429808e-02 -9.95288610e-01 1.39178336e-01 -6.80872440e-01 6.11040592e-01 -7.73604274e-01 5.45127630e-01 3.59878868e-01 -1.15571702e+00 1.14854462e-01 -9.63475406e-01 1.00659025e+00 -3.70785967e-02 2.40557447e-01 -1.07433271e+00 3.67516816e-01 -3.25827420e-01 3.90153110e-01 9.55309153e-01 3.24457645e-01 -8.04120004e-01 -4.56568450e-01 -6.18480265e-01 5.19398190e-02 2.10026160e-01 -1.82844043e-01 -3.00713211e-01 -5.32975197e-01 -4.71774042e-01 5.49167514e-01 3.28149348e-01 6.08166039e-01 8.44869673e-01 1.06031978e+00 -5.01756907e-01 -5.42934835e-01 1.05096102e+00 1.75479043e+00 7.84100518e-02 3.30986232e-01 3.37784320e-01 2.88650692e-01 4.98931766e-01 9.68695462e-01 7.29695141e-01 3.56625803e-02 9.36057866e-01 8.53065968e-01 -2.29486637e-03 5.35384417e-01 1.26993865e-01 3.44514221e-01 4.68071625e-02 9.80855972e-02 -3.33636969e-01 -9.90983844e-01 5.67016125e-01 -1.81153250e+00 -9.57580090e-01 -2.37683132e-01 2.82283354e+00 2.29145512e-01 -6.12899922e-02 2.78573483e-01 -1.12616383e-01 8.69236469e-01 -1.60904929e-01 -3.02284688e-01 -2.36892313e-01 -1.25350058e-01 -7.42963701e-02 1.28069103e+00 8.58230233e-01 -1.14809489e+00 1.41504362e-01 5.68700123e+00 8.90373766e-01 -8.55144501e-01 -3.48860957e-02 3.13293666e-01 -7.83403888e-02 -1.07007869e-01 1.63170785e-01 -9.85650361e-01 5.85907817e-01 9.23841715e-01 -5.34332097e-01 1.99504942e-01 9.21391010e-01 2.37203851e-01 -3.98273349e-01 -8.70862246e-01 9.40659046e-01 4.39014167e-01 -8.95771146e-01 -5.50946414e-01 5.04749417e-01 6.52430058e-01 -9.29728225e-02 1.89980611e-01 -1.21656612e-01 9.46643054e-02 -1.02285719e+00 6.64720178e-01 7.39807487e-01 5.08905172e-01 -1.02404881e+00 9.70752239e-01 3.02183032e-01 -1.21976316e+00 4.40387391e-02 -3.36199880e-01 3.92674506e-01 3.00672531e-01 1.17576432e+00 -9.44599152e-01 9.48055625e-01 4.22434300e-01 3.47524256e-01 -3.32757831e-01 1.91350198e+00 6.58934265e-02 4.11478221e-01 -9.58389163e-01 1.09512359e-01 8.99074301e-02 -4.12274241e-01 1.07305956e+00 1.15911543e+00 1.06802475e+00 3.74054238e-02 2.51604259e-01 4.90844607e-01 1.69223085e-01 2.06246942e-01 -9.67248559e-01 5.62470734e-01 5.53479433e-01 9.73867178e-01 -7.54432440e-01 1.34875059e-01 -1.37137726e-01 4.06135231e-01 -2.12083951e-01 2.54009157e-01 -8.77971888e-01 -7.20948398e-01 3.78618270e-01 1.99489623e-01 3.66514623e-01 -3.36997032e-01 -4.25418675e-01 -8.83604944e-01 3.17139000e-01 -4.97487009e-01 4.07128811e-01 -1.29044428e-01 -1.17267025e+00 5.91563046e-01 4.29839641e-01 -1.61344624e+00 -2.71035373e-01 -5.16399264e-01 -5.96346200e-01 9.07357514e-01 -1.21195936e+00 -6.52645051e-01 -2.97078311e-01 5.75491846e-01 1.25416191e-02 -1.68703780e-01 5.24693251e-01 2.40551412e-01 -3.19079727e-01 5.36413133e-01 5.87818265e-01 -1.36514962e-01 5.66695929e-01 -1.37790656e+00 1.47142196e-02 1.19368196e+00 -2.79501807e-02 2.62260258e-01 1.00314975e+00 -6.34728849e-01 -1.24497271e+00 -1.15880299e+00 6.16305828e-01 -5.69420338e-01 7.59210467e-01 -4.11126792e-01 -3.66044492e-01 5.40624380e-01 -2.49700427e-01 2.11478725e-01 4.55892146e-01 -8.85465816e-02 8.62700120e-02 -2.13911504e-01 -1.45846474e+00 2.98280478e-01 6.85868502e-01 1.11643583e-01 -4.72613841e-01 6.92424238e-01 3.40044260e-01 -5.01276135e-01 -9.26053166e-01 4.45404142e-01 1.30746603e-01 -8.14500332e-01 1.09672678e+00 -3.38806473e-02 -4.65711623e-01 -5.97250581e-01 -5.01001596e-01 -1.52565300e+00 2.31103733e-01 -9.19529915e-01 4.54941243e-02 1.20322025e+00 3.92206877e-01 -1.05087054e+00 6.94461346e-01 7.09109008e-02 -2.78394699e-01 -7.53672004e-01 -1.46865106e+00 -1.24765325e+00 4.29733191e-03 -5.13267338e-01 3.98392498e-01 6.17305219e-01 -1.61305234e-01 -3.03779215e-01 -3.40023935e-01 7.84075975e-01 1.25104153e+00 3.11477184e-02 7.88262069e-01 -1.44129229e+00 -4.13373232e-01 -3.87529701e-01 -5.43569922e-01 -8.54717016e-01 -3.87910455e-02 -7.77403176e-01 2.57722437e-01 -1.13949275e+00 -4.34202820e-01 -8.19168568e-01 4.64286702e-03 -5.42935580e-02 3.23883623e-01 -9.28504318e-02 1.89195096e-01 1.70030802e-01 -4.14919883e-01 2.72556722e-01 5.58262825e-01 1.68962758e-02 -2.44299889e-01 6.85121655e-01 -7.19335735e-01 9.20756400e-01 7.19701588e-01 -8.18785191e-01 -2.41650015e-01 -1.52991772e-01 3.11494380e-01 3.07048947e-01 3.57038677e-01 -1.18921757e+00 3.47778291e-01 -5.49498983e-02 2.22565293e-01 -7.36378193e-01 3.18051487e-01 -1.23890591e+00 3.24261665e-01 4.96400446e-01 1.72034308e-01 2.37170056e-01 1.87630087e-01 8.09780657e-01 -2.72785008e-01 -5.40641010e-01 6.71351910e-01 2.66837299e-01 -2.14889064e-01 2.92292356e-01 -6.05214611e-02 1.70351326e-01 1.59714782e+00 -1.96960643e-01 -2.64646441e-01 -5.25223851e-01 -7.94747472e-01 2.47113109e-01 5.39367676e-01 -2.65524536e-01 4.61647272e-01 -1.34990764e+00 -7.43755579e-01 5.03123775e-02 4.88470383e-02 2.17508152e-01 -1.20139234e-01 1.19055510e+00 -5.57002366e-01 3.39615941e-01 2.68799454e-01 -8.32289279e-01 -9.04873431e-01 4.79654551e-01 4.91071463e-01 -2.31463388e-01 -2.94052601e-01 7.16687500e-01 -8.59079212e-02 -1.55198961e-01 6.00174069e-01 -3.82833093e-01 4.29905020e-02 2.55885720e-01 2.50847012e-01 7.55462229e-01 1.31141752e-01 -6.19825363e-01 -5.75411499e-01 1.02816629e+00 6.11422360e-01 -2.92446673e-01 1.30325270e+00 6.14453964e-02 -9.40996483e-02 1.10590853e-01 1.15979970e+00 8.63349020e-01 -1.13855875e+00 5.97452261e-02 3.94002497e-01 -5.85960388e-01 -1.42842829e-01 -3.87821317e-01 -9.33735371e-01 4.66488212e-01 6.62696421e-01 4.90457654e-01 1.22417140e+00 -1.08216256e-01 1.34070903e-01 4.02327776e-01 4.86116529e-01 -8.28630090e-01 -1.41587749e-01 6.00691736e-01 1.06983721e+00 -9.13163245e-01 3.72172557e-02 -3.08364034e-01 -3.09465259e-01 1.08007276e+00 1.48570150e-01 -5.73393881e-01 9.84765887e-01 3.86560947e-01 -2.36097366e-01 -7.66893500e-04 -1.72287878e-02 -2.39271328e-01 3.05138856e-01 6.27943993e-01 -1.63630649e-01 2.21247390e-01 -3.87455881e-01 4.54254031e-01 -2.61305422e-01 -4.70117688e-01 6.82364106e-01 6.80631161e-01 -7.10329890e-01 -7.55017221e-01 -1.05467808e+00 3.47540796e-01 -5.29125988e-01 2.71436367e-02 -8.39454457e-02 1.05476427e+00 1.19332761e-01 8.70499074e-01 -1.10088758e-01 -4.99329269e-02 7.81374037e-01 -2.76055187e-01 2.08361715e-01 -4.22927201e-01 -1.30021542e-01 1.42318234e-01 -9.72613767e-02 -5.01399338e-01 -8.34686309e-02 -9.40607190e-01 -8.60222936e-01 -1.83423713e-01 -4.74402606e-01 6.72592878e-01 1.01106024e+00 6.30207777e-01 1.51774064e-01 1.41212404e-01 9.23852265e-01 -8.34494889e-01 -9.90911126e-01 -7.02649653e-01 -8.37730289e-01 -2.73735553e-01 5.38072109e-01 -8.35935235e-01 -8.51938963e-01 -5.51815331e-01]
[6.747450351715088, 3.9457013607025146]
ed333c6d-8fd9-4013-8194-d788e1b081d5
scam-transferring-humans-between-images-with
2210.04883
null
https://arxiv.org/abs/2210.04883v1
https://arxiv.org/pdf/2210.04883v1.pdf
SCAM! Transferring humans between images with Semantic Cross Attention Modulation
A large body of recent work targets semantically conditioned image generation. Most such methods focus on the narrower task of pose transfer and ignore the more challenging task of subject transfer that consists in not only transferring the pose but also the appearance and background. In this work, we introduce SCAM (Semantic Cross Attention Modulation), a system that encodes rich and diverse information in each semantic region of the image (including foreground and background), thus achieving precise generation with emphasis on fine details. This is enabled by the Semantic Attention Transformer Encoder that extracts multiple latent vectors for each semantic region, and the corresponding generator that exploits these multiple latents by using semantic cross attention modulation. It is trained only using a reconstruction setup, while subject transfer is performed at test time. Our analysis shows that our proposed architecture is successful at encoding the diversity of appearance in each semantic region. Extensive experiments on the iDesigner and CelebAMask-HD datasets show that SCAM outperforms SEAN and SPADE; moreover, it sets the new state of the art on subject transfer.
['Vicky Kalogeiton', 'David Picard', 'Nicolas Dufour']
2022-10-10
null
null
null
null
['reconstruction', 'pose-transfer']
['computer-vision', 'computer-vision']
[ 6.19308054e-01 3.58312011e-01 2.06330106e-01 -4.86023098e-01 -1.02110410e+00 -3.89295101e-01 9.49302316e-01 -8.11174214e-01 -3.76043469e-02 8.18126500e-01 2.19232589e-01 1.31504536e-01 3.29056889e-01 -7.22549558e-01 -1.08597445e+00 -1.01628542e+00 3.56160909e-01 6.61248326e-01 2.65890539e-01 -4.00617659e-01 -7.48748481e-02 2.96074580e-02 -1.69331956e+00 6.44855082e-01 7.10995555e-01 9.83445406e-01 3.74372095e-01 4.39580023e-01 2.22711116e-02 6.80928707e-01 -9.52107668e-01 -6.61213458e-01 9.85062718e-02 -9.49758530e-01 -8.36531162e-01 2.06103012e-01 1.03804541e+00 -3.74273181e-01 -1.66427717e-01 1.07575548e+00 6.43364906e-01 -8.90463814e-02 8.32693338e-01 -1.47858274e+00 -8.29371870e-01 5.61724663e-01 -6.13262951e-01 -8.68277624e-02 1.25071332e-01 2.61721849e-01 8.07805359e-01 -9.00648236e-01 7.66275167e-01 1.54824162e+00 1.91591859e-01 9.60807025e-01 -1.23809814e+00 -8.35585237e-01 2.33990848e-01 2.34798402e-01 -1.27904749e+00 -3.76263052e-01 7.68768609e-01 -4.11443084e-01 4.38952208e-01 2.66837358e-01 4.42637205e-01 1.72026110e+00 2.28108376e-01 1.08463526e+00 1.41153884e+00 -4.36593592e-01 5.61758392e-02 2.56033063e-01 -1.87796324e-01 6.36161208e-01 -8.81218538e-02 2.01388299e-01 -7.31606066e-01 4.33753967e-01 8.44152272e-01 -3.59973609e-01 -4.83190566e-01 -3.73817921e-01 -1.36212969e+00 7.14907706e-01 5.45328796e-01 2.11362526e-01 -1.72933236e-01 2.44414017e-01 2.18524441e-01 1.92593381e-01 5.91377497e-01 1.86957017e-01 -3.59770566e-01 3.65851372e-01 -1.03071129e+00 2.23027259e-01 6.02456093e-01 1.30977464e+00 5.75961769e-01 2.31999382e-01 -8.18411231e-01 6.00770593e-01 3.02703738e-01 8.50288808e-01 4.77089286e-01 -5.72319388e-01 5.13879418e-01 2.74735183e-01 1.70949548e-02 -7.23316133e-01 4.26200591e-02 -6.05279207e-01 -7.50307620e-01 5.24629474e-01 1.75517380e-01 -2.00555548e-02 -1.41838062e+00 2.17146730e+00 4.04446363e-01 2.99655348e-01 1.47268400e-01 1.13076591e+00 1.02467728e+00 7.40128160e-01 3.92820656e-01 3.05279166e-01 1.54932892e+00 -1.21524429e+00 -7.10196495e-01 -3.74490499e-01 -3.67032625e-02 -7.99137056e-01 1.12466192e+00 2.36323118e-01 -1.22047842e+00 -7.81410277e-01 -1.07418549e+00 -1.91114664e-01 -3.61482501e-01 4.21259493e-01 4.51911807e-01 3.31947625e-01 -1.15704989e+00 3.53941351e-01 -5.11845589e-01 -2.65285224e-01 4.13677871e-01 2.08942890e-01 -4.40252542e-01 -1.68341681e-01 -1.43831575e+00 1.04569089e+00 4.12970841e-01 1.64644629e-01 -1.34785247e+00 -7.18670428e-01 -1.05154085e+00 1.75436795e-01 1.44321099e-01 -1.19121182e+00 1.09037161e+00 -1.51124728e+00 -1.75172162e+00 1.12387192e+00 2.21260801e-01 -3.49661320e-01 6.78807259e-01 -2.96250701e-01 -1.95161760e-01 2.63977647e-01 1.86776757e-01 1.18960464e+00 1.42360234e+00 -1.51651025e+00 -5.40194213e-01 -4.39556450e-01 -5.01630642e-02 4.61627632e-01 -6.82883635e-02 1.05661929e-01 -5.81805348e-01 -8.21356952e-01 -3.35777283e-01 -8.75135899e-01 2.91175507e-02 -8.74880105e-02 -5.13943255e-01 -3.73964082e-03 8.95130336e-01 -1.07706964e+00 6.76634490e-01 -2.26768589e+00 6.51599169e-01 -4.90289330e-02 -5.50631918e-02 2.17866004e-01 -4.01640058e-01 1.12687245e-01 -2.04418764e-01 -3.81649822e-01 -3.72900963e-01 -5.57674348e-01 1.93862811e-01 6.98243976e-02 -6.37871027e-01 3.25973779e-01 4.12430018e-01 1.17827392e+00 -7.46262789e-01 -5.51577389e-01 3.30122918e-01 8.40324044e-01 -5.65368712e-01 4.21804667e-01 -4.30593431e-01 8.32601666e-01 -3.28081280e-01 4.62512136e-01 7.64869094e-01 -3.75793606e-01 4.13651504e-02 -4.24006402e-01 2.79610872e-01 -6.55252263e-02 -8.57224584e-01 2.02952790e+00 -6.22758567e-01 3.65301818e-01 7.47820884e-02 -8.07060897e-01 6.92903638e-01 3.37526679e-01 8.93153176e-02 -9.81034279e-01 2.39829510e-01 8.61440599e-02 -2.18193769e-01 -1.95633814e-01 2.86476344e-01 -1.52977824e-01 -2.57397085e-01 8.35215375e-02 6.32992983e-01 -1.14290968e-01 -1.51752960e-02 3.00069004e-01 5.33840179e-01 7.02156067e-01 -7.42005557e-02 -4.23382908e-01 5.87884367e-01 -3.24352443e-01 2.57474601e-01 4.16222990e-01 1.96586043e-01 9.79278266e-01 2.53867388e-01 6.71586022e-02 -9.50731575e-01 -1.45604455e+00 4.43033390e-02 1.13771331e+00 3.71695369e-01 -1.68842822e-01 -1.15872836e+00 -9.29705501e-01 -1.53377458e-01 9.08524871e-01 -9.85278308e-01 -4.52748477e-01 -5.59019625e-01 -6.10407233e-01 2.12602347e-01 6.30629957e-01 5.61585724e-01 -1.47376192e+00 -4.91600156e-01 2.23953929e-02 -5.39545536e-01 -1.10134399e+00 -7.17072427e-01 -3.31220865e-01 -4.20684367e-01 -9.21883404e-01 -1.15434420e+00 -7.85808742e-01 8.82537723e-01 -8.37201402e-02 1.32431555e+00 -3.58447582e-01 -5.89549243e-01 3.18742782e-01 -2.65897304e-01 -3.73705178e-01 -4.64884669e-01 9.16613638e-02 -5.40186346e-01 6.85406208e-01 -1.30301341e-01 -3.83071512e-01 -7.95613587e-01 3.12040478e-01 -8.67301047e-01 6.53886318e-01 9.16151047e-01 9.94001985e-01 4.85286325e-01 -3.67946804e-01 4.38124508e-01 -9.45281923e-01 -2.33870987e-02 -3.04662704e-01 -5.70702434e-01 3.66835237e-01 -1.15692705e-01 5.12211677e-03 3.41245890e-01 -2.43460551e-01 -1.52167857e+00 6.93761259e-02 -1.46167502e-01 -6.03083789e-01 -2.78151959e-01 -1.37381166e-01 -6.42678857e-01 1.23316169e-01 4.22246963e-01 5.48769772e-01 8.61492008e-02 -5.04013598e-01 5.99250138e-01 4.42401350e-01 6.90372169e-01 -5.73976338e-01 7.97600985e-01 6.09923780e-01 -8.31586588e-03 -4.96776909e-01 -1.12721097e+00 -1.63263887e-01 -5.61243355e-01 -2.97908574e-01 1.27113032e+00 -1.01501644e+00 -4.46288705e-01 6.89580441e-01 -1.07888985e+00 -5.18023729e-01 -5.43574750e-01 1.66627139e-01 -8.58478665e-01 -1.01414863e-02 -5.75469255e-01 -3.92952472e-01 -4.51738060e-01 -1.18881273e+00 1.80009413e+00 -2.37620361e-02 -8.43472183e-02 -8.38757455e-01 -1.71087861e-01 5.45198500e-01 4.29004848e-01 3.36038560e-01 7.00048566e-01 -2.15966240e-01 -9.91864085e-01 1.80911526e-01 -3.38117331e-01 5.13604820e-01 -6.59516230e-02 -4.18561310e-01 -1.21162271e+00 -5.56592286e-01 -1.98728159e-01 -3.45720708e-01 1.04854286e+00 3.38775188e-01 1.16539264e+00 -2.87547797e-01 -2.88659811e-01 7.55081832e-01 1.40052795e+00 -4.14949320e-02 1.03837335e+00 -3.11688650e-02 8.62605572e-01 7.84617603e-01 6.44821048e-01 -6.46577850e-02 2.92069435e-01 1.03256750e+00 5.19081831e-01 -7.34840393e-01 -8.38673234e-01 -3.97387445e-01 5.92169940e-01 4.68283743e-01 -2.23813653e-02 -4.13031369e-01 -3.75954270e-01 5.59877872e-01 -1.77792513e+00 -1.00535429e+00 1.19308546e-01 2.03359437e+00 9.56979752e-01 -3.27133119e-01 -1.16011575e-01 -1.40905201e-01 8.34821522e-01 5.65919951e-02 -2.28224069e-01 7.35564530e-02 -1.96529061e-01 5.88655472e-01 2.22071260e-01 5.72568119e-01 -1.14263856e+00 1.28134906e+00 5.90999126e+00 9.57859755e-01 -1.12529206e+00 2.76973188e-01 6.09617889e-01 -5.03488444e-02 -2.15690270e-01 -2.61661828e-01 -9.68477368e-01 7.35327005e-01 7.37513721e-01 1.51947320e-01 1.66955635e-01 7.44223595e-01 -3.39350879e-01 1.72071800e-01 -1.11526775e+00 7.61772811e-01 6.54473245e-01 -9.69476521e-01 6.02549672e-01 -5.21007851e-02 9.37422514e-01 -4.54744935e-01 4.92115885e-01 4.04127568e-01 2.72632629e-01 -9.86130953e-01 1.13839972e+00 6.34252727e-01 1.29647624e+00 -5.55542707e-01 5.95837414e-01 -1.33999810e-03 -8.92317712e-01 1.33417889e-01 5.03847227e-02 4.68552500e-01 2.72216201e-01 2.99725860e-01 -7.05241621e-01 6.05268598e-01 6.49002433e-01 5.70942819e-01 -6.59992397e-01 7.85927773e-01 -7.11128533e-01 5.55500507e-01 7.68210068e-02 4.91347134e-01 1.25115678e-01 -5.72124831e-02 4.43649322e-01 1.29751205e+00 3.95268142e-01 -2.48072356e-01 6.61171749e-02 1.10207725e+00 2.53416207e-02 4.41707298e-02 -3.44876617e-01 4.17993277e-01 -1.33873075e-01 1.21148133e+00 -4.58430260e-01 -7.13506341e-01 -2.65389770e-01 1.63271809e+00 1.20961189e-01 4.88315701e-01 -1.30411828e+00 -2.21754745e-01 5.01734138e-01 1.43489331e-01 5.76633334e-01 3.26878756e-01 -2.55698059e-02 -1.28824174e+00 -1.11752354e-01 -8.50309074e-01 4.52124655e-01 -1.14458919e+00 -1.33425391e+00 6.76254034e-01 1.41896531e-01 -1.22056162e+00 -3.01904798e-01 -6.44428790e-01 -3.34061682e-01 1.15750265e+00 -1.63765085e+00 -2.00960636e+00 -5.21456003e-01 8.37523997e-01 7.66887963e-01 -1.24373354e-01 7.07114935e-01 5.31641304e-01 -3.80028486e-01 7.68734574e-01 -2.90559739e-01 -2.78630443e-02 1.01828718e+00 -1.28019178e+00 4.19652015e-01 7.69582272e-01 4.77208085e-02 3.68524194e-01 6.42244339e-01 -5.35824478e-01 -7.68891692e-01 -1.48273635e+00 8.50583315e-01 -6.18808210e-01 1.91294312e-01 -7.30514407e-01 -6.63824022e-01 8.43666852e-01 5.27782500e-01 -3.21933627e-02 3.82965535e-01 -3.23560447e-01 -4.12666619e-01 -1.65384963e-01 -1.01378250e+00 5.49194038e-01 1.05231154e+00 -2.78002381e-01 -5.23579180e-01 2.87218064e-01 8.19840848e-01 -6.29943430e-01 -4.75838989e-01 5.10699451e-01 3.18696111e-01 -7.55023539e-01 1.10100818e+00 -6.14671111e-01 7.75986433e-01 -4.75181878e-01 -1.48777634e-01 -1.34548998e+00 -2.83950001e-01 -3.84772658e-01 -7.77301416e-02 1.46692574e+00 2.15808958e-01 -4.49314088e-01 5.52742600e-01 7.24116713e-02 -2.99062610e-01 -4.22164410e-01 -7.20650136e-01 -6.26786470e-01 2.35531833e-02 2.11004123e-01 6.60111070e-01 9.11715090e-01 -7.42403448e-01 6.94898546e-01 -6.70020282e-01 1.35793343e-01 9.16758120e-01 4.16946352e-01 8.81648540e-01 -8.40824068e-01 -5.74440122e-01 -1.08378604e-01 -4.81590122e-01 -8.14427853e-01 3.27915519e-01 -9.88331795e-01 2.13208258e-01 -1.41269326e+00 4.27922159e-01 4.28652838e-02 -3.57235849e-01 4.71635014e-01 -2.50762492e-01 7.09888577e-01 2.79074401e-01 -7.97320232e-02 -6.39265001e-01 8.85032594e-01 1.63560963e+00 -2.12219983e-01 3.32546055e-01 -1.07452519e-01 -7.45868862e-01 5.62960863e-01 4.15542603e-01 -6.79643825e-02 -5.96023440e-01 -5.49768090e-01 -3.67575705e-01 5.57644442e-02 1.01004803e+00 -9.87920344e-01 -2.22560853e-01 8.60438868e-02 6.26091659e-01 -4.87715513e-01 5.57203591e-01 -7.66759217e-01 1.74051315e-01 2.87207007e-01 -2.84484833e-01 -4.25928414e-01 1.52066052e-01 5.02270758e-01 -3.64753515e-01 3.25605184e-01 1.13171935e+00 -1.24544755e-01 -9.33646798e-01 4.26340789e-01 1.93234563e-01 7.10551962e-02 1.23103189e+00 -2.29823757e-02 -2.59534180e-01 -2.03562781e-01 -1.06644118e+00 6.50999174e-02 2.58856297e-01 7.25061059e-01 4.68571693e-01 -1.51903546e+00 -1.03541410e+00 4.51367587e-01 2.90251166e-01 -1.27612144e-01 6.77594364e-01 5.40197849e-01 -1.44145787e-01 2.92071670e-01 -5.19574702e-01 -6.90824986e-01 -1.30691803e+00 5.42132676e-01 3.99484187e-01 -1.99039310e-01 -7.90371835e-01 1.07661462e+00 1.21174574e+00 -1.59657180e-01 8.62912834e-02 1.12432197e-01 -2.50061363e-01 -4.82033826e-02 5.79219759e-01 -7.01161250e-02 -1.45250827e-01 -9.46928144e-01 -1.96564272e-01 6.32199228e-01 -5.69724031e-02 -2.90718943e-01 1.05469489e+00 -7.82023445e-02 -7.33439699e-02 2.06973195e-01 1.14271677e+00 -2.61748731e-01 -1.53070593e+00 -9.33145359e-02 -6.09177828e-01 -6.24681473e-01 -1.70381486e-01 -1.25971246e+00 -1.32411671e+00 1.10861981e+00 6.23228371e-01 -4.98941630e-01 1.35730386e+00 2.06707880e-01 7.28406429e-01 -4.75064754e-01 6.45398796e-01 -8.12684000e-01 2.55455703e-01 3.14413369e-01 1.40972412e+00 -1.00294077e+00 -4.46302652e-01 -6.31358027e-01 -7.67939985e-01 5.10837436e-01 1.07676017e+00 -1.06048174e-01 2.36035004e-01 6.94457293e-02 1.04395404e-01 -1.09995939e-01 -6.04768157e-01 -2.88575888e-01 5.85128546e-01 6.85664177e-01 3.31191540e-01 6.79997280e-02 1.00400019e-02 5.27526021e-01 -2.94769466e-01 -2.33848020e-01 -2.67989561e-02 5.59328914e-01 -7.50505552e-02 -1.07673395e+00 -3.97057772e-01 2.04327293e-02 -3.90316904e-01 -2.83476472e-01 -3.81710559e-01 7.92504251e-01 3.47501487e-01 3.81726146e-01 1.77905709e-01 -1.77676231e-02 3.58390659e-01 1.18461698e-01 9.09409821e-01 -7.43862212e-01 -4.62405711e-01 7.66108260e-02 -1.35363648e-02 -7.85962582e-01 -2.32049063e-01 -5.68519652e-01 -8.71689439e-01 2.00668842e-01 -1.65850610e-01 8.05105194e-02 6.59695089e-01 6.64370358e-01 3.35164815e-01 1.03246260e+00 4.77186501e-01 -9.70040917e-01 -5.97851753e-01 -9.94709551e-01 -6.15953147e-01 8.94069731e-01 3.55610907e-01 -8.42458248e-01 -1.15619995e-01 5.04874229e-01]
[11.61235237121582, -0.5833448171615601]
50b5d7ec-1d4b-4ad0-ba83-a4380a2eec7f
learning-thermodynamically-constrained
2306.17004
null
https://arxiv.org/abs/2306.17004v1
https://arxiv.org/pdf/2306.17004v1.pdf
Learning thermodynamically constrained equations of state with uncertainty
Numerical simulations of high energy-density experiments require equation of state (EOS) models that relate a material's thermodynamic state variables -- specifically pressure, volume/density, energy, and temperature. EOS models are typically constructed using a semi-empirical parametric methodology, which assumes a physics-informed functional form with many tunable parameters calibrated using experimental/simulation data. Since there are inherent uncertainties in the calibration data (parametric uncertainty) and the assumed functional EOS form (model uncertainty), it is essential to perform uncertainty quantification (UQ) to improve confidence in the EOS predictions. Model uncertainty is challenging for UQ studies since it requires exploring the space of all possible physically consistent functional forms. Thus, it is often neglected in favor of parametric uncertainty, which is easier to quantify without violating thermodynamic laws. This work presents a data-driven machine learning approach to constructing EOS models that naturally captures model uncertainty while satisfying the necessary thermodynamic consistency and stability constraints. We propose a novel framework based on physics-informed Gaussian process regression (GPR) that automatically captures total uncertainty in the EOS and can be jointly trained on both simulation and experimental data sources. A GPR model for the shock Hugoniot is derived and its uncertainties are quantified using the proposed framework. We apply the proposed model to learn the EOS for the diamond solid state of carbon, using both density functional theory data and experimental shock Hugoniot data to train the model and show that the prediction uncertainty reduces by considering the thermodynamic constraints.
['Michael D. Shields', 'Dimitrios Tsapetis', 'Jim A. Gaffney', 'Himanshu Sharma']
2023-06-29
null
null
null
null
['gpr', 'gpr']
['computer-vision', 'miscellaneous']
[-1.80523172e-01 -1.15758076e-01 -6.63438961e-02 -3.61685395e-01 -7.79089928e-01 -6.20875321e-02 6.48826420e-01 5.26567400e-01 -2.95535475e-01 9.60267544e-01 -2.79744625e-01 -3.78518969e-01 -3.65740478e-01 -9.39375699e-01 -7.91543841e-01 -9.83087182e-01 1.86594054e-01 1.07459438e+00 2.93700218e-01 1.04324080e-01 3.40672970e-01 7.16495335e-01 -1.39447725e+00 -5.03510237e-01 1.12016916e+00 1.20198631e+00 6.84918314e-02 4.10960793e-01 -1.55604631e-01 3.96451384e-01 -1.40953243e-01 5.33497036e-02 1.11877859e-01 -2.19153419e-01 -5.01476049e-01 -6.48679912e-01 -4.68187243e-01 -7.81519432e-03 -2.18320578e-01 1.00108039e+00 2.27580592e-01 5.77312291e-01 1.48067379e+00 -1.27605164e+00 -2.78249323e-01 4.15658534e-01 -3.51433009e-01 -2.10244641e-01 -1.72527939e-01 2.68310308e-01 9.33329582e-01 -6.53193414e-01 2.40239531e-01 1.05239236e+00 2.36646309e-01 3.11823070e-01 -1.31976259e+00 -5.15546679e-01 -3.63267541e-01 -1.09957539e-01 -1.46040440e+00 4.45464738e-02 8.99768829e-01 -7.50563562e-01 9.47798193e-01 1.76859468e-01 4.63851184e-01 8.17826033e-01 7.96411753e-01 2.33891323e-01 1.19892406e+00 -5.63152313e-01 8.98800492e-01 2.73619443e-01 1.70255721e-01 3.09415996e-01 5.38346529e-01 6.93443239e-01 -3.31387967e-01 -2.30031446e-01 7.27974772e-01 -1.73986688e-01 -9.69439000e-02 -7.24932313e-01 -5.29701948e-01 9.21289504e-01 3.39322358e-01 9.62952599e-02 -3.24176610e-01 3.74264449e-01 3.81147802e-01 -1.48450330e-01 3.10686767e-01 5.57750344e-01 -6.59683645e-01 -2.98461080e-01 -1.02706349e+00 3.71663123e-01 1.01487339e+00 7.47873366e-01 1.00468218e+00 7.74975196e-02 1.11643948e-01 5.04467547e-01 9.41991210e-01 7.15353847e-01 2.72363126e-01 -6.68404102e-01 1.06371418e-01 2.67323107e-01 4.45134699e-01 -5.06084800e-01 -3.45530272e-01 2.63745207e-02 -7.56361306e-01 4.57071304e-01 2.23730579e-01 -1.12313494e-01 -9.54081357e-01 1.50560582e+00 4.81329113e-01 2.34701578e-02 1.08025342e-01 6.52221382e-01 6.54328823e-01 8.31415057e-01 4.51693147e-01 -3.89255106e-01 9.09347653e-01 -4.66800928e-01 -4.86313462e-01 -4.12166268e-02 3.90537828e-01 -3.80749255e-01 7.71424294e-01 1.41795814e-01 -8.77161086e-01 -4.18506116e-01 -1.30419385e+00 1.70127079e-01 -3.37875605e-01 -3.62253755e-01 5.02618492e-01 6.68539822e-01 -4.61243182e-01 1.32912779e+00 -1.15689087e+00 1.26378331e-02 -2.53268272e-01 3.18596572e-01 -2.03098021e-02 3.25615168e-01 -1.41696024e+00 1.09332538e+00 7.74105847e-01 5.58874235e-02 -9.07577813e-01 -9.08951700e-01 -9.12776530e-01 8.42771381e-02 4.23474103e-01 -6.14491761e-01 1.20899403e+00 1.33587033e-01 -1.87485969e+00 1.62019268e-01 9.30716321e-02 -3.51611108e-01 5.74607730e-01 -1.36461824e-01 -4.57496703e-01 8.50523487e-02 -4.26698506e-01 2.36639410e-01 9.83522713e-01 -1.33260167e+00 5.87281436e-02 3.34786996e-02 -4.78194088e-01 -1.32274240e-01 3.60265315e-01 -1.49703100e-01 1.35731827e-02 -3.60704958e-02 2.92551368e-01 -9.10925567e-01 -4.24882293e-01 -2.85289347e-01 -4.45128977e-01 -3.31501663e-02 4.77402180e-01 -4.81280655e-01 9.34939742e-01 -1.79037905e+00 2.17012927e-01 6.03868067e-01 3.72253098e-02 -1.52057335e-01 6.61248207e-01 5.69777966e-01 -9.51354280e-02 2.85551816e-01 -8.29598427e-01 -2.56115943e-01 3.02135617e-01 2.12829739e-01 -1.52397916e-01 3.71901989e-01 3.51340264e-01 7.16085494e-01 -7.29749739e-01 -5.29630542e-01 8.20055008e-01 4.61169869e-01 -3.10907066e-01 4.94909734e-01 -6.24261737e-01 6.53792143e-01 -4.61559176e-01 3.89292896e-01 8.32849085e-01 1.03344165e-01 -1.39747947e-01 -2.92145640e-01 -2.40330070e-01 7.13668391e-02 -1.27783871e+00 1.15565276e+00 -6.41573131e-01 -1.64186433e-01 -1.38492137e-01 -9.00761068e-01 1.38351762e+00 1.67947382e-01 7.79858947e-01 -1.73521936e-01 5.13750672e-01 5.28834641e-01 -3.42036001e-02 -2.95394838e-01 4.95199233e-01 -9.69943643e-01 -2.00812086e-01 2.26651043e-01 2.46985778e-01 -1.10308743e+00 -1.47304252e-01 2.28299052e-02 6.44766986e-01 4.36039716e-01 4.10691261e-01 -6.78790689e-01 5.53675652e-01 -2.92716771e-01 4.23515648e-01 5.01154184e-01 -1.19318098e-01 5.05966723e-01 5.62481940e-01 2.37222183e-02 -1.43217039e+00 -1.25039411e+00 -6.85865641e-01 2.15420634e-01 1.61804155e-01 -3.39637220e-01 -5.12217820e-01 -2.02601209e-01 4.11744714e-01 1.09102321e+00 -5.69498122e-01 -5.01088738e-01 -2.33522907e-01 -9.26964939e-01 -2.59103198e-02 6.12467945e-01 2.81180620e-01 -8.31676006e-01 -3.83030623e-01 2.36370996e-01 5.44305146e-01 -8.45551252e-01 8.62673149e-02 5.88067889e-01 -7.46427774e-01 -1.04690433e+00 -2.08312973e-01 3.19818676e-01 2.44759619e-01 -6.90802693e-01 8.61570060e-01 -2.85235554e-01 -2.67794490e-01 3.45995992e-01 -7.50156417e-02 -6.34041488e-01 -9.44649577e-01 -2.80589670e-01 3.35347801e-01 -3.32106143e-01 1.23806909e-01 -6.42367959e-01 -4.17562515e-01 1.75538629e-01 -8.53964627e-01 -1.86651647e-01 4.47269976e-01 5.98111272e-01 7.86823094e-01 4.43402469e-01 4.98147666e-01 -7.42540359e-01 3.73663098e-01 -6.05638623e-01 -9.79200542e-01 1.67500138e-01 -1.09070122e+00 6.50068045e-01 6.67847514e-01 -8.41571614e-02 -1.30989373e+00 -3.11554372e-01 -2.01545998e-01 -6.12747908e-01 -1.82961389e-01 6.07226968e-01 -3.44011545e-01 -1.52044818e-01 5.05952179e-01 -9.32337046e-02 -1.65350989e-01 -4.22647834e-01 8.03726376e-04 5.05251527e-01 5.77341378e-01 -1.40358210e+00 8.63074601e-01 7.33703226e-02 7.29806960e-01 -8.02397311e-01 -5.49079716e-01 -3.30337465e-01 -8.15079331e-01 -2.49899104e-01 7.03786910e-01 -5.33556998e-01 -8.96443725e-01 2.90022224e-01 -6.75127983e-01 -1.33093983e-01 -4.28135663e-01 8.67889822e-01 -8.10943902e-01 5.75775623e-01 -4.99352217e-01 -1.34390211e+00 -5.05153000e-01 -1.37423825e+00 9.10808921e-01 3.85460228e-01 -2.17089191e-01 -1.25621819e+00 2.66231775e-01 -4.93790023e-02 3.41077626e-01 7.23286808e-01 1.09781086e+00 -5.06692410e-01 -4.87981677e-01 -1.88112170e-01 -1.90225065e-01 4.59423542e-01 1.27592655e-02 6.56297982e-01 -8.81849051e-01 -1.28257126e-01 4.01938468e-01 -1.50104985e-01 8.80699217e-01 6.71029270e-01 1.04857850e+00 2.05848843e-01 -2.81365365e-01 4.88814741e-01 1.75643253e+00 2.12139562e-01 4.83067393e-01 7.22318292e-02 5.92893541e-01 4.07619387e-01 6.39388025e-01 6.11774147e-01 -1.61576658e-01 2.87084341e-01 3.24536860e-01 4.62639928e-01 5.20601571e-01 -3.94925058e-01 1.49823934e-01 7.08042860e-01 -1.62684858e-01 -1.10025346e-01 -1.01338553e+00 1.23623665e-02 -1.66948414e+00 -5.64764082e-01 -2.06903771e-01 2.62120581e+00 8.83340061e-01 4.51766044e-01 -1.56124979e-01 7.66876191e-02 6.40435636e-01 -9.19231847e-02 -8.24783206e-01 -7.26703763e-01 3.30609709e-01 4.46902335e-01 7.21539080e-01 6.99020028e-01 -9.32686925e-01 4.89676207e-01 6.27251863e+00 8.30924809e-01 -1.15868819e+00 -1.23390257e-01 4.59364742e-01 2.13949978e-01 -5.16972899e-01 3.33283126e-01 -7.63028860e-01 6.07383192e-01 1.42535162e+00 -5.34841061e-01 1.90976039e-01 9.21613395e-01 2.05697447e-01 -5.38414180e-01 -1.18187392e+00 7.25941360e-01 -5.28472066e-01 -1.00131309e+00 -2.81395167e-01 -2.32262760e-02 6.76926374e-01 -1.37914345e-01 -1.50455222e-01 4.99324679e-01 3.75833601e-01 -1.00215244e+00 7.63254523e-01 9.37749088e-01 9.67513561e-01 -7.48053551e-01 9.00720060e-01 3.32463145e-01 -9.76066351e-01 1.41810387e-01 -5.61410367e-01 2.50880569e-01 3.89207751e-01 9.53597128e-01 -1.04069090e+00 9.29079592e-01 4.43122208e-01 3.32844943e-01 -2.37879939e-02 9.34537590e-01 -1.32379398e-01 5.89377224e-01 -8.81264865e-01 -1.31063759e-01 9.90568381e-03 -7.46031404e-01 4.07942921e-01 9.48720574e-01 5.64339340e-01 1.48451691e-02 -9.48262513e-02 1.39862275e+00 1.38671219e-01 6.67162985e-02 -3.42021197e-01 -1.16571374e-01 5.55172622e-01 1.12274611e+00 -7.37781763e-01 1.52062643e-02 -3.93510796e-02 2.32716113e-01 -4.38833870e-02 2.93068349e-01 -1.02123523e+00 -6.84439577e-03 4.15674984e-01 1.87303022e-01 5.06306365e-02 -3.96947861e-01 -3.45649600e-01 -8.27025473e-01 -2.80347258e-01 -5.09736082e-03 2.42509976e-01 -7.30028749e-01 -1.40267503e+00 1.82364836e-01 7.77403712e-01 -9.00161266e-01 -4.35227662e-01 -9.47531819e-01 -8.74155998e-01 1.15577912e+00 -1.42571163e+00 -8.03703368e-01 -8.40293691e-02 -2.86233349e-04 -4.03214879e-02 2.22524688e-01 6.27896786e-01 -1.85385421e-01 -7.04352796e-01 1.95456669e-01 5.61852336e-01 -4.92058009e-01 6.67701066e-01 -1.50985932e+00 4.82706502e-02 4.40100163e-01 -6.22024059e-01 5.01143157e-01 1.46575034e+00 -1.02943647e+00 -1.32449710e+00 -1.01278603e+00 2.17840537e-01 -2.30162814e-01 1.01449609e+00 -1.77582100e-01 -1.16895139e+00 3.79517943e-01 -4.25848663e-01 2.10578457e-01 5.65863192e-01 -1.18364925e-02 -7.90021792e-02 3.91274132e-02 -1.43308830e+00 1.34048030e-01 3.88787597e-01 -5.56635261e-01 -5.44689476e-01 1.80896774e-01 8.56056571e-01 -3.30572963e-01 -1.37584746e+00 7.95968235e-01 3.10656518e-01 -7.17152774e-01 8.73537362e-01 -4.79186147e-01 3.97810787e-01 -3.04116756e-01 -2.67475158e-01 -1.34102893e+00 -5.51291220e-02 -3.46434742e-01 -5.05686581e-01 1.29937458e+00 3.93642336e-01 -6.14187777e-01 4.64289933e-01 1.53653073e+00 -1.48236230e-01 -8.32393110e-01 -1.22332525e+00 -1.05443978e+00 6.68420434e-01 -7.79763937e-01 7.54627645e-01 5.61217844e-01 6.02648221e-02 9.99739766e-03 -2.24231794e-01 3.67995799e-01 7.35212624e-01 -3.51739787e-02 3.78175557e-01 -1.49825466e+00 -3.40312898e-01 -2.39626646e-01 -2.73831338e-01 -2.65556008e-01 2.19685689e-01 -5.98062694e-01 2.26161733e-01 -1.22270918e+00 1.43147381e-02 -6.90531611e-01 -3.69243532e-01 -1.84109956e-01 -1.00490395e-02 -5.11555314e-01 -2.83727585e-03 2.77574241e-01 -3.81270014e-02 1.24698496e+00 9.60304320e-01 8.33078474e-02 -2.54475653e-01 -1.21882819e-02 -9.49001089e-02 6.59255564e-01 8.35139096e-01 -4.21506643e-01 -3.23303938e-01 6.43328130e-01 1.76754370e-01 3.75972182e-01 3.87228191e-01 -1.26813245e+00 6.26120195e-02 -5.97284496e-01 2.67296344e-01 -6.59859836e-01 4.33400482e-01 -9.54444110e-01 5.42935073e-01 2.22750455e-01 -6.77911714e-02 -6.04301631e-01 2.62929589e-01 7.25491762e-01 -2.83437762e-02 -7.81415045e-01 1.07327902e+00 -4.74381121e-03 -3.92102361e-01 4.35416460e-01 -5.57564609e-02 -3.13621283e-01 1.09475863e+00 8.87535587e-02 3.80139472e-03 -7.97799975e-02 -7.47153044e-01 1.09306514e-01 5.86625814e-01 -1.02873959e-01 3.80862117e-01 -1.13490260e+00 -4.91842508e-01 6.46103472e-02 5.53398877e-02 2.71653295e-01 5.65768182e-01 5.16539752e-01 -6.57191217e-01 4.33393568e-01 1.12032302e-01 -5.73848188e-01 -3.74890924e-01 8.96896660e-01 6.23051763e-01 -4.15780067e-01 -1.76329166e-01 5.89913189e-01 -5.58846593e-02 -6.20479643e-01 -4.40570563e-01 -6.40445888e-01 9.29713249e-02 -3.54387730e-01 -3.20908725e-02 4.62877482e-01 1.35974422e-01 -5.44048965e-01 -1.14912964e-01 4.45521116e-01 3.22602779e-01 -1.32846087e-01 1.15610290e+00 -8.78026802e-03 -5.88172814e-03 8.48261952e-01 1.12383008e+00 -1.23176612e-01 -1.46266735e+00 -8.94676894e-02 9.63529199e-02 -2.73583591e-01 3.01862538e-01 -5.69384515e-01 -5.97489417e-01 8.38714778e-01 2.28345096e-01 3.87895405e-02 5.84960520e-01 9.34510306e-02 5.69915593e-01 3.78614306e-01 3.76242071e-01 -1.42416847e+00 -2.49239862e-01 4.10135359e-01 9.20735955e-01 -1.29967284e+00 1.87224254e-01 -5.31100333e-01 -5.11364043e-01 1.28039634e+00 6.74215972e-01 -3.60713489e-02 1.23039317e+00 4.70471978e-01 -5.95066011e-01 -1.04979165e-01 -4.78179842e-01 3.13217908e-01 3.00109535e-01 3.16993296e-01 2.15197891e-01 2.35662371e-01 -1.89800471e-01 8.84604752e-01 -2.32342303e-01 -1.97686985e-01 2.54905283e-01 8.37610662e-01 -6.76194787e-01 -1.06404531e+00 -6.03485405e-01 5.59690177e-01 8.51712227e-02 3.21328938e-02 1.89038664e-01 8.41215789e-01 -7.50444531e-02 7.12106466e-01 -6.28171712e-02 -2.81439424e-01 1.64681196e-01 5.29261649e-01 3.67778867e-01 -4.33852494e-01 9.12016332e-02 -5.30882701e-02 -7.15694800e-02 -1.32797882e-01 -2.45959044e-01 -8.22318971e-01 -1.67062175e+00 -5.29279709e-01 -5.67548931e-01 5.45296490e-01 9.82082427e-01 1.18103588e+00 -1.18692636e-01 4.83753443e-01 6.73524797e-01 -8.74062836e-01 -8.60670269e-01 -1.06121194e+00 -1.15989780e+00 1.98971570e-01 -1.94676574e-02 -1.19178867e+00 -8.51978660e-01 -5.72624385e-01]
[6.368881702423096, 3.473710536956787]
79ee40f5-64a8-4ce2-8e1a-9e6fc8e32cc2
world-to-words-grounded-open-vocabulary
2306.08685
null
https://arxiv.org/abs/2306.08685v1
https://arxiv.org/pdf/2306.08685v1.pdf
World-to-Words: Grounded Open Vocabulary Acquisition through Fast Mapping in Vision-Language Models
The ability to connect language units to their referents in the physical world, referred to as grounding, is crucial to learning and understanding grounded meanings of words. While humans demonstrate fast mapping in new word learning, it remains unclear whether modern vision-language models can truly represent language with their grounded meanings and how grounding may further bootstrap new word learning. To this end, we introduce Grounded Open Vocabulary Acquisition (GOVA) to examine grounding and bootstrapping in open-world language learning. As an initial attempt, we propose object-oriented BERT (OctoBERT), a novel visually-grounded language model by pre-training on image-text pairs highlighting grounding as an objective. Through extensive experiments and analysis, we demonstrate that OctoBERT is a more coherent and fast grounded word learner, and that the grounding ability acquired during pre-training helps the model to learn unseen words more rapidly and robustly. Our code is available at https://github.com/sled-group/world-to-words
['Joyce Chai', 'Jiayi Pan', 'Ziqiao Ma']
2023-06-14
null
null
null
null
['grounded-open-vocabulary-acquisition']
['natural-language-processing']
[ 9.91790444e-02 2.46037871e-01 -2.06236113e-02 -2.02660784e-01 -4.39029843e-01 -7.23296404e-01 7.48387098e-01 2.59154975e-01 -4.55500871e-01 5.41804492e-01 1.98008522e-01 -4.75112259e-01 1.28132448e-01 -8.54749382e-01 -9.97490287e-01 -2.05896899e-01 -3.04255873e-01 4.54496771e-01 1.44156098e-01 -5.74689746e-01 1.96344107e-01 2.32363313e-01 -1.40580773e+00 1.78005055e-01 1.05057728e+00 1.61948025e-01 7.56872118e-01 6.71376705e-01 -6.31056130e-01 6.65518045e-01 -2.72532463e-01 -1.95697963e-01 3.03030550e-01 -5.04530668e-01 -1.01871848e+00 -2.89303839e-01 7.84244835e-01 -4.72518772e-01 -3.41284037e-01 9.62303042e-01 1.89830020e-01 3.85072917e-01 3.58575910e-01 -1.17192030e+00 -1.60558677e+00 7.94615388e-01 -2.72229612e-01 2.49040291e-01 5.18890917e-01 2.80132353e-01 1.24104559e+00 -1.40771508e+00 9.42728758e-01 1.37100816e+00 5.52047610e-01 9.60220575e-01 -1.14960396e+00 -7.73921251e-01 4.78712052e-01 1.84112296e-01 -1.56061864e+00 -3.38645935e-01 6.02112591e-01 -4.73272264e-01 1.08543420e+00 -2.37169772e-01 1.35855830e+00 1.17490101e+00 1.99141219e-01 7.86047101e-01 1.22864938e+00 -8.32531929e-01 1.69988737e-01 5.83894132e-03 3.49039495e-01 1.21527529e+00 7.40764797e-01 4.53874558e-01 -1.28717959e+00 3.90817612e-01 9.08684731e-01 -2.13656738e-01 -3.70098472e-01 -7.75728524e-01 -1.75594687e+00 7.08760738e-01 1.04524302e+00 3.21212173e-01 -1.55305773e-01 6.92942679e-01 -3.02123159e-01 2.01757818e-01 2.19259426e-01 8.38390768e-01 -6.88309669e-02 1.67639494e-01 -7.93005943e-01 -1.96373295e-02 3.01264435e-01 1.06329834e+00 1.07279634e+00 2.18307853e-01 -3.44192609e-02 4.35015500e-01 5.85416496e-01 7.86060870e-01 4.09144133e-01 -6.54246747e-01 8.57547969e-02 4.27341074e-01 -2.14707509e-01 -5.94566047e-01 -1.06852837e-01 -4.81756181e-01 -2.96352237e-01 3.52913499e-01 3.47225159e-01 1.62643358e-01 -1.15064573e+00 2.07490921e+00 1.33047029e-01 4.97014791e-01 3.58686835e-01 7.74162054e-01 1.10747385e+00 7.30912864e-01 2.62612283e-01 1.34880945e-01 1.35954392e+00 -8.47829103e-01 -6.10931635e-01 -5.59594393e-01 7.08323538e-01 -4.58612978e-01 1.59313476e+00 2.75501341e-01 -9.26602483e-01 -7.29738533e-01 -1.31238794e+00 -5.12826383e-01 -8.36336613e-01 -7.14290857e-01 7.71138370e-01 3.96457493e-01 -1.40755975e+00 3.43589783e-01 -7.56293058e-01 -8.34797859e-01 8.65890503e-01 6.62978664e-02 -4.20963168e-01 -1.45273700e-01 -1.25816655e+00 1.12693059e+00 6.71963036e-01 -4.67111617e-01 -1.25478518e+00 -8.04510355e-01 -1.06617296e+00 -3.01327348e-01 2.33048141e-01 -1.20949090e+00 1.30492318e+00 -7.79776275e-01 -1.03519988e+00 1.23006320e+00 -1.70466110e-01 -4.86755639e-01 -5.34641221e-02 -3.92697036e-01 -1.09944567e-01 1.97847173e-01 1.12351917e-01 1.58192909e+00 7.72394001e-01 -1.81765556e+00 -3.44201744e-01 -2.08369330e-01 3.76491278e-01 2.96812087e-01 -4.77769613e-01 -4.52938467e-01 -3.87481898e-02 -6.16119802e-01 2.33512536e-01 -7.69339442e-01 9.18928757e-02 4.84153718e-01 -1.25553772e-01 -1.83129460e-01 2.78868139e-01 -4.83399540e-01 8.23194981e-01 -1.97685921e+00 1.64463609e-01 -3.94021086e-02 7.57014096e-01 1.12116240e-01 -5.75035155e-01 5.24022639e-01 3.97712504e-03 2.53929615e-01 -1.31821306e-02 -5.30286878e-02 -1.03391938e-01 5.47472000e-01 -6.68384433e-01 8.61163512e-02 2.66519308e-01 1.61673415e+00 -1.59682775e+00 -5.53379416e-01 2.05697522e-01 4.60055411e-01 -4.54415888e-01 8.40733126e-02 -5.25171578e-01 5.24432600e-01 -2.62557507e-01 3.32424760e-01 3.90318543e-01 -4.37385559e-01 -1.43454105e-01 -1.09010432e-02 -1.24876797e-01 1.14381589e-01 -8.37692976e-01 2.35585999e+00 -5.61955035e-01 1.05426693e+00 -4.58812416e-01 -7.19082832e-01 7.65488684e-01 5.86223416e-02 -1.83650166e-01 -9.53495264e-01 2.35086773e-02 2.02073753e-01 9.73611027e-02 -4.16018486e-01 7.44730413e-01 -4.77868885e-01 2.19902426e-01 5.26628196e-01 4.63075012e-01 -6.81082428e-01 6.76508620e-02 6.58516586e-01 5.73353589e-01 3.86286139e-01 5.20773113e-01 -4.69216615e-01 -2.19888408e-02 3.82071406e-01 -1.71984434e-01 8.61109495e-01 -6.47282675e-02 4.61820990e-01 -2.74540782e-01 -4.69569981e-01 -8.62357676e-01 -1.76559198e+00 -4.51875553e-02 1.36764669e+00 5.14557242e-01 -4.55656528e-01 -5.46637535e-01 -2.73991346e-01 -1.77180842e-01 1.20572591e+00 -5.41484416e-01 -4.03962195e-01 -2.62696296e-01 1.66693509e-01 4.09081459e-01 5.26246846e-01 4.27932918e-01 -1.22112119e+00 -8.48871887e-01 -8.52646232e-02 -5.83239347e-02 -9.67323303e-01 -3.96734148e-01 2.79861629e-01 -8.23685765e-01 -7.80127943e-01 -8.95009875e-01 -1.30561554e+00 9.17159021e-01 7.25401819e-01 1.47443628e+00 4.60291773e-01 -3.51506621e-01 9.41680551e-01 -4.95899022e-01 -8.32642496e-01 -5.28653800e-01 -1.77770421e-01 7.09775090e-02 -4.49228138e-01 3.88129085e-01 -3.73950452e-01 -3.64967734e-01 -2.18460768e-01 -9.39893961e-01 6.62191510e-01 3.82532477e-01 8.25853169e-01 5.36699593e-01 -6.91950977e-01 3.41014892e-01 -4.20170277e-01 7.90883362e-01 -2.44240209e-01 -5.40161908e-01 3.89321923e-01 -5.62230051e-01 4.08551097e-01 8.94272402e-02 -5.72513044e-01 -7.01399565e-01 -2.66240597e-01 4.44214698e-03 -2.11373359e-01 -1.58470236e-02 6.21212602e-01 2.22355813e-01 -1.88865736e-01 1.07822669e+00 4.67589796e-01 2.31857281e-02 -4.25740145e-02 1.23752022e+00 2.31876597e-01 6.40263855e-01 -8.22407484e-01 1.33830464e+00 6.09469056e-01 -2.54292041e-01 -1.02942038e+00 -1.10550237e+00 -3.06623518e-01 -7.87359059e-01 -2.18367547e-01 1.07726836e+00 -1.15649855e+00 -1.00591406e-01 2.47641191e-01 -1.21425140e+00 -8.11370850e-01 -8.46094310e-01 4.33964729e-01 -5.39792776e-01 2.41523743e-01 -1.04710897e-02 -4.79587555e-01 -2.21614838e-01 -6.91322982e-01 8.93078804e-01 3.28969598e-01 -4.18192178e-01 -1.30771220e+00 2.60732591e-01 9.35233161e-02 2.07390338e-01 -7.97443911e-02 9.35016811e-01 -4.52740550e-01 -8.74058723e-01 3.89666885e-01 -1.67911351e-01 1.01029657e-01 1.01731084e-01 -3.76339287e-01 -9.49344397e-01 -5.09121537e-01 -4.88578349e-01 -7.73073912e-01 9.37494934e-01 8.91076848e-02 7.39437878e-01 1.80130743e-03 -3.11396539e-01 7.38733351e-01 1.42710924e+00 -5.18112443e-02 5.74571788e-01 3.26630861e-01 7.47782230e-01 3.85078609e-01 3.06974709e-01 4.50370386e-02 5.39395869e-01 1.99719280e-01 3.26999068e-01 -2.85847515e-01 -7.46730864e-01 -7.46633112e-01 3.79620433e-01 1.08856463e+00 1.63235113e-01 -3.37588161e-01 -1.34565413e+00 8.57620597e-01 -1.30288506e+00 -8.64401817e-01 2.33831659e-01 2.01633668e+00 1.13395250e+00 2.71642625e-01 -3.87772888e-01 -3.20370346e-01 4.36995506e-01 8.27409625e-02 -6.46008432e-01 -1.97385773e-01 -3.66118491e-01 6.61111236e-01 -1.82850398e-02 9.16490316e-01 -4.20005739e-01 1.62227321e+00 6.32441711e+00 4.91379589e-01 -1.06228483e+00 2.61683464e-01 7.75307268e-02 4.77152839e-02 -8.18370998e-01 2.19409034e-01 -5.31689882e-01 -4.11947489e-01 4.67966974e-01 -5.19240022e-01 6.60999119e-01 4.47102696e-01 -1.69706106e-01 1.40565261e-01 -1.37874579e+00 1.20974863e+00 2.82391250e-01 -1.62768507e+00 6.82619452e-01 -2.23521423e-02 8.73132825e-01 1.63051695e-01 2.58638918e-01 3.99916887e-01 6.73241973e-01 -1.25949359e+00 1.10683215e+00 5.44007421e-01 9.49081302e-01 -4.28994596e-02 -2.25030947e-02 1.19490430e-01 -1.28947091e+00 1.56657249e-01 -3.36444199e-01 -3.23994160e-01 7.98356384e-02 1.66806113e-02 -8.84367466e-01 1.99387833e-01 4.43255723e-01 8.16910744e-01 -7.76881337e-01 7.11595654e-01 -6.93681717e-01 5.41435301e-01 -9.76121798e-02 -3.05149078e-01 2.35077232e-01 1.95964545e-01 5.32753587e-01 7.72246897e-01 3.83273900e-01 2.33534247e-01 2.04991713e-01 1.22054708e+00 -1.77325025e-01 2.49777690e-01 -9.10785377e-01 -3.00803602e-01 4.84448910e-01 8.85280192e-01 -7.99257994e-01 -4.78740066e-01 -5.05620241e-01 9.21683729e-01 5.50371826e-01 5.99211693e-01 -5.43502033e-01 -1.37671739e-01 6.57829165e-01 1.91793934e-01 1.16170339e-01 -8.15916181e-01 -2.24752948e-01 -1.36868131e+00 -3.17233741e-01 -7.98076689e-01 5.16536422e-02 -1.54258287e+00 -1.19000769e+00 4.57178712e-01 2.24051952e-01 -9.94727969e-01 1.89156085e-01 -1.04873037e+00 -3.29275161e-01 7.29613006e-01 -1.51991868e+00 -1.33915627e+00 -7.21735060e-01 7.19224215e-01 7.78423429e-01 -2.08768755e-01 8.96925807e-01 -5.04167199e-01 3.50675099e-02 4.98308599e-01 -1.69346958e-01 1.33003309e-01 5.99251986e-01 -1.30255532e+00 9.13291037e-01 9.13469017e-01 1.18157530e+00 1.13902891e+00 7.14453936e-01 -7.56949008e-01 -1.47492397e+00 -8.26378882e-01 6.58283770e-01 -7.64743507e-01 8.49693358e-01 -7.12710440e-01 -1.05411625e+00 8.67331207e-01 6.38749421e-01 2.80311964e-02 7.17907667e-01 1.87076166e-01 -8.07927907e-01 2.10038215e-01 -6.23834074e-01 1.08187294e+00 1.57428837e+00 -7.98484445e-01 -1.38966238e+00 3.89568061e-01 1.11003625e+00 -2.84511149e-01 -4.02070761e-01 -1.20407924e-01 5.33801794e-01 -6.51343226e-01 1.25612605e+00 -6.65657282e-01 1.35468170e-01 -3.85965049e-01 -3.17066461e-01 -1.53334773e+00 -3.99604797e-01 -6.26316249e-01 -6.88475147e-02 7.25455165e-01 3.53583485e-01 -7.12051988e-01 4.46737736e-01 -2.26796083e-02 -1.83636323e-01 -4.92107362e-01 -8.46879482e-01 -1.02927744e+00 7.28222668e-01 -6.81285083e-01 4.98921961e-01 1.06425011e+00 1.22093655e-01 6.18966520e-01 1.12771660e-01 1.68709308e-01 6.84012175e-01 -1.14329852e-01 6.81617975e-01 -1.31643999e+00 -6.40001744e-02 -3.59638184e-01 -3.87896359e-01 -1.37960291e+00 2.24503487e-01 -1.44531655e+00 1.96976289e-01 -2.11215758e+00 1.97068349e-01 -2.04759836e-01 -4.26008642e-01 7.22156048e-01 -2.20932990e-01 3.45833272e-01 5.18258393e-01 9.93385687e-02 -7.87288725e-01 5.93858778e-01 1.69664288e+00 -3.85402411e-01 -1.49381340e-01 -9.96118903e-01 -8.74467611e-01 7.10454166e-01 7.36543655e-01 -2.62544930e-01 -8.23341072e-01 -8.96807492e-01 6.67275727e-01 -5.40391386e-01 7.68031836e-01 -1.11097336e+00 2.25490868e-01 -2.51206756e-01 2.96251714e-01 -3.76455694e-01 1.92888200e-01 -5.12456417e-01 -2.95901656e-01 6.91233635e-01 -4.98695314e-01 3.30551229e-02 5.41137457e-01 3.93413216e-01 2.08121926e-01 -2.09460482e-01 6.22821629e-01 -4.37424809e-01 -1.50824952e+00 1.40505344e-01 -3.77783626e-01 6.82267189e-01 9.36436176e-01 -6.06373787e-01 -5.00033081e-01 -3.69872659e-01 -7.59835839e-01 2.82042950e-01 6.79985404e-01 7.85135627e-01 1.10077763e+00 -1.33254623e+00 -7.56255090e-01 1.71298951e-01 7.44147718e-01 -1.77707765e-02 -7.34145939e-02 1.69028476e-01 -6.75574541e-01 3.46939147e-01 -3.18111628e-01 -7.24221945e-01 -9.63249862e-01 8.25955510e-01 3.94973397e-01 2.95601487e-01 -7.64630139e-01 1.38838351e+00 6.08522475e-01 -2.62265563e-01 8.13978836e-02 -7.46018708e-01 1.11175636e-02 -4.49894182e-02 4.63763148e-01 -2.06561625e-01 -4.59359050e-01 -6.35946512e-01 -2.47291490e-01 1.02768469e+00 -4.45682444e-02 -4.02173519e-01 8.58833849e-01 -2.45740041e-01 1.13966633e-02 8.22886705e-01 7.96706140e-01 -7.60270655e-02 -1.22265542e+00 -4.82104570e-01 -6.88247383e-02 -3.43484610e-01 -4.59735375e-03 -7.51915634e-01 -4.33300555e-01 1.24168289e+00 7.88370609e-01 -1.61700621e-01 7.26120591e-01 4.53732908e-01 7.91991293e-01 8.09612155e-01 7.26389289e-01 -6.85704231e-01 8.55631530e-01 7.26677239e-01 1.16786444e+00 -1.24696159e+00 -4.46933061e-02 -1.96439147e-01 -3.62805724e-01 9.67456937e-01 7.65389264e-01 6.73862081e-03 5.59358239e-01 -1.44086286e-01 4.06992972e-01 -3.23305041e-01 -6.39417827e-01 -6.38700783e-01 4.33555931e-01 1.13234603e+00 3.35203081e-01 -1.03363477e-01 3.08332324e-01 -2.56035253e-02 -6.55932903e-01 1.58783905e-02 4.63623762e-01 9.55447018e-01 -8.37244093e-01 -8.62849891e-01 -3.77578706e-01 -2.95282472e-02 3.86616707e-01 -7.11560667e-01 -4.02039737e-01 8.75475943e-01 4.00839239e-01 7.94665754e-01 1.19014651e-01 -1.61461383e-01 1.53289974e-01 3.33587945e-01 9.83663082e-01 -1.18380654e+00 -6.91523179e-02 -5.31450093e-01 -2.72816867e-01 -2.80091286e-01 -2.75371701e-01 -3.20232034e-01 -1.86314034e+00 -2.89508644e-02 -3.19089085e-01 -1.76837686e-02 3.96162778e-01 9.74744678e-01 2.00117826e-01 5.98369002e-01 -1.81966111e-01 -6.46644354e-01 -2.76838213e-01 -6.11244500e-01 -2.27529213e-01 5.31799853e-01 6.07998967e-01 -8.14596415e-01 -3.91425639e-02 4.21877116e-01]
[10.642595291137695, 1.8430994749069214]
12d7feff-d131-425a-9645-51d5780a10f2
data-uncertainty-guided-multi-phase-learning
2103.16368
null
https://arxiv.org/abs/2103.16368v1
https://arxiv.org/pdf/2103.16368v1.pdf
Data-Uncertainty Guided Multi-Phase Learning for Semi-Supervised Object Detection
In this paper, we delve into semi-supervised object detection where unlabeled images are leveraged to break through the upper bound of fully-supervised object detection models. Previous semi-supervised methods based on pseudo labels are severely degenerated by noise and prone to overfit to noisy labels, thus are deficient in learning different unlabeled knowledge well. To address this issue, we propose a data-uncertainty guided multi-phase learning method for semi-supervised object detection. We comprehensively consider divergent types of unlabeled images according to their difficulty levels, utilize them in different phases and ensemble models from different phases together to generate ultimate results. Image uncertainty guided easy data selection and region uncertainty guided RoI Re-weighting are involved in multi-phase learning and enable the detector to concentrate on more certain knowledge. Through extensive experiments on PASCAL VOC and MS COCO, we demonstrate that our method behaves extraordinarily compared to baseline approaches and outperforms them by a large margin, more than 3% on VOC and 2% on COCO.
['Shengjin Wang', 'Lu Fang', 'Ye Guo', 'YaLi Li', 'Zhenyu Wang']
2021-03-29
null
http://openaccess.thecvf.com//content/CVPR2021/html/Wang_Data-Uncertainty_Guided_Multi-Phase_Learning_for_Semi-Supervised_Object_Detection_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Wang_Data-Uncertainty_Guided_Multi-Phase_Learning_for_Semi-Supervised_Object_Detection_CVPR_2021_paper.pdf
cvpr-2021-1
['semi-supervised-object-detection']
['computer-vision']
[ 1.51092723e-01 2.01484114e-01 -4.43283528e-01 -5.68955958e-01 -1.23345208e+00 -8.05543363e-01 5.00987411e-01 -3.57303098e-02 -4.53885823e-01 7.18219221e-01 -1.87063277e-01 -1.90211624e-01 1.91830024e-01 -1.20887443e-01 -6.38877988e-01 -7.52201796e-01 3.92429113e-01 7.59442925e-01 5.66687524e-01 3.96310180e-01 1.58097848e-01 3.40782195e-01 -1.48535275e+00 1.19797409e-01 9.72457528e-01 8.27604771e-01 1.62050441e-01 4.64030087e-01 -6.42359704e-02 8.40304673e-01 -4.46243167e-01 -1.71381563e-01 3.05312008e-01 -8.17084983e-02 -8.04947197e-01 6.31563485e-01 6.32636249e-01 -3.05503398e-01 1.35917544e-01 1.29763544e+00 4.54083115e-01 -6.92906305e-02 1.09796274e+00 -1.15621710e+00 -3.64919662e-01 7.70097196e-01 -9.80872631e-01 1.54342368e-01 -1.26807317e-02 2.79741079e-01 9.84117806e-01 -1.42963588e+00 6.04876876e-01 1.19537115e+00 5.59600294e-01 5.77181458e-01 -1.43137932e+00 -6.20819271e-01 3.64009470e-01 -3.28575186e-02 -1.54319322e+00 -4.25466686e-01 7.37155080e-01 -5.98661482e-01 3.10232341e-01 8.06874782e-02 1.43136322e-01 9.15968120e-01 -3.52742612e-01 1.36016333e+00 1.50283253e+00 -5.38426459e-01 2.20596179e-01 6.71772003e-01 2.93192506e-01 8.07205856e-01 5.09625673e-01 4.11068857e-01 -4.49622661e-01 -2.08118483e-02 4.46508586e-01 -1.54869393e-01 -7.86961764e-02 -6.87757730e-01 -1.16680980e+00 6.39658689e-01 5.18879533e-01 -1.16652191e-01 -1.37303118e-03 -4.84848581e-02 2.66360134e-01 -9.64217708e-02 4.04059917e-01 4.40573603e-01 -3.90088350e-01 4.17190254e-01 -1.18505681e+00 -1.02432311e-01 5.27105033e-01 1.14157689e+00 8.88523042e-01 5.81563860e-02 -4.75555599e-01 8.13073218e-01 6.22896969e-01 5.36663473e-01 2.43789539e-01 -8.65985692e-01 2.03009695e-01 6.83441818e-01 3.13148946e-01 -3.87703717e-01 -3.68089050e-01 -6.45576000e-01 -3.91898245e-01 5.51247358e-01 6.37855113e-01 -1.68829262e-01 -1.46573865e+00 1.30104721e+00 5.43045580e-01 -2.14699693e-02 -8.54439363e-02 9.68142748e-01 8.40955973e-01 2.72696555e-01 2.28139356e-01 -2.83352047e-01 1.24685919e+00 -1.42195833e+00 -4.86387610e-01 -3.75410736e-01 6.77971303e-01 -7.72153378e-01 9.00151134e-01 4.76649255e-01 -7.10681617e-01 -7.91347384e-01 -1.35124505e+00 2.60689169e-01 -2.85062522e-01 5.28041840e-01 4.22201902e-01 7.93685377e-01 -5.99939883e-01 3.80897790e-01 -6.83902204e-01 -8.97496119e-02 8.51820469e-01 1.52173117e-01 -1.82917595e-01 -2.04961315e-01 -6.33039296e-01 8.83368552e-01 8.83459151e-01 8.74727443e-02 -1.31095552e+00 -4.47166771e-01 -6.87051535e-01 -3.84651750e-01 9.07289863e-01 -3.52260798e-01 1.13881707e+00 -8.67737293e-01 -9.98801529e-01 1.12409377e+00 1.44354463e-01 -5.05304813e-01 7.86846817e-01 -3.32331508e-01 -1.97712958e-01 1.58296898e-01 3.34782869e-01 1.04799366e+00 1.03794837e+00 -1.73564517e+00 -9.63726461e-01 -3.88987720e-01 -3.20007689e-02 3.83238047e-01 7.22192973e-02 3.89724448e-02 -5.81397116e-01 -5.35873353e-01 3.03405672e-01 -9.46698725e-01 -4.46446717e-01 1.40160367e-01 -7.70101905e-01 -2.12620571e-01 9.48978126e-01 4.32168087e-03 8.55748892e-01 -1.87651205e+00 -1.37794971e-01 4.44468744e-02 3.23686332e-01 3.83025378e-01 8.58018547e-02 -2.71418631e-01 2.58971721e-01 1.79912582e-01 -2.94240743e-01 -5.89235365e-01 -7.53756240e-02 3.03081661e-01 -7.92495683e-02 6.03129685e-01 4.76660818e-01 9.50940013e-01 -1.10281777e+00 -1.20154536e+00 4.16087419e-01 6.63126186e-02 -8.94488916e-02 2.75717825e-01 -3.27281862e-01 4.69876140e-01 -6.36597872e-01 1.29907417e+00 8.37611854e-01 -4.32190210e-01 -1.04225837e-01 -3.27247947e-01 3.24555449e-02 -3.03780437e-01 -1.43952239e+00 1.47373235e+00 -7.96219613e-03 3.53018403e-01 1.22226149e-01 -7.81355381e-01 6.77394629e-01 1.00897051e-01 2.49508291e-01 3.47485803e-02 2.16243386e-01 3.63294542e-01 -2.32688129e-01 -2.67572910e-01 3.41102868e-01 5.42342104e-02 1.41524360e-01 5.64976215e-01 3.93278271e-01 -3.61843318e-01 2.98883855e-01 3.09394211e-01 6.04165018e-01 4.38464373e-01 4.22652304e-01 -3.61849129e-01 3.91675413e-01 2.50506759e-01 6.23108566e-01 1.06991732e+00 -8.52971017e-01 1.01253593e+00 2.04884127e-01 -2.65837014e-01 -8.57179761e-01 -1.14857924e+00 -6.36824131e-01 1.09701288e+00 4.64764893e-01 1.27285048e-01 -7.61177301e-01 -1.38017476e+00 5.43269627e-02 6.28149092e-01 -6.78653598e-01 -9.30504501e-03 -8.92104506e-02 -9.39593971e-01 4.47795540e-01 7.35579073e-01 3.70886445e-01 -8.48302484e-01 -3.32345009e-01 5.01994155e-02 -1.28840106e-02 -1.29929674e+00 -4.16922897e-01 7.64741778e-01 -7.52087593e-01 -1.20714903e+00 -8.41252685e-01 -7.59491742e-01 9.92511749e-01 4.79414076e-01 1.11027920e+00 -1.31330863e-01 -5.24517894e-01 3.95950228e-01 -5.06456554e-01 -6.15033090e-01 -4.65379357e-01 -1.03910409e-01 3.42905343e-01 1.69382066e-01 3.97415608e-01 -6.21324591e-02 -5.80612957e-01 8.06475461e-01 -7.70458519e-01 -1.64954469e-01 8.49478602e-01 1.08210719e+00 9.44766164e-01 -1.46353751e-01 5.56255877e-01 -1.20762479e+00 -6.16964325e-02 -3.33777130e-01 -7.40511000e-01 5.42449415e-01 -1.03305113e+00 1.88580558e-01 1.25681847e-01 -5.66895008e-01 -1.32618392e+00 7.98098922e-01 3.04411203e-01 -6.01140440e-01 -2.22435042e-01 -1.23347312e-01 -1.60941258e-01 -3.75603735e-01 1.09575438e+00 4.85657565e-02 -2.75425732e-01 -2.83180416e-01 7.02083588e-01 8.03239942e-01 5.30467451e-01 -6.63387835e-01 1.00131965e+00 8.33222032e-01 -4.51544136e-01 -3.79328996e-01 -1.43865037e+00 -9.00214672e-01 -9.93416011e-01 -4.33814436e-01 7.05435216e-01 -1.14036477e+00 1.01092897e-01 2.28912786e-01 -7.78317153e-01 -2.42999997e-02 -5.12175143e-01 5.41886151e-01 -4.88132954e-01 5.25985718e-01 -3.05221200e-01 -1.11040294e+00 -1.57469437e-01 -1.29415309e+00 1.27557480e+00 4.37417537e-01 1.75447330e-01 -7.56336033e-01 -1.77327082e-01 7.61285067e-01 6.11429214e-02 8.33064690e-02 1.06853843e-01 -1.02392566e+00 -7.06207097e-01 -3.29242051e-01 -6.01934671e-01 5.07194102e-01 -2.13330165e-01 1.12494072e-02 -1.42530549e+00 -2.97675669e-01 -3.77918817e-02 -1.04732680e+00 1.30605078e+00 4.01537955e-01 1.00873733e+00 2.59858072e-01 -5.46503663e-01 4.61355150e-01 1.37152922e+00 -2.97853470e-01 1.14917621e-01 9.07716453e-02 7.71359026e-01 7.22358167e-01 1.26593697e+00 2.87094682e-01 6.47529215e-02 3.55607212e-01 6.56203747e-01 -1.55467138e-01 -3.59888166e-01 -3.12044948e-01 2.31245264e-01 1.89240813e-01 1.02971345e-01 -9.07883793e-02 -7.91657269e-01 6.32488668e-01 -1.87120032e+00 -4.49246287e-01 -1.58532351e-01 2.03609633e+00 8.29121351e-01 5.61875999e-01 1.62160635e-01 3.15984935e-02 9.15794730e-01 1.87959503e-02 -8.04401278e-01 3.40522349e-01 -3.25557828e-01 -4.03861463e-01 8.43126416e-01 3.85872364e-01 -1.49060011e+00 1.25514317e+00 6.62794161e+00 1.23155487e+00 -6.76292360e-01 2.40961313e-01 9.40741479e-01 -4.35412005e-02 -2.40111891e-02 1.10621355e-01 -1.22452545e+00 1.66228756e-01 3.26248914e-01 2.45339319e-01 -9.59225520e-02 1.15178418e+00 -1.60867214e-01 -4.83585745e-01 -1.21531379e+00 1.02113104e+00 6.39629364e-02 -9.29774761e-01 -1.09295577e-01 -1.97448343e-01 1.16466033e+00 1.93041369e-01 4.69770432e-02 3.67026865e-01 6.83199525e-01 -1.04058874e+00 7.39589453e-01 1.60031080e-01 7.21659064e-01 -3.47406894e-01 7.52324581e-01 4.98945832e-01 -1.05536699e+00 -1.06086656e-01 -5.06381512e-01 4.60447580e-01 2.89857447e-01 7.94824898e-01 -8.85623395e-01 3.93107533e-01 6.34626210e-01 3.53459209e-01 -7.81159639e-01 1.01520014e+00 -3.89297575e-01 7.45485067e-01 -3.80328327e-01 2.89689493e-03 2.61023015e-01 1.45700455e-01 5.03416181e-01 1.19427884e+00 -1.61700472e-01 -1.31041974e-01 5.84244847e-01 1.08883882e+00 -3.45704582e-04 8.19414482e-02 -3.24447453e-01 6.58198968e-02 4.91155684e-01 1.59162366e+00 -9.81144428e-01 -4.70409662e-01 -2.89922804e-01 7.50423789e-01 3.95895362e-01 4.41902131e-01 -8.61927927e-01 4.23970707e-02 -2.44600356e-01 -7.82890990e-02 2.02156112e-01 6.47505969e-02 -4.13644284e-01 -1.19277608e+00 -6.04435578e-02 -5.62600613e-01 4.83493835e-01 -6.94191456e-01 -1.35489202e+00 6.59417212e-01 2.90778633e-02 -1.32206297e+00 -6.61691204e-02 -8.69811058e-01 -3.84301424e-01 5.69058895e-01 -1.81748891e+00 -1.52964687e+00 -1.50988117e-01 3.47728401e-01 8.06063354e-01 -1.49329975e-01 4.48604465e-01 1.42710775e-01 -6.03065014e-01 5.43741405e-01 1.97845966e-01 1.11546077e-01 1.00465977e+00 -1.68021166e+00 -1.42167926e-01 9.67390537e-01 6.69024408e-01 2.22601116e-01 6.21739209e-01 -4.94149655e-01 -9.18418407e-01 -1.21008515e+00 3.46064895e-01 -8.51016402e-01 5.74881434e-01 -2.62250632e-01 -6.80290341e-01 4.87955153e-01 -1.45623133e-01 6.59802496e-01 4.94425356e-01 9.64971632e-02 -4.86558288e-01 3.77722830e-02 -1.22281766e+00 3.95993322e-01 8.77279341e-01 -4.00893152e-01 -6.41802430e-01 5.85285306e-01 6.31180763e-01 -3.86585176e-01 -5.58849871e-01 8.27185810e-01 3.49846154e-01 -9.76026416e-01 8.12763333e-01 -4.83771950e-01 -6.32657036e-02 -7.13160038e-01 -5.43037951e-02 -1.00824440e+00 -1.66810974e-01 -3.44537079e-01 -1.88426584e-01 1.30700517e+00 6.49536788e-01 -1.87482715e-01 1.07586217e+00 3.99163902e-01 6.79104328e-02 -6.91928685e-01 -7.38133073e-01 -7.85549283e-01 -2.23908752e-01 -5.24316549e-01 -1.79323688e-01 7.39679694e-01 -2.28724062e-01 4.15150285e-01 -4.05862957e-01 2.82269925e-01 1.15203071e+00 2.65453279e-01 7.88125634e-01 -1.18942475e+00 -2.80597866e-01 -3.35389614e-01 -1.80825666e-01 -1.20639658e+00 -1.04064941e-01 -6.88124061e-01 6.04116499e-01 -1.06303322e+00 8.22200656e-01 -6.79184020e-01 -4.78062183e-01 3.86295199e-01 -5.25225043e-01 6.64086998e-01 -3.80801372e-02 5.55479825e-01 -1.18530428e+00 3.06304663e-01 1.16782868e+00 -3.37476790e-01 -2.61353068e-02 1.30161628e-01 -7.21551538e-01 9.63467240e-01 4.91877615e-01 -6.28882408e-01 -3.11577886e-01 -6.04134845e-03 -1.75107196e-01 -3.08134854e-01 2.44613498e-01 -1.04170632e+00 2.61245728e-01 -8.96104127e-02 4.05746222e-01 -9.27945137e-01 9.06847492e-02 -8.67515802e-01 -4.37389433e-01 3.58068764e-01 -3.44424397e-01 -7.63762951e-01 -7.46613145e-02 9.53178644e-01 -2.40198076e-01 -5.81869364e-01 1.09620166e+00 -1.50471002e-01 -9.38319743e-01 3.70134860e-01 -3.40647181e-03 1.33931354e-01 1.25197017e+00 -3.83254588e-01 -2.67665694e-03 -2.81732697e-02 -1.03230035e+00 6.99207306e-01 3.92640620e-01 2.82503188e-01 3.87818933e-01 -1.17208672e+00 -7.47737408e-01 -5.95314875e-02 6.61918879e-01 3.71020526e-01 1.61502928e-01 7.82174170e-01 -3.15143436e-01 3.73496473e-01 1.45199373e-01 -1.14809394e+00 -1.25939536e+00 6.91497505e-01 4.01282072e-01 -3.97804528e-01 -1.18040711e-01 1.11189318e+00 3.27577591e-01 -5.77858448e-01 6.07805550e-01 -7.77656585e-02 -2.57589996e-01 1.71398863e-01 4.29594666e-01 2.58990854e-01 -2.66124129e-01 -6.26706004e-01 -1.93041652e-01 5.21902323e-01 -3.56685311e-01 -8.12735967e-03 7.02799559e-01 -3.20498109e-01 4.35587615e-01 5.23532867e-01 8.55198741e-01 -1.19665572e-02 -1.76912856e+00 -6.26040161e-01 2.25489393e-01 -3.73168409e-01 9.26534906e-02 -9.10987735e-01 -8.42174947e-01 7.58546293e-01 8.61169696e-01 -1.29879341e-01 6.86860859e-01 3.09890628e-01 9.44295973e-02 3.90032351e-01 3.90229017e-01 -1.47378552e+00 4.43168670e-01 1.80371508e-01 4.23416376e-01 -2.11504388e+00 2.85294741e-01 -5.46168506e-01 -8.92954886e-01 8.54485869e-01 9.32308555e-01 -4.96359356e-02 5.50223172e-01 3.52604598e-01 3.49244475e-01 -1.31353006e-01 -3.87101561e-01 -7.00880885e-01 5.46735108e-01 6.79282486e-01 1.05572313e-01 -1.94123909e-02 -1.78105772e-01 3.62868071e-01 5.72437108e-01 -1.26972467e-01 3.26432317e-01 9.40011561e-01 -1.02486563e+00 -9.10509825e-01 -7.05909669e-01 4.52494621e-01 -4.77416039e-01 3.40016792e-03 -3.27968866e-01 7.97585785e-01 2.08424777e-01 8.85139048e-01 -3.94926220e-01 -1.91864491e-01 -1.01143643e-01 4.00522463e-02 3.78692627e-01 -9.22928810e-01 -1.42008141e-01 2.68956989e-01 -3.06910202e-02 -2.63620317e-01 -7.22326577e-01 -6.34279907e-01 -1.23288739e+00 5.63018024e-01 -1.16467476e+00 1.19014628e-01 6.13810599e-01 9.33272660e-01 4.59181257e-02 3.94935608e-01 7.15039253e-01 -9.77026284e-01 -1.07462180e+00 -1.11224258e+00 -7.03422189e-01 2.23801464e-01 9.45221409e-02 -8.09437335e-01 -6.96342945e-01 1.21223010e-01]
[9.202065467834473, 1.2710559368133545]
de415047-aa5c-46ce-b1f1-32a0c6f856a9
leveraging-deep-learning-techniques-on
2304.09282
null
https://arxiv.org/abs/2304.09282v1
https://arxiv.org/pdf/2304.09282v1.pdf
Leveraging Deep Learning Techniques on Collaborative Filtering Recommender Systems
With the exponentially increasing volume of online data, searching and finding required information have become an extensive and time-consuming task. Recommender Systems as a subclass of information retrieval and decision support systems by providing personalized suggestions helping users access what they need more efficiently. Among the different techniques for building a recommender system, Collaborative Filtering (CF) is the most popular and widespread approach. However, cold start and data sparsity are the fundamental challenges ahead of implementing an effective CF-based recommender. Recent successful developments in enhancing and implementing deep learning architectures motivated many studies to propose deep learning-based solutions for solving the recommenders' weak points. In this research, unlike the past similar works about using deep learning architectures in recommender systems that covered different techniques generally, we specifically provide a comprehensive review of deep learning-based collaborative filtering recommender systems. This in-depth filtering gives a clear overview of the level of popularity, gaps, and ignored areas on leveraging deep learning techniques to build CF-based systems as the most influential recommenders.
['Javad Mohammadzadeh', 'Ali Fallahi RahmatAbadi']
2023-04-18
null
null
null
null
['collaborative-filtering']
['miscellaneous']
[-4.21741158e-01 -4.97451544e-01 -3.83848727e-01 -6.19793415e-01 -2.82130599e-01 -5.01284182e-01 4.25976038e-01 7.95174111e-03 -3.36457044e-01 2.56862760e-01 6.20341480e-01 -4.89222825e-01 -8.66196156e-01 -1.00343323e+00 -1.63736045e-01 -3.94843727e-01 -4.53089289e-02 6.39748275e-01 3.36142406e-02 -7.87519395e-01 8.39079320e-01 5.28339028e-01 -1.86499631e+00 7.61603236e-01 1.10079348e+00 1.13312805e+00 3.16485345e-01 5.06707132e-01 -6.06684804e-01 7.40654349e-01 -5.25961161e-01 -4.38224971e-01 3.21839899e-01 -7.42392316e-02 -6.67298555e-01 -5.81395447e-01 5.61324716e-01 -4.07314181e-01 -5.47847092e-01 6.41897738e-01 5.99958539e-01 9.72881496e-01 6.44672573e-01 -9.64938700e-01 -1.53354299e+00 9.99261022e-01 -2.89222766e-02 5.52943170e-01 5.16888320e-01 -5.68090677e-01 1.35929251e+00 -1.14084101e+00 2.62185156e-01 8.92581880e-01 8.52700412e-01 6.02375805e-01 -5.00351131e-01 -7.32697964e-01 6.65099204e-01 6.85997844e-01 -1.00140762e+00 -9.84874740e-02 4.99240607e-01 -3.96546811e-01 1.20606840e+00 3.82216126e-01 8.18879068e-01 1.01818180e+00 2.57840216e-01 9.67245638e-01 4.17750299e-01 -1.30327493e-01 1.95538893e-01 1.63236856e-01 6.40165150e-01 1.61766991e-01 3.63636538e-02 3.14628512e-01 -2.61935592e-01 -2.87317038e-01 5.75693548e-01 1.08670413e+00 -1.13180608e-01 3.25062335e-03 -5.86032152e-01 1.22620094e+00 7.91208327e-01 5.87587774e-01 -3.08086485e-01 -1.65342733e-01 1.55490637e-01 8.91922891e-01 4.63942647e-01 9.40169811e-01 -6.72327340e-01 1.16422117e-01 -1.08853030e+00 4.30739045e-01 1.01661098e+00 6.78910196e-01 3.55141073e-01 7.37241432e-02 -1.18239395e-01 9.62794900e-01 6.30353451e-01 2.60568172e-01 5.94269395e-01 -8.72043431e-01 -1.11903854e-01 6.05272412e-01 2.81882286e-01 -1.05988669e+00 -4.84076172e-01 -1.05806613e+00 -8.59804332e-01 7.28399009e-02 4.21539433e-02 -2.90788740e-01 -4.88391966e-01 9.78555501e-01 1.06768847e-01 1.03481740e-01 -1.41000092e-01 1.11248422e+00 1.31978178e+00 7.01196074e-01 -6.49698451e-02 -5.46631962e-02 8.50856185e-01 -1.36370695e+00 -6.18904173e-01 6.65794536e-02 4.75193620e-01 -1.17218149e+00 8.44891250e-01 7.26001084e-01 -8.02886605e-01 -8.38536024e-01 -8.02052259e-01 -1.87473461e-01 -6.02605760e-01 1.46319166e-01 1.14844072e+00 7.50172555e-01 -1.19963193e+00 1.03856730e+00 -2.72093505e-01 -2.48193383e-01 3.42965990e-01 6.23702407e-01 7.61702210e-02 -2.72524655e-01 -1.39450312e+00 9.27682877e-01 -2.19518453e-01 7.40025416e-02 -7.19254911e-01 -1.02463663e+00 1.29864022e-01 4.07540113e-01 3.73257637e-01 -6.76698864e-01 1.30556011e+00 -9.27910209e-01 -1.60973179e+00 1.49200652e-02 2.83348054e-01 -3.94230783e-01 5.01647219e-02 -8.90085340e-01 -7.49506414e-01 -5.85610569e-01 -4.26759690e-01 5.81157357e-02 5.97035348e-01 -6.55797958e-01 -9.84882236e-01 -1.82025984e-01 4.93152171e-01 3.88102233e-01 -5.07101178e-01 3.20491105e-01 -2.25965604e-02 -4.77729470e-01 -1.90355182e-01 -6.33840263e-01 -5.82234800e-01 -2.42398918e-01 2.47423097e-01 -8.58993053e-01 5.55019557e-01 -1.47042543e-01 1.59085345e+00 -1.81248260e+00 -2.39511882e-03 1.20276168e-01 5.02761602e-01 6.00895524e-01 -3.04485291e-01 9.12257195e-01 2.69076109e-01 -2.91087441e-02 7.15653598e-01 3.05984821e-02 9.83414799e-02 2.03419521e-01 -5.19282699e-01 1.41161531e-01 -5.99646986e-01 7.67106295e-01 -1.11957169e+00 2.64145076e-01 3.11238736e-01 7.96028972e-01 -1.06850648e+00 4.30880100e-01 -1.68277174e-01 2.83868283e-01 -6.65279746e-01 3.50924462e-01 4.71241534e-01 -4.21517789e-01 7.17600286e-02 -1.54920131e-01 -3.29407871e-01 6.64868951e-01 -1.30534303e+00 1.62264550e+00 -7.17720628e-01 4.91209328e-01 -9.68113542e-02 -1.15620780e+00 1.10954940e+00 7.63395578e-02 6.46502376e-01 -7.65194893e-01 2.39394128e-01 1.81698769e-01 3.11651707e-01 -3.27996343e-01 9.11838114e-01 2.29550958e-01 3.73784333e-01 8.47146869e-01 1.27091378e-01 5.09880662e-01 9.71844718e-02 4.39550281e-01 1.08460510e+00 -1.40442297e-01 -3.98418494e-02 -3.44370753e-01 5.77259719e-01 -2.43631691e-01 3.51002067e-01 1.07644928e+00 2.00583786e-01 3.47396702e-01 -4.69245911e-01 -9.95515049e-01 -5.31105936e-01 -5.31073570e-01 1.67144090e-01 2.06064677e+00 -1.56292856e-01 -8.15405071e-01 -5.80648184e-02 -6.85271025e-01 1.91248104e-01 5.76018810e-01 -7.88911402e-01 -7.05779493e-02 -3.40046912e-01 -4.44662869e-01 -1.31689787e-01 7.23933578e-01 3.62210125e-02 -1.19528782e+00 -9.17919427e-02 3.80924821e-01 1.51254207e-01 -1.49807334e-01 -5.80846667e-01 1.87687963e-01 -9.03680861e-01 -1.17316413e+00 -7.21300542e-01 -6.62610471e-01 2.84414977e-01 9.91791844e-01 1.31299639e+00 6.56678021e-01 1.83147982e-01 2.91789353e-01 -9.17841554e-01 -2.38323897e-01 2.31588688e-02 2.75702983e-01 2.85897106e-01 -3.45140278e-01 7.36686409e-01 -4.98828441e-01 -1.09485531e+00 5.11740983e-01 -3.62256825e-01 -5.63882709e-01 5.46664357e-01 6.68602645e-01 1.72330275e-01 -1.39147013e-01 7.80602336e-01 -1.38476741e+00 1.22601652e+00 -9.10265326e-01 -3.97724599e-01 1.55287713e-01 -1.16282499e+00 -3.00302923e-01 9.70963717e-01 -4.62578923e-01 -9.81359184e-01 -4.72226739e-01 -7.11268902e-01 -2.14872032e-01 -7.87303597e-02 8.09756517e-01 3.37810248e-01 -1.58890948e-01 1.07611012e+00 -1.29491221e-02 -4.95700508e-01 -1.02275026e+00 7.80933738e-01 8.51126969e-01 -5.19214012e-02 -3.24199557e-01 3.18731248e-01 4.59213927e-02 -5.79339862e-01 -5.04373968e-01 -1.38786387e+00 -9.68958914e-01 -5.61617970e-01 -2.55011052e-01 1.77121386e-01 -6.02367043e-01 -7.72414386e-01 -1.29093468e-01 -7.26742446e-01 -2.99827773e-02 -3.13900054e-01 5.40806592e-01 5.95632568e-03 3.13610822e-01 -6.03598714e-01 -7.09921956e-01 -9.28207219e-01 -1.08409107e+00 2.71683455e-01 5.21682501e-01 -1.42143860e-01 -1.08928812e+00 5.57186484e-01 6.24189258e-01 1.24279284e+00 -7.12521374e-01 4.74014372e-01 -1.26058698e+00 -3.16814035e-01 -4.11486059e-01 -1.17164008e-01 2.33779073e-01 9.05418321e-02 2.72059292e-02 -7.55760133e-01 -3.15384179e-01 -2.75630563e-01 9.78283361e-02 8.36661935e-01 5.25050104e-01 1.21391070e+00 -3.74564826e-01 -4.21546787e-01 6.62564397e-01 1.18149948e+00 5.21903813e-01 3.95280898e-01 1.68581635e-01 6.37888849e-01 3.13297242e-01 6.49609685e-01 4.54585075e-01 3.93041313e-01 3.74364108e-01 3.48603040e-01 1.79616034e-01 -9.85303223e-02 2.56773978e-02 2.40347415e-01 1.18278170e+00 -5.35421968e-01 -4.92794007e-01 -4.38409597e-01 1.47079349e-01 -2.01001287e+00 -1.24385750e+00 -3.05662453e-01 2.01413035e+00 3.97302538e-01 -4.40015346e-01 2.10530877e-01 -3.51454727e-02 3.14402550e-01 -1.28898248e-01 -6.28367662e-01 -4.90488648e-01 2.00845629e-01 5.32592714e-01 8.59744027e-02 3.78910393e-01 -1.07421613e+00 8.48633885e-01 7.13074350e+00 5.36330104e-01 -1.02463353e+00 2.27102712e-01 7.14126080e-02 -3.46949130e-01 -2.92010337e-01 -2.24497810e-01 -1.04454899e+00 3.80516380e-01 1.21507525e+00 -1.46399677e-01 6.06272578e-01 1.17815363e+00 1.11882746e-01 3.04068565e-01 -1.09244537e+00 9.46311414e-01 5.00074476e-02 -1.91778898e+00 1.40413612e-01 9.67201218e-02 1.14444578e+00 6.79387748e-01 3.03838342e-01 7.36152291e-01 9.66564536e-01 -8.64381492e-01 8.10725540e-02 8.38325202e-01 7.48260766e-02 -7.33432889e-01 9.88308668e-01 2.11359575e-01 -1.02477026e+00 -7.27725089e-01 -9.48057890e-01 -4.08834755e-01 -6.21857606e-02 6.69891357e-01 -1.47782847e-01 4.53253537e-01 1.04141974e+00 1.21700466e+00 -1.97526321e-01 1.74594998e+00 -8.75149965e-02 6.84376538e-01 -1.51217759e-01 -4.12289828e-01 2.61534691e-01 -5.35129428e-01 7.96375498e-02 1.19126534e+00 5.27741015e-01 2.45406240e-01 5.79904675e-01 3.52682590e-01 -2.21552700e-01 5.08693874e-01 -3.68400961e-01 -5.83772846e-02 5.10390878e-01 1.46668684e+00 -4.73747939e-01 -3.18735778e-01 -7.80071616e-01 4.84432936e-01 3.73997539e-01 1.05013072e-01 -4.77168918e-01 -2.31729805e-01 1.01659715e+00 2.14335218e-01 3.89786661e-01 -5.80822453e-02 1.36160273e-02 -1.15845191e+00 -7.91348994e-01 -1.02649021e+00 7.51369953e-01 -4.57316458e-01 -1.82227063e+00 6.23468459e-01 -6.20628715e-01 -1.30416405e+00 2.01889463e-02 -4.42936689e-01 -7.82991767e-01 8.26614201e-01 -1.48227954e+00 -7.85109878e-01 -4.20885682e-01 7.75482655e-01 7.98280478e-01 -6.96512520e-01 9.89798665e-01 8.85535717e-01 -2.48761624e-01 6.23391271e-01 7.66837776e-01 -2.18146130e-01 8.73528719e-01 -1.13208258e+00 3.40230048e-01 4.15102303e-01 4.96143371e-01 1.22264028e+00 4.09191966e-01 -4.16230619e-01 -1.56035960e+00 -7.43142784e-01 7.83816516e-01 -5.01244247e-01 7.26870239e-01 -9.24347192e-02 -9.50461507e-01 3.54983270e-01 3.83148640e-01 -3.61464858e-01 1.39634025e+00 1.04520917e+00 -3.60248417e-01 -2.68604189e-01 -8.24705243e-01 2.68922687e-01 1.16707850e+00 -2.73833603e-01 -6.84412122e-01 6.25234127e-01 4.74201083e-01 -7.61887357e-02 -1.06528604e+00 -1.35140017e-01 8.73732924e-01 -1.17713594e+00 1.09341025e+00 -9.60807621e-01 2.78434515e-01 -1.08170874e-01 -1.70387328e-01 -1.56122613e+00 -1.26132321e+00 -5.86929560e-01 -3.46586049e-01 5.42398214e-01 3.42560083e-01 -3.54288787e-01 1.01450872e+00 4.66374665e-01 -5.28491735e-01 -8.76392126e-01 -1.73657790e-01 -2.13103876e-01 2.01791495e-01 -4.29812402e-01 6.58567667e-01 1.09831572e+00 7.69035518e-02 5.11595905e-01 -6.22771382e-01 -2.45413795e-01 1.76885664e-01 4.92711961e-01 7.14904964e-01 -2.11305881e+00 -3.21866214e-01 -6.88561797e-01 1.83997661e-01 -1.48072410e+00 -3.37096035e-01 -1.12642097e+00 -3.75254065e-01 -2.07271528e+00 -1.41856238e-01 -8.65460396e-01 -1.00709665e+00 3.11303258e-01 1.03707075e-01 1.07208513e-01 1.49548054e-01 5.36041915e-01 -9.25763726e-01 3.25871378e-01 1.22094285e+00 -1.97965190e-01 -3.24527264e-01 7.19059408e-01 -1.30726278e+00 6.08761847e-01 5.91292739e-01 -2.80437291e-01 -6.77755833e-01 -6.63190007e-01 1.03685367e+00 -2.42085144e-01 -6.69206500e-01 -8.38428736e-01 7.35239565e-01 -2.66502827e-01 4.70177948e-01 -8.48299563e-01 1.54876083e-01 -8.28250945e-01 4.59608696e-02 2.34151036e-01 -7.30222523e-01 1.00184686e-01 -2.18529463e-01 6.25554144e-01 5.30636273e-02 -2.43872762e-01 5.07816374e-01 -2.87820578e-01 -8.90384436e-01 6.69236004e-01 -7.60827720e-01 -2.75250882e-01 4.09834236e-01 1.48952445e-02 -3.55149984e-01 -5.86168408e-01 -9.88154471e-01 2.19391420e-01 -1.21399529e-01 9.81325507e-01 8.90457749e-01 -1.01007426e+00 -5.82248747e-01 1.52666315e-01 -1.91533566e-01 -7.67626703e-01 4.41388071e-01 6.24746203e-01 -1.70269012e-01 6.29070640e-01 -3.85877162e-01 1.57321453e-01 -1.06813490e+00 7.89523721e-01 2.95253843e-01 -2.13459313e-01 -7.58432984e-01 1.29107308e+00 -5.95398434e-02 -6.40946567e-01 6.05328560e-01 -1.82248384e-01 -1.16876638e+00 2.36868590e-01 1.04559386e+00 6.14670455e-01 3.34904522e-01 -1.64877698e-01 -2.15253770e-01 2.80812562e-01 -7.07281113e-01 6.90660536e-01 1.69274724e+00 -7.17611089e-02 -1.37525648e-02 -5.44922762e-02 7.66010344e-01 -5.63628227e-02 -5.65985322e-01 -4.32878524e-01 -9.10594463e-02 -5.64894855e-01 5.83961010e-01 -1.24588394e+00 -1.51582277e+00 8.36207330e-01 6.63256407e-01 4.65760887e-01 8.52991223e-01 -2.58299202e-01 8.46143126e-01 9.05317247e-01 2.63559669e-01 -1.27544808e+00 -9.60525647e-02 9.80866969e-01 9.38324749e-01 -1.15956688e+00 2.51338661e-01 1.84645683e-01 -3.23779464e-01 1.34917331e+00 5.97611845e-01 -6.49041057e-01 1.51327395e+00 7.44841341e-03 2.40448162e-01 -3.68894279e-01 -1.02812910e+00 -1.95449144e-01 7.87264466e-01 6.65816128e-01 1.17108512e+00 -5.25721759e-02 -4.81865764e-01 9.51934457e-01 -2.42419556e-01 1.28823653e-01 3.09276909e-01 6.93575025e-01 -8.53329062e-01 -1.45621252e+00 8.47297832e-02 1.12204087e+00 -4.20592308e-01 -3.16179931e-01 -3.99334490e-01 5.97223155e-02 6.58211336e-02 1.35783160e+00 -4.41361517e-02 -8.56384933e-01 2.57403016e-01 -3.39926392e-01 1.92590088e-01 -9.04097080e-01 -1.38489068e+00 -1.60581946e-01 -6.16872944e-02 -6.81234837e-01 -4.10538316e-01 -2.28164464e-01 -8.54655147e-01 -6.04357958e-01 -7.96005607e-01 6.78050756e-01 6.62957311e-01 1.02362025e+00 8.11007440e-01 5.42906284e-01 6.57699943e-01 -7.54653037e-01 -5.05967677e-01 -1.01531470e+00 -5.69435775e-01 1.00841880e-01 -2.49651298e-02 -8.51465940e-01 -1.94000900e-01 -6.47222817e-01]
[10.097545623779297, 5.738048553466797]
bf6c11f3-5bc4-45b1-b01f-f88f9f3194d3
trade-off-between-communication-and-1
2305.04423
null
https://arxiv.org/abs/2305.04423v1
https://arxiv.org/pdf/2305.04423v1.pdf
Trade-off Between Communication and Positioning in Millimeter Wave Systems with Bounded and Unbounded Positioning Errors
Millimeter wave has proven to be effective in the integrated positioning and communication (IPAC) system. In this work, we establish a millimeter wave IPAC system by leveraging the inner coupling relationship between estimated data rate and positioning error. Moreover, we formulate robust power allocation problems by minimizing the positioning performance metric, i.e., the Cram\'er-Rao bound (CRB), subject to the data rate and total transmit power constraints to investigate the trade-off between positioning and communication of the IPAC system. Specifically, in the first case, we characterize the positioning error as a bounded ellipsoidal model to formulate the power allocation problem, which can be solved by the ${\cal{S}}$-Procedure and the alternating optimization (AO) algorithm. In the second case, we formulate the optimization problem by characterizing the positioning error as an unbounded Gaussian model, which can be solved by the Bernstein-type inequality and successive convex approximation (SCA) methods. Numerical results verify the robustness against the positioning error of the proposed robust design by comparing with the non-robust design, and analyze the trade-off between communication and positioning of the integrated system.
['Shiyin Li', 'Ruixin Yang', 'Shuai Ma', 'Junchang Sun']
2023-05-08
null
null
null
null
['robust-design']
['miscellaneous']
[ 1.23750746e-01 4.50149447e-01 2.71451026e-01 1.26733780e-01 -5.97903132e-01 -6.81687653e-01 -1.63020104e-01 -2.77342588e-01 -3.70459378e-01 8.89605224e-01 -2.25439548e-01 -7.31491804e-01 -9.01606858e-01 -5.98872960e-01 -5.98362684e-01 -1.10245943e+00 -9.70891640e-02 -1.29639357e-01 -5.70740461e-01 1.82561949e-01 2.12447092e-01 3.07866246e-01 -6.47101998e-01 -1.14625359e+00 1.12469065e+00 1.61623454e+00 1.20895401e-01 4.63106930e-01 4.45162773e-01 6.83759525e-02 -8.07959855e-01 -2.14120254e-01 3.46294641e-01 -3.50903869e-01 1.02563530e-01 -1.44909248e-01 -2.74308443e-01 -9.43348110e-02 -1.71609327e-01 1.21401966e+00 9.65358853e-01 -1.36004552e-01 4.45351839e-01 -1.44192958e+00 -1.97352931e-01 8.64485279e-02 -6.86914623e-01 -2.13836804e-02 1.10415705e-01 -5.63917577e-01 4.31335270e-01 -6.50352538e-01 2.35033825e-01 8.29946578e-01 6.20382845e-01 -1.93500593e-02 -9.18630242e-01 -8.36312175e-01 -1.17317744e-01 -4.30796117e-01 -1.84417593e+00 -3.57597262e-01 4.97798800e-01 -3.08950573e-01 2.27839127e-02 5.15964568e-01 4.77113456e-01 3.22287709e-01 4.58367020e-01 2.20360950e-01 6.49857223e-01 -3.61645818e-01 4.74614650e-01 6.45441934e-02 -1.69354320e-01 7.00268328e-01 1.09016645e+00 2.31152356e-01 9.53260139e-02 -2.73881495e-01 7.72686422e-01 -2.52972960e-01 -6.36370301e-01 -3.27548385e-01 -9.70727742e-01 5.03372788e-01 3.93212497e-01 1.58744439e-01 -4.03469861e-01 3.42917144e-01 -4.04041827e-01 6.94124922e-02 2.87359446e-01 4.48277861e-01 -2.99180120e-01 -9.45335478e-02 -7.64989436e-01 3.57036516e-02 8.41916859e-01 1.63400483e+00 1.07796378e-01 9.09536853e-02 -5.65212630e-02 5.23940146e-01 1.04609132e+00 1.20571744e+00 -1.45113409e-01 -7.81361520e-01 1.05357587e+00 1.38450816e-01 6.37122571e-01 -1.38189852e+00 -6.11654937e-01 -1.14937043e+00 -9.02398109e-01 -2.65603155e-01 4.42068428e-01 -1.11850655e+00 -2.69021511e-01 1.66056108e+00 2.18529209e-01 -2.34423708e-02 1.75100073e-01 7.25480199e-01 2.23837286e-01 7.82194734e-01 -5.93622386e-01 -8.72784615e-01 1.09812212e+00 -3.63795161e-01 -1.08317590e+00 -1.87060431e-01 4.25999254e-01 -6.24514103e-01 6.16278537e-02 1.70999452e-01 -1.01771832e+00 -2.35437192e-02 -1.68711555e+00 4.38736379e-01 1.09767810e-01 4.29333389e-01 9.62101109e-03 1.23217928e+00 -8.02136123e-01 -1.50776535e-01 -6.66961789e-01 -5.23990281e-02 1.47503465e-01 5.06944239e-01 1.12775698e-01 -1.77073944e-02 -8.15188766e-01 5.06214321e-01 8.14772248e-02 7.45192766e-01 -3.80140692e-02 -8.33447754e-01 -6.93465114e-01 6.37828335e-02 4.11187202e-01 -5.59694469e-01 7.78510630e-01 -3.81628685e-02 -1.76763201e+00 1.05672525e-02 -7.49739036e-02 -1.86125085e-01 4.60769385e-01 -9.25500616e-02 -1.67825684e-01 -1.08974323e-01 7.89542794e-02 -2.14552835e-01 3.00967693e-01 -1.32377160e+00 -8.10425162e-01 -7.07580209e-01 4.72818241e-02 1.26529247e-01 2.71491140e-01 -3.28033626e-01 -6.37993991e-01 -5.22234201e-01 7.52223551e-01 -1.25537896e+00 -4.40073580e-01 -3.05000722e-01 -6.60671949e-01 3.78189087e-01 4.84123319e-01 -5.24274230e-01 1.36050451e+00 -2.47029281e+00 1.98526397e-01 8.28100979e-01 -2.02805787e-01 -2.13533834e-01 4.33754593e-01 2.88011312e-01 3.20261508e-01 2.67741859e-01 -2.56667528e-02 -4.09676582e-01 1.97044015e-01 -5.78765906e-02 -3.33799645e-02 1.08276594e+00 -6.10931754e-01 4.03198928e-01 -5.87661445e-01 -1.17955878e-01 -1.31981552e-01 1.62395164e-01 -4.45345670e-01 3.18853259e-02 4.85278875e-01 5.55030525e-01 -8.22107136e-01 6.28847420e-01 1.08471394e+00 1.49022520e-01 2.18206614e-01 -2.54996449e-01 -4.33270097e-01 -4.05478597e-01 -1.42617929e+00 1.25161839e+00 -7.81300902e-01 3.07287514e-01 8.83659780e-01 -8.09472680e-01 1.01504827e+00 2.51790136e-01 4.39444035e-01 -7.34792054e-01 4.64916915e-01 4.86957610e-01 -5.70745766e-02 -2.13815525e-01 1.83665454e-01 -1.93714231e-01 -4.41507399e-01 2.01748624e-01 -3.35493833e-01 -9.19031873e-02 -2.66330689e-01 -2.50538699e-02 7.75899529e-01 -3.50317776e-01 4.13477749e-01 -6.90810323e-01 6.45052612e-01 -4.84374374e-01 8.22599113e-01 6.60636544e-01 3.66976783e-02 1.92576841e-01 5.70588291e-01 4.53024775e-01 -6.19436502e-01 -7.32505560e-01 -4.11374778e-01 1.26180068e-01 8.01569581e-01 -1.01685263e-01 -4.61413652e-01 -2.26636440e-01 1.83646068e-01 6.23221397e-01 -1.90409437e-01 -2.55806409e-02 -1.49377197e-01 -7.17851579e-01 5.34345031e-01 1.01811305e-01 6.54268086e-01 2.36446336e-01 -4.84988064e-01 4.98039536e-02 1.17915915e-02 -1.09310043e+00 -4.10882652e-01 2.12503314e-01 -4.00301963e-01 -9.02775347e-01 -7.63780653e-01 -5.75139463e-01 1.04699898e+00 3.24968606e-01 1.08050540e-01 -3.03867638e-01 1.00530237e-01 7.20544159e-01 -2.24109873e-01 -5.68003297e-01 5.56778967e-01 -6.41083866e-02 4.52688992e-01 1.69305027e-01 -5.46570063e-01 -5.50472379e-01 -7.76788652e-01 7.96508551e-01 -2.80140549e-01 -2.96809256e-01 5.49217880e-01 3.67504448e-01 5.77625930e-01 6.03359461e-01 6.53491318e-01 -2.60358572e-01 5.89695156e-01 -5.51350534e-01 -1.24439633e+00 1.13417536e-01 -5.62231243e-01 -8.19356740e-02 4.59197611e-01 9.74034145e-02 -9.48416770e-01 7.74366185e-02 6.07002340e-02 1.76206172e-01 7.16068029e-01 7.32135653e-01 -9.58298206e-01 -5.92608213e-01 -1.13357186e-01 2.83522397e-01 -2.09593579e-01 -8.17523599e-02 3.24455261e-01 1.06385946e+00 5.18114686e-01 -5.94553888e-01 1.28521371e+00 4.13982719e-01 4.32459950e-01 -9.68063474e-01 -6.77090347e-01 -3.51966381e-01 -6.60548508e-02 -2.63478667e-01 6.74595952e-01 -7.85034955e-01 -1.16554332e+00 4.47060987e-02 -1.04741204e+00 1.99488297e-01 3.27433527e-01 8.05014193e-01 -4.73362178e-01 3.18347484e-01 1.23846985e-01 -1.42407560e+00 -2.01696277e-01 -1.02248836e+00 8.35910678e-01 3.90694559e-01 1.49826244e-01 -7.78680086e-01 -1.60386488e-01 2.24528059e-01 5.22099018e-01 7.26064146e-01 5.17881930e-01 -5.09273522e-02 -7.77188838e-01 -6.09551787e-01 -2.85149753e-01 -8.66008699e-02 -1.14855029e-01 -6.68716609e-01 -2.32476577e-01 -6.44833744e-01 3.65471184e-01 5.39866149e-01 -2.48726726e-01 5.46208203e-01 8.63411546e-01 -5.98790407e-01 -6.14313424e-01 1.17316103e+00 1.71018672e+00 8.11581016e-01 7.54486620e-01 1.99218303e-01 1.05237886e-01 2.01051787e-01 9.01002884e-01 6.93841279e-01 2.05534264e-01 7.94744968e-01 5.58327317e-01 2.96770841e-01 8.10086787e-01 -5.31049818e-02 -1.07304998e-01 6.39407635e-01 -5.25714364e-03 -5.18678606e-01 -4.59993362e-01 1.02431864e-01 -1.94198287e+00 -3.83897096e-01 -2.47704178e-01 2.46000409e+00 2.53165156e-01 -7.62442648e-02 -4.45779353e-01 1.27244473e-01 6.88335717e-01 -1.38999730e-01 -2.80549884e-01 6.06845021e-02 1.09679870e-01 -2.57073432e-01 1.39926910e+00 7.67675161e-01 -9.07966435e-01 -4.06532399e-02 5.50327253e+00 7.41657376e-01 -8.98126900e-01 -4.42859866e-02 2.99884528e-01 7.67464638e-02 -1.96148455e-01 -5.66010624e-02 -8.57385278e-01 9.63824570e-01 7.35691905e-01 -7.29667723e-01 2.19534382e-01 7.37730622e-01 4.60419536e-01 -4.14983422e-01 -7.67460167e-01 1.37959063e+00 5.97109981e-02 -8.89700055e-01 -7.99743533e-01 7.35041082e-01 4.98215675e-01 -5.80670476e-01 1.35242313e-01 1.20936878e-01 -2.40411043e-01 -5.45707881e-01 7.36717105e-01 6.34326220e-01 6.83865666e-01 -7.78570831e-01 1.04716432e+00 4.41910267e-01 -1.36641121e+00 -4.13064688e-01 -3.24862510e-01 1.82989910e-02 4.40030009e-01 8.85112584e-01 -1.48117155e-01 1.17928183e+00 2.22775519e-01 5.13783395e-02 1.48960454e-02 1.44933319e+00 -2.31871977e-01 -1.12371389e-02 -7.01150119e-01 -3.10487449e-01 3.85848060e-02 -8.60346675e-01 7.43494093e-01 6.63727522e-01 1.06976342e+00 6.07131898e-01 -1.18007891e-01 5.83698392e-01 4.77036759e-02 9.28568318e-02 -3.75705689e-01 3.58311385e-01 1.07261992e+00 1.03182721e+00 -4.97334689e-01 2.30853260e-01 -2.20980719e-01 6.19015455e-01 -4.06677008e-01 6.32072270e-01 -1.02890909e+00 -1.00297809e+00 4.69546914e-01 -2.03532681e-01 2.24853694e-01 -7.10079193e-01 -7.03319073e-01 -5.63803613e-01 3.77489895e-01 -9.55187008e-02 -2.12417930e-01 -4.39504236e-01 -5.38689971e-01 1.52809665e-01 -2.03065977e-01 -1.52315140e+00 1.44918874e-01 -4.04746354e-01 -4.48141426e-01 9.93365645e-01 -1.22603631e+00 -5.56691766e-01 -3.02408397e-01 2.93121696e-01 -2.89154589e-01 3.80664226e-03 3.93944949e-01 7.27083385e-01 -9.72196639e-01 7.58591354e-01 8.68571877e-01 -1.84625134e-01 1.00924253e-01 -8.96489024e-01 -6.21099412e-01 1.00910771e+00 -7.12348402e-01 8.82117033e-01 8.74533474e-01 -3.81561607e-01 -1.92213929e+00 -1.12806678e+00 5.02300858e-01 1.58933531e-02 4.82975155e-01 -4.43159580e-01 2.35788614e-01 3.66615832e-01 -2.76976734e-01 -2.77262386e-02 7.23397434e-01 -1.68297768e-01 3.23865116e-01 -4.14036095e-01 -1.28987062e+00 5.73116243e-01 9.20234203e-01 -4.52064499e-02 1.26080170e-01 7.84180164e-02 6.10800862e-01 -5.56753933e-01 -1.05107021e+00 5.88673174e-01 7.33468413e-01 -1.22666821e-01 8.75022411e-01 2.67494619e-01 -8.87282416e-02 -6.72842443e-01 -6.29685283e-01 -1.29256999e+00 -3.12033385e-01 -9.55578506e-01 7.15990663e-02 1.42817783e+00 7.58883715e-01 -8.32774401e-01 7.31552601e-01 5.04130065e-01 -1.08112931e-01 -8.92015517e-01 -1.47002923e+00 -1.15582454e+00 -3.70852441e-01 -3.06398958e-01 3.40131998e-01 3.55232149e-01 3.95300955e-01 2.57692337e-01 -5.03559291e-01 1.06235278e+00 7.99295664e-01 -3.28723699e-01 7.74769723e-01 -9.49241400e-01 -3.05228710e-01 -9.06017572e-02 -4.30070668e-01 -1.58054590e+00 -3.39220196e-01 -3.28094661e-01 4.25004154e-01 -1.70259452e+00 -5.56568444e-01 -9.06348765e-01 -2.18669884e-02 -2.73536056e-01 1.17536947e-01 -6.80757090e-02 9.22712311e-02 3.97300767e-03 -4.40757245e-01 6.70777440e-01 9.70003128e-01 -2.19297558e-01 -4.02526975e-01 6.60089433e-01 -9.78802681e-01 5.49524069e-01 5.92707098e-01 -1.44393787e-01 -5.06020010e-01 -3.97990823e-01 6.44288242e-01 6.31473124e-01 -1.41869843e-01 -1.09483552e+00 5.74286461e-01 4.40536290e-02 5.14121503e-02 -6.56410635e-01 3.71925116e-01 -1.48750603e+00 3.99528325e-01 6.07423425e-01 3.38880986e-01 -3.22022200e-01 -2.38661431e-02 9.97775435e-01 2.07253829e-01 -2.10237861e-01 7.30453312e-01 5.87661088e-01 2.47691453e-01 2.34532163e-01 -4.60309386e-01 -2.65986830e-01 1.34316015e+00 -2.69626170e-01 -5.12855053e-01 -7.53004134e-01 -4.62368369e-01 6.98372602e-01 -1.06965676e-01 -5.03728241e-02 3.79220426e-01 -1.51856661e+00 -2.71155447e-01 -1.06261130e-02 -3.26156840e-02 -5.46363629e-02 1.83665052e-01 1.26612628e+00 -5.88477671e-01 9.41011190e-01 4.71276015e-01 -3.90567124e-01 -1.06787562e+00 9.44612771e-02 5.20361304e-01 8.59480277e-02 4.52183895e-02 7.44235039e-01 -9.72030610e-02 -2.00384766e-01 5.49138963e-01 -2.25954473e-01 4.60084043e-02 -1.51079059e-01 2.58639187e-01 4.42818016e-01 -6.93329573e-02 -2.72244394e-01 -6.16995394e-01 8.87183845e-01 6.63419902e-01 -3.36241841e-01 9.43518043e-01 -7.86512971e-01 -1.66144490e-01 -1.98747903e-01 1.19767463e+00 6.13445282e-01 -9.12178159e-01 -1.56653840e-02 -1.22615799e-01 -5.97099364e-01 1.01143241e-01 -6.59141958e-01 -9.46673036e-01 1.53879104e-02 6.84222639e-01 4.35296625e-01 8.50151122e-01 -3.57078463e-01 4.36273098e-01 4.46716577e-01 9.07489777e-01 -9.44384575e-01 -3.70435297e-01 3.58507484e-01 7.94218004e-01 -6.33349955e-01 3.03370863e-01 -6.80467188e-01 1.71325371e-01 8.39566469e-01 3.06740284e-01 -8.31167027e-02 1.07077837e+00 3.61084342e-01 -2.02161878e-01 1.27672166e-01 1.44943252e-01 1.36377171e-01 7.49991983e-02 3.41515392e-01 2.21391052e-01 2.11666942e-01 -9.33100164e-01 1.37606931e+00 -5.69846630e-01 -3.19981575e-01 4.84183639e-01 9.06774461e-01 -5.12852252e-01 -7.19995022e-01 -7.02339530e-01 1.10436119e-01 -4.39827085e-01 2.29331806e-01 2.11804714e-02 6.89732909e-01 4.22902107e-01 1.45661318e+00 -1.24257870e-01 -2.47829556e-01 6.59656405e-01 -5.13966262e-01 3.51150632e-01 -1.62299886e-01 4.87291723e-01 2.09387764e-01 1.38912082e-01 -3.61779541e-01 -1.27992824e-01 -3.71550530e-01 -8.72799754e-01 -1.25727341e-01 -7.78856754e-01 6.26905978e-01 1.02303255e+00 1.04767811e+00 3.96622390e-01 8.58461380e-01 1.06204855e+00 -5.41911840e-01 -4.43092465e-01 -5.90427458e-01 -9.71194863e-01 -7.25736558e-01 1.33019030e-01 -5.08575678e-01 -5.89215636e-01 -7.11290419e-01]
[6.170117378234863, 1.3630609512329102]
146717c5-dbe0-4a96-933b-a610093e93d5
dccrn-kws-an-audio-bias-based-model-for-noise
2305.12331
null
https://arxiv.org/abs/2305.12331v3
https://arxiv.org/pdf/2305.12331v3.pdf
DCCRN-KWS: an audio bias based model for noise robust small-footprint keyword spotting
Real-world complex acoustic environments especially the ones with a low signal-to-noise ratio (SNR) will bring tremendous challenges to a keyword spotting (KWS) system. Inspired by the recent advances of neural speech enhancement and context bias in speech recognition, we propose a robust audio context bias based DCCRN-KWS model to address this challenge. We form the whole architecture as a multi-task learning framework for both denosing and keyword spotting, where the DCCRN encoder is connected with the KWS model. Helped with the denoising task, we further introduce an audio context bias module to leverage the real keyword samples and bias the network to better iscriminate keywords in noisy conditions. Feature merge and complex context linear modules are also introduced to strength such discrimination and to effectively leverage contextual information respectively. Experiments on the internal challenging dataset and the HIMIYA public dataset show that our DCCRN-KWS system is superior in performance, while ablation study demonstrates the good design of the whole model.
['Lei Xie', 'Long Ma', 'Sining Sun', 'Xiong Wang', 'Shubo Lv']
2023-05-21
null
null
null
null
['small-footprint-keyword-spotting', 'keyword-spotting', 'speech-enhancement']
['speech', 'speech', 'speech']
[ 0.46059433 -0.3345368 0.09412286 -0.47058883 -1.4259113 -0.2109551 0.4651138 -0.49191523 -0.5974118 0.24134283 0.63674015 -0.31841922 0.01922458 -0.20693499 -0.6801752 -0.92093635 0.26747593 -0.27709922 0.01289 -0.32145017 0.0316256 0.08560326 -1.5020516 0.51968557 0.68344975 1.299815 0.83118904 0.97921836 0.05527418 0.58646715 -0.83320063 -0.11820081 0.10725147 -0.24697192 -0.24444331 -0.2107928 0.46560892 -0.3411319 -0.5283951 0.94399756 1.3409127 0.16735066 0.30515182 -0.96694773 -0.85051644 1.1800001 -0.3233424 0.32474253 0.24818465 0.1758148 1.1429704 -1.4587512 0.19529647 1.4210777 0.7656986 0.48899034 -0.94225895 -0.93446094 0.52642995 0.8028342 -1.543241 -1.0438517 0.9503283 0.2171393 0.8881805 0.4519977 0.469304 1.842395 -0.24682944 1.2629224 0.98938286 -0.36611286 0.19459589 -0.1275579 -0.11876158 -0.03818943 -0.50419647 0.1971695 -1.2329351 0.09857043 0.11707727 -0.15069291 -0.5558332 0.3132976 -1.1480095 0.47248477 0.25110698 0.35974768 -0.23188515 0.4835452 0.35568723 0.5254736 0.35279086 0.2781267 -0.66800594 -0.17802511 -1.2248169 0.2962628 0.38425964 1.0732569 0.22966051 0.6405163 -0.6335832 1.3637141 0.37904206 0.75934166 0.6518023 -0.572904 0.50115156 -0.14814293 -0.26355442 -0.659321 -0.07396176 -0.88057774 -0.96447766 -0.35872325 -0.24780126 -0.06524016 -1.1431192 1.8608496 0.20207132 0.7385126 0.14174862 0.9995499 1.0224351 0.81800365 -0.03970408 -0.0382022 1.5199404 -1.0108179 -1.2962425 -0.34458804 0.17868903 -0.90505636 1.3632021 0.56325525 -0.91170985 -0.72077996 -1.2800158 -0.18907799 -0.53458077 0.41694203 0.17725988 0.49008268 -1.1117026 0.22041604 -0.35211152 0.18216638 0.22409402 -0.05720466 -0.1587604 -0.07981758 -1.7532123 0.5531698 0.3481071 0.7523815 -1.3373269 -0.7508634 -0.789092 0.15999049 0.7945337 -0.44389462 1.4252598 -0.6322879 -1.6048303 0.27007008 -0.17215006 -0.64517415 0.3376076 -0.47127748 -0.8610484 0.08014145 -0.13321911 0.3774547 1.4323537 -1.1728088 -0.7426241 -0.09227803 -0.39618206 0.33782557 -0.56731176 0.19313538 -0.7742668 -1.4294579 0.16799527 -0.44874707 -0.1689718 -0.36839712 -0.58968234 -0.14059958 1.1235793 -1.0756713 1.3978505 -2.3979037 0.11122963 0.10328213 -0.12377758 0.4508229 -0.47782925 0.44412243 -0.15350267 0.01372388 -0.16835655 -0.6202078 0.2630727 0.09906985 -0.65883124 0.2543765 0.40797544 0.857156 -0.7219218 -0.20387682 0.03308083 0.7311136 -0.48118514 0.6363278 -0.03429322 0.17595446 -0.15030009 0.7879547 0.6681701 0.36486462 -0.06528123 -0.31727448 0.19025695 0.41841903 -1.4665941 1.9508232 -0.6954749 0.43187144 0.7282885 -1.0296183 0.78146577 0.59240115 -0.09704822 -0.8877203 -0.0349366 0.3516414 -0.10876927 -0.6141902 0.69762963 -0.02323157 -0.03748033 -0.09895474 0.33408925 -0.3732814 -0.3781949 0.34947708 1.0254472 -0.07333747 -0.04829338 0.02091546 0.6770141 -0.76242864 0.60210794 1.0875471 -0.2037089 0.87383676 0.02065645 0.05062744 -0.7090411 -0.9178498 -0.08660846 1.4514717 0.01378275 -0.586146 -0.6688791 -0.44621706 -0.11769356 0.61280566 -0.42871013 -0.48129213 -0.57453775 -0.4815045 0.87773955 0.53272086 0.5010235 -0.92967445 0.10024461 0.24639781 -0.4335144 -1.4516561 -0.8578897 0.6164831 -0.11892236 -0.53489435 -0.8255051 -0.828245 -0.05280752 0.302786 0.74767494 -0.12046624 -0.23085144 0.46570617 -0.76931846 -0.47515187 -0.412713 0.17754164 0.22528578 0.24900164 0.08188317 -0.78683525 -0.69442576 0.25169146 -0.9885528 -0.09651451 0.9082485 1.2747675 0.6250657 -0.02377063 1.1390239 -0.38443762 0.9415805 -0.4636391 -0.3214267 0.24596046 -0.7140912 -0.15565051 0.5794226 -0.72276366 -1.0780407 -0.16294952 -0.73073345 -0.39735118 -0.0697865 0.5157408 -0.6703962 0.18315795 0.28656176 0.7373394 -0.3600553 -0.86396825 0.55433846 1.2838293 0.8496858 -0.42640546 0.9153438 -0.00891282 -0.48787844 -0.85371745 -0.6789133 -0.5063124 -0.1016735 -0.08240257 0.58654594 -1.3942177 -0.4868633 0.6208666 -1.1400466 -0.17381541 -0.19088715 0.2990676 -0.3323051 0.41207823 -0.44762796 -1.0940007 -0.5806895 -1.3035504 1.4081641 -0.05986487 0.2761601 -0.48458812 -0.2922387 0.3784564 0.93148804 -0.5689275 0.5775432 -0.89422816 -0.38530368 -0.15368253 -0.01534982 0.9206239 -0.07093696 -0.45378023 -1.6393225 -0.23460269 0.25435394 -0.39828143 1.1534448 0.19395143 1.6744316 -0.27235815 0.03413196 0.8265371 0.987257 0.17049323 0.601878 0.06548981 0.572579 0.19056724 0.65734863 0.27358273 0.3009495 0.7432854 0.35971394 -0.16996172 -0.7385297 -0.43062744 0.6167945 1.4830137 0.5219559 -0.35045794 -0.6495813 0.6783359 -1.6728511 -0.64070374 0.33620602 1.5992947 1.3721992 -0.02473077 -0.420816 0.38832036 0.6617408 0.57224447 -0.4436865 -0.16735084 -0.52692544 0.24384941 0.08922338 0.5053693 -1.0860575 1.0252901 6.1983933 1.7282994 -1.0540651 0.27319002 0.66540563 -0.17112824 -0.3191556 -0.1682483 -0.9351691 0.5709722 1.0765086 0.2213539 0.6689225 0.73661137 0.3591516 0.22405992 -0.96096855 1.2699159 0.05013393 -1.0810913 0.04690285 -0.3447556 0.34268755 -0.01684912 0.40211126 0.67303413 0.01070139 -1.0740975 1.0110435 0.47023773 0.95642406 -0.69359046 0.60821664 0.20217474 -1.252827 -0.3615196 -0.2373364 0.4341258 0.05136193 0.798946 -1.1254168 0.71113527 1.0084513 0.5300055 -0.3791658 0.7667635 -0.44206935 0.9548984 -0.32923833 0.05163909 0.15440883 0.12156753 0.81293935 1.800525 0.30723676 -0.1726864 -0.04791918 0.5579621 -0.34398407 0.12718879 -0.23470229 0.13736059 0.7870077 1.3522179 -0.07235494 -0.3441444 -0.14605787 0.974788 0.02297479 0.69795954 -0.832245 -0.64356846 0.8076363 -0.31066075 0.624101 -0.08749577 -0.05191974 -1.1767495 0.19099711 -1.3567687 0.01799973 -0.90301555 -1.1950178 0.6045462 -0.2753117 -1.0439531 -0.04601032 -0.5526223 -0.53435904 1.0429757 -2.2062085 -1.2944728 -0.05399158 0.79941005 0.97073394 -0.39105374 0.5976629 0.64672065 -0.49734864 1.1071913 0.07109347 0.07763984 1.0320821 -1.193014 0.47667193 1.1840559 0.238087 0.5663239 0.59187627 -0.49324754 -1.7570256 -1.0966077 0.5538867 -0.02415608 0.8720416 -1.0425898 -0.80753946 0.16396825 0.42514852 0.14870308 0.50109 0.05370344 -0.5375576 -0.52185583 -0.77941525 0.78498924 1.237958 -1.0771192 -0.55505824 0.13428253 1.4432648 -0.31198817 -0.5594019 0.5612577 0.38532263 -0.5082662 1.2562114 -0.48119974 0.05292938 -0.33734006 -0.6822705 -1.595796 0.08775851 -0.9118609 -0.3694076 1.7202989 0.4865825 -0.4105614 0.30818167 -0.18319497 -0.59333295 -0.67263156 -1.4276274 -0.67226225 -0.19244374 -1.1023836 0.75446737 0.83245575 -0.39416623 0.34163556 -0.77004755 0.3166365 0.41708195 -0.3593199 0.5019119 -0.5439802 -0.5343327 -0.15485877 -0.06286273 -1.3856112 0.03698143 -0.8306063 0.5008064 -1.0991132 -0.21658014 -0.3311204 -0.6007364 0.4414033 -0.5051476 0.03439978 0.1669825 -0.1425855 -0.59946537 1.1021613 1.0611498 -0.39921916 0.11476367 0.03165947 -0.9483432 0.45117232 0.39016744 -0.35374972 -0.38885176 -0.46602023 0.15450345 -0.00951633 0.24507354 -0.8601689 0.48753676 0.11579929 0.30369917 -0.7026056 0.70430243 -1.0231792 -0.314128 -0.02417753 -0.61588556 -0.24886067 0.10411628 0.8346737 -0.5024595 -0.01704519 0.58368045 0.09703713 -0.7233053 0.05435571 -0.44397968 -0.03359329 0.3440742 0.14501074 -0.160365 -0.65502435 -0.70236826 0.30254468 -0.4746925 0.6511169 1.0007169 -1.4601258 -0.8639728 0.36526707 0.1352032 0.06903792 0.36794728 0.721054 0.30932742 0.13419466 0.6758574 -0.5711323 -1.1794579 0.37879878 0.34396815 0.02962776 -0.5013493 1.2592303 0.09028342 -0.5158788 0.9095095 -0.5872744 -0.14368017 0.18198235 0.7987711 0.11554352 0.5309884 -0.39492068 -0.2705696 0.2125589 -0.28459302 -0.36057466 1.4855682 -0.36998332 0.19643912 0.27386636 1.1444441 0.16521476 -1.2675263 -0.6483014 0.03495218 -0.33652502 0.39391017 -1.2424107 -1.0125043 0.9991855 0.8225757 0.10160955 1.556181 -0.1563767 0.9044314 0.47575748 -0.09432704 -1.6014286 0.33494478 0.5924468 1.3374897 -1.0803325 -0.28000194 -0.12211494 -0.5909777 0.95163125 0.29254076 0.18704905 0.7596944 0.7935035 0.26431125 0.19191818 -0.8499786 -0.17964946 0.24686624 0.6858783 -0.03070233 -0.21794142 0.12021024 1.3468685 -0.35469678 -0.41865 0.23202439 0.68283486 -0.36826473 -1.0700905 -0.5083799 0.02726359 -0.6225349 -0.6877579 -0.25983974 0.19821708 -0.07700011 1.2011696 -0.42578405 -0.7546568 0.4911356 0.34974793 -0.19764295 -0.32915893 -0.8606474 0.6978684 0.14181823 -0.5180364 -0.08773387 -0.34412792 -0.86634344 0.2896894 -0.6881307 0.02488148 1.0386666 0.8405428 0.48406005 0.98382413 1.0170989 -0.86223596 -0.9253133 -1.3696352 -0.56112194 0.13393222 0.7917762 -0.2722159 -0.6514834 -0.12222095]
[14.70246696472168, 6.122887134552002]
261366e6-1c43-4be0-809f-83c1ffc03e2e
score-based-generative-models-for
2306.13843
null
https://arxiv.org/abs/2306.13843v1
https://arxiv.org/pdf/2306.13843v1.pdf
Score-based Generative Models for Photoacoustic Image Reconstruction with Rotation Consistency Constraints
Photoacoustic tomography (PAT) is a newly emerged imaging modality which enables both high optical contrast and acoustic depth of penetration. Reconstructing images of photoacoustic tomography from limited amount of senser data is among one of the major challenges in photoacoustic imaging. Previous works based on deep learning were trained in supervised fashion, which directly map the input partially known sensor data to the ground truth reconstructed from full field of view. Recently, score-based generative models played an increasingly significant role in generative modeling. Leveraging this probabilistic model, we proposed Rotation Consistency Constrained Score-based Generative Model (RCC-SGM), which recovers the PAT images by iterative sampling between Langevin dynamics and a constraint term utilizing the rotation consistency between the images and the measurements. Our proposed method can generalize to different measurement processes (32.29 PSNR with 16 measurements under random sampling, whereas 28.50 for supervised counterpart), while supervised methods need to train on specific inverse mappings.
['Fei Gao', 'Jianwen Luo', 'Liming Nie', 'Hengrong Lan', 'Shangqing Tong']
2023-06-24
null
null
null
null
['image-reconstruction']
['computer-vision']
[ 6.81303203e-01 -4.44554053e-02 5.58959901e-01 -3.18277955e-01 -1.21179950e+00 -3.26871425e-01 4.60252762e-01 -7.51811028e-01 -5.54460883e-01 6.40374243e-01 3.09842139e-01 1.25626639e-01 -4.12730306e-01 -7.20533609e-01 -6.92084134e-01 -1.39490759e+00 3.45616579e-01 4.84328419e-01 2.56867975e-01 4.68208313e-01 2.35480905e-01 2.12356865e-01 -1.20412683e+00 -1.53062925e-01 4.78395015e-01 8.88922870e-01 7.01454878e-01 9.03837502e-01 5.78538291e-02 6.93798840e-01 -3.05583745e-01 -1.47012129e-01 1.33323342e-01 -4.53024894e-01 -3.45831007e-01 3.41341011e-02 8.66183117e-02 -4.65550065e-01 -6.88173294e-01 1.14156961e+00 7.54438102e-01 2.06363961e-01 8.60220611e-01 -9.68739271e-01 -8.64551604e-01 6.43592060e-01 -7.19310045e-01 -4.41021025e-02 1.85375214e-01 2.81779885e-01 6.91630065e-01 -8.72670829e-01 3.45386654e-01 9.94227409e-01 3.49026471e-01 6.74029648e-01 -1.00771511e+00 -6.47729576e-01 -6.05508566e-01 4.97344345e-01 -1.15069687e+00 -5.96312404e-01 8.96030962e-01 -3.94743532e-01 4.63057131e-01 5.05967624e-02 5.74630678e-01 1.06987226e+00 2.18713775e-01 3.71719986e-01 1.67957497e+00 -4.17037964e-01 3.26933891e-01 -4.33557890e-02 -4.03633744e-01 5.11228323e-01 2.21286014e-01 4.11691576e-01 -8.88711989e-01 1.02512635e-01 1.00704467e+00 -3.38300854e-01 -6.03072286e-01 -2.80737907e-01 -1.12359166e+00 5.48545897e-01 2.65839696e-01 -4.21578698e-05 -4.25625414e-01 3.73199791e-01 -2.38010108e-01 -1.99158207e-01 5.92187941e-02 4.96098369e-01 7.11281598e-02 -2.51156658e-01 -9.22439635e-01 -4.14519578e-01 5.96314251e-01 7.73565114e-01 8.51715088e-01 3.40059280e-01 7.64361694e-02 5.30888259e-01 6.86518490e-01 1.13021541e+00 6.44509494e-01 -1.39600372e+00 -3.45002790e-03 -1.10882021e-01 8.81284475e-02 -5.12643754e-01 -5.72964177e-03 -3.91178012e-01 -9.49732780e-01 1.17013074e-01 2.59474039e-01 -1.01854108e-01 -1.09941494e+00 1.72526968e+00 3.28571260e-01 6.70255721e-01 2.94463843e-01 1.04267514e+00 9.24368083e-01 6.27921283e-01 -1.80011868e-01 -4.47825521e-01 9.67641532e-01 -5.75900495e-01 -7.19242275e-01 -1.99866936e-01 -3.09991147e-02 -9.35853243e-01 8.39256942e-01 5.54221690e-01 -1.14325559e+00 -3.81780863e-01 -1.10396850e+00 -1.08317584e-01 2.20642731e-01 1.46294370e-01 1.48530558e-01 6.23184144e-01 -1.27029026e+00 4.23011661e-01 -1.16505659e+00 1.62652522e-01 2.41646901e-01 1.07928395e-01 -3.15611690e-01 -3.72264028e-01 -8.11246693e-01 7.81219780e-01 -2.01035645e-02 1.10923745e-01 -1.37150621e+00 -6.88796341e-01 -6.20986760e-01 -1.62274465e-01 2.07805812e-01 -9.31586981e-01 1.11550283e+00 6.93899319e-02 -2.35096025e+00 5.20576656e-01 -3.37836117e-01 -3.18304390e-01 3.93440962e-01 -7.64567927e-02 1.73651055e-03 5.59857190e-01 7.94490054e-03 5.48971117e-01 1.06736386e+00 -1.36149299e+00 4.93694581e-02 -2.18388557e-01 -4.39417720e-01 1.45772636e-01 -4.61129658e-02 -3.68533522e-01 -1.91725895e-01 3.24916653e-02 6.58039451e-01 -7.80505478e-01 -1.59767225e-01 9.72625520e-03 -4.48670596e-01 2.44089022e-01 6.81075811e-01 -6.15893304e-01 5.07923841e-01 -1.82599545e+00 3.13770384e-01 -1.95749164e-01 2.69064397e-01 2.86998097e-02 -1.62599877e-01 4.12789345e-01 1.92229271e-01 -2.18773916e-01 -4.58918691e-01 -7.88801491e-01 -2.90441569e-02 3.31469208e-01 -3.75137299e-01 6.82468116e-01 -9.51643661e-02 9.68768418e-01 -9.46699321e-01 -4.73526597e-01 4.75989342e-01 6.16620839e-01 -3.82537723e-01 4.75104958e-01 3.57378908e-02 1.14140439e+00 -3.66314411e-01 5.35290956e-01 1.05373299e+00 -3.80512208e-01 -3.04627940e-02 -5.33864737e-01 -2.64558405e-01 -3.19473408e-02 -9.46162045e-01 1.87691629e+00 -7.58812666e-01 4.53774691e-01 2.10377201e-01 -9.67516124e-01 7.82772541e-01 4.23168331e-01 6.82385743e-01 -4.45515275e-01 1.09445751e-01 1.29611328e-01 1.81286763e-02 -8.20907295e-01 2.92713106e-01 -6.12318099e-01 2.52103597e-01 6.33531988e-01 3.02175283e-01 -1.01307356e+00 -4.84902352e-01 1.38200432e-01 8.31204176e-01 2.73042440e-01 1.59381077e-01 3.10451537e-01 3.17829102e-01 -3.60678434e-01 2.95723885e-01 9.16352212e-01 -5.22543630e-03 8.46345663e-01 -2.22550616e-01 6.26158714e-02 -1.21792209e+00 -1.28749526e+00 -2.00491175e-01 1.92829058e-01 2.10915253e-01 1.59850657e-01 -6.10826671e-01 1.59745649e-01 -5.32659829e-01 4.64555234e-01 -3.56886297e-01 -1.39077067e-01 -3.55824620e-01 -9.16657269e-01 5.44722378e-01 1.65062308e-01 9.05311286e-01 -7.74845064e-01 -3.93830061e-01 2.37744436e-01 -3.58368903e-01 -1.65123498e+00 -6.54316172e-02 -9.18529853e-02 -5.93525171e-01 -8.31293166e-01 -7.72549570e-01 -1.87553659e-01 4.88024801e-01 4.36927408e-01 6.67092621e-01 -7.07158387e-01 -2.52826184e-01 7.32178271e-01 -7.51683637e-02 -5.47525048e-01 -5.48667669e-01 -4.79694754e-01 2.01334491e-01 1.21363558e-01 1.42184729e-02 -1.19851053e+00 -8.23630750e-01 3.97795111e-01 -8.46954644e-01 2.00591698e-01 9.73441303e-01 8.55587959e-01 8.32170248e-01 -1.11151703e-01 3.43136042e-01 -3.05800349e-01 2.03356981e-01 -2.97602624e-01 -6.98805332e-01 9.80807170e-02 -3.65206569e-01 1.67866647e-01 3.42451602e-01 -6.16854489e-01 -1.24246764e+00 4.62090373e-02 -2.78939128e-01 -5.60006440e-01 -2.36298844e-01 4.92842674e-01 -1.81724504e-01 -4.45923716e-01 4.01101381e-01 8.82486820e-01 1.85675666e-01 -3.22019070e-01 3.19589704e-01 5.52071512e-01 8.14463437e-01 -6.23825550e-01 1.08587253e+00 9.48177516e-01 5.98462522e-01 -9.66891944e-01 -9.05222237e-01 -5.49206972e-01 -4.30335611e-01 -2.46418163e-01 1.00339913e+00 -1.03905559e+00 -9.28800285e-01 8.31279576e-01 -9.23853576e-01 -3.99902552e-01 -3.72255683e-01 1.04379356e+00 -5.91917694e-01 9.05458689e-01 -4.11235452e-01 -1.12410498e+00 -2.05501854e-01 -1.26957941e+00 1.24698603e+00 3.36429924e-01 4.39938545e-01 -8.77661765e-01 3.08416277e-01 5.62515259e-01 5.59517503e-01 -2.45238375e-02 3.40382457e-01 2.67493337e-01 -1.01431096e+00 -1.36018738e-01 -1.85470566e-01 5.57873964e-01 1.67620733e-01 -1.95479453e-01 -1.45985162e+00 -2.45095089e-01 5.87499321e-01 -4.10226494e-01 9.41264331e-01 9.89698470e-01 1.13641512e+00 -2.93385726e-03 -1.25805080e-01 9.05860126e-01 1.43890202e+00 1.54783323e-01 7.70658314e-01 -2.09970191e-01 6.65031254e-01 3.27175915e-01 2.10916638e-01 5.12288988e-01 3.74743909e-01 5.04174531e-01 5.68276465e-01 1.95734739e-01 -3.83789539e-01 -5.18766105e-01 4.03638691e-01 1.20302248e+00 -1.19684704e-01 -3.41381460e-01 -3.98678988e-01 3.67812067e-01 -1.29880631e+00 -8.21168125e-01 -6.42901137e-02 2.17835140e+00 7.34658480e-01 -3.13351631e-01 -4.83076364e-01 1.97235987e-01 4.66869593e-01 4.70210314e-02 -7.25861251e-01 3.57273430e-01 -1.79993033e-01 6.30038202e-01 4.45713758e-01 7.47003853e-01 -4.85912830e-01 6.23118639e-01 6.37325287e+00 8.03698242e-01 -1.40573943e+00 3.39779705e-01 2.23970607e-01 1.48658648e-01 -4.68991309e-01 6.83881566e-02 -6.03774011e-01 4.48060513e-01 8.24547291e-01 8.31502303e-02 3.31249684e-01 2.68671751e-01 4.79846358e-01 -4.92083877e-01 -7.33716011e-01 1.14394963e+00 -1.62279478e-03 -1.07623041e+00 1.00928612e-01 3.35241765e-01 7.75665283e-01 3.32887620e-02 4.90549535e-01 -3.00962210e-01 5.11810124e-01 -1.02708340e+00 4.37938452e-01 8.34109962e-01 1.18861830e+00 -8.05429742e-02 4.97159839e-01 5.57487011e-01 -7.65747249e-01 2.63765782e-01 -7.99449623e-01 3.85537446e-02 7.01738060e-01 8.72308433e-01 -1.09587049e+00 6.54413640e-01 6.27947628e-01 7.50971735e-01 4.68727909e-02 9.82912004e-01 -5.20872772e-01 9.98880923e-01 -5.41989088e-01 -3.62638943e-02 -7.67878294e-02 -4.96264189e-01 8.26801598e-01 5.47325313e-01 1.12111080e+00 2.34865203e-01 -2.69566029e-01 1.02785766e+00 1.62960067e-01 -5.28245270e-01 -4.26140815e-01 1.56912044e-01 4.61612791e-01 1.43790734e+00 -2.13729978e-01 6.93366453e-02 -2.04139147e-02 5.94595134e-01 -1.93594769e-01 4.53390270e-01 -7.14954078e-01 2.66684890e-01 2.82651275e-01 6.12190962e-02 2.18564227e-01 -5.75486600e-01 -9.18828249e-02 -1.18646193e+00 -2.37922266e-01 -1.41024351e-01 -1.81299880e-01 -1.38688910e+00 -1.20577252e+00 3.88582796e-01 8.70795473e-02 -1.32744944e+00 -1.86776996e-01 -5.43550253e-01 -6.85564458e-01 1.04786623e+00 -2.03910732e+00 -1.29331064e+00 -7.21028626e-01 5.58035135e-01 1.59866184e-01 5.90037555e-02 8.56976330e-01 -1.85797084e-02 -2.26883873e-01 2.03714445e-01 2.27129936e-01 -1.45110101e-01 7.88039684e-01 -1.23336709e+00 -7.21972510e-02 9.52614605e-01 3.13228220e-01 4.89651114e-01 8.14066350e-01 -2.47655556e-01 -1.49817836e+00 -7.37915516e-01 1.10104181e-01 -2.49379352e-01 6.70176923e-01 -6.35797605e-02 -7.02260554e-01 3.11757535e-01 2.94773430e-01 1.48978189e-01 6.98420227e-01 -8.92296791e-01 -1.39963612e-01 -2.03729853e-01 -1.15977550e+00 9.42081511e-02 8.14630270e-01 -6.31297171e-01 -1.44963995e-01 1.56523526e-01 6.06673181e-01 -7.65927494e-01 -8.39503467e-01 1.51401982e-01 6.12879097e-01 -8.96619618e-01 1.15246546e+00 1.64738283e-01 5.69719732e-01 -4.58933115e-01 -2.73635775e-01 -1.49248636e+00 -8.94020200e-02 -8.85430992e-01 -5.46768270e-02 8.20534229e-01 6.33155778e-02 -7.32492626e-01 8.79106283e-01 1.68813869e-01 -5.75040519e-01 -3.82042736e-01 -1.26831901e+00 -5.54909647e-01 2.21439525e-02 -5.78965366e-01 4.52750355e-01 4.28275108e-01 -4.32291061e-01 2.58079525e-02 -5.95944107e-01 7.96063244e-01 1.30964994e+00 -6.20240159e-02 6.32168949e-01 -9.21747506e-01 -7.12964356e-01 1.25314796e-03 -4.36480433e-01 -1.58625555e+00 -1.74810365e-02 -5.71164668e-01 2.81868160e-01 -1.48940074e+00 4.27977473e-01 -3.61327916e-01 -8.60793795e-03 1.40812127e-02 1.12455092e-01 5.20421565e-01 -2.56403357e-01 3.58067483e-01 -6.60840645e-02 7.28581727e-01 1.88953400e+00 2.22188863e-03 2.27051541e-01 1.96127102e-01 -6.33243740e-01 5.26146472e-01 5.30314028e-01 -4.80084985e-01 -7.01733172e-01 -5.95416725e-01 3.56123894e-01 5.03433466e-01 6.74595594e-01 -1.16383421e+00 7.99132645e-01 -3.06405097e-01 1.12304665e-01 -4.23167139e-01 9.28637445e-01 -4.98468250e-01 4.95238692e-01 1.85715482e-01 -2.43266463e-01 -6.16901517e-01 -2.70418614e-01 8.90630960e-01 -3.61087471e-01 -3.59038204e-01 1.04337156e+00 -3.47948641e-01 -4.02735978e-01 4.36266422e-01 -2.60689080e-01 -7.86429271e-02 4.75781530e-01 -1.05795979e-01 -3.11339051e-01 -7.75703192e-01 -7.99906313e-01 -5.21834075e-01 1.29777610e-01 -3.04480523e-01 1.05906630e+00 -9.71203983e-01 -7.72447824e-01 8.92755985e-02 -2.48951316e-01 3.85516405e-01 5.99870682e-01 1.00353181e+00 -3.04302067e-01 2.60190517e-01 5.85021786e-02 -9.09962237e-01 -8.99376869e-01 -3.22208405e-02 5.80679774e-01 -1.94294319e-01 -4.35198694e-01 1.12189758e+00 2.75146216e-01 -3.33460480e-01 -3.05640727e-01 -3.22251439e-01 -3.80202569e-02 -6.51825249e-01 3.18667382e-01 3.00095826e-01 -7.39055499e-02 -5.87217867e-01 -1.53289735e-02 1.04114449e+00 2.48573124e-01 -6.58781767e-01 1.47816563e+00 -3.11929673e-01 2.31359843e-02 2.80672103e-01 1.13438511e+00 1.29499003e-01 -1.61752379e+00 -4.80232030e-01 -7.15329230e-01 -4.52244878e-01 3.47391576e-01 -6.43889368e-01 -9.54586804e-01 1.38096809e+00 4.67675298e-01 -1.46591082e-01 1.02087486e+00 1.45202652e-01 7.57160127e-01 1.98117673e-01 5.07989168e-01 -4.64678675e-01 6.50571585e-01 1.56573012e-01 7.06009686e-01 -1.24699008e+00 1.21174909e-01 -4.66962188e-01 -3.51309091e-01 9.79595006e-01 3.23363423e-01 -7.20982626e-02 8.16845834e-01 4.51134354e-01 3.80097814e-02 -1.33031607e-01 -5.30153692e-01 2.56452952e-02 1.00203551e-01 1.06933916e+00 1.91808119e-02 9.49102268e-02 3.27114016e-01 4.21345681e-01 -3.38619679e-01 -8.26239809e-02 9.77485359e-01 4.98729825e-01 -3.03919882e-01 -9.58754599e-01 -2.57195324e-01 9.61766765e-02 -2.37685725e-01 -4.89675440e-02 2.28153273e-01 3.63248140e-01 -1.32995725e-01 8.34206641e-01 -2.08194777e-01 -3.08326393e-01 -1.42939808e-02 -3.39788049e-01 8.21802616e-01 -5.08759558e-01 5.03458023e-01 3.85069519e-01 -4.06702518e-01 -2.03679681e-01 -1.04074442e+00 -9.06633615e-01 -9.77889717e-01 1.71354488e-01 -5.19977272e-01 1.90521359e-01 9.30271327e-01 1.00291741e+00 1.67729005e-01 2.25350097e-01 7.80489266e-01 -8.68183911e-01 -7.88335145e-01 -1.36901951e+00 -8.14563692e-01 -1.30255833e-01 3.96649957e-01 -6.50745809e-01 -8.86639416e-01 2.24303976e-01]
[11.73662281036377, -2.3468804359436035]
d4b3876c-9ed8-4c85-823c-c0e33304e019
visual-semantic-slam-with-landmarks-for-large
2001.01028
null
https://arxiv.org/abs/2001.01028v1
https://arxiv.org/pdf/2001.01028v1.pdf
Visual Semantic SLAM with Landmarks for Large-Scale Outdoor Environment
Semantic SLAM is an important field in autonomous driving and intelligent agents, which can enable robots to achieve high-level navigation tasks, obtain simple cognition or reasoning ability and achieve language-based human-robot-interaction. In this paper, we built a system to creat a semantic 3D map by combining 3D point cloud from ORB SLAM with semantic segmentation information from Convolutional Neural Network model PSPNet-101 for large-scale environments. Besides, a new dataset for KITTI sequences has been built, which contains the GPS information and labels of landmarks from Google Map in related streets of the sequences. Moreover, we find a way to associate the real-world landmark with point cloud map and built a topological map based on semantic map.
['Zirui Zhao', 'Yijun Mao', 'Pengju Ren', 'Yan Ding', 'Nanning Zheng']
2020-01-04
null
null
null
null
['semantic-slam']
['computer-vision']
[-3.62316221e-01 -1.50990576e-01 5.84874339e-02 -8.38017225e-01 -1.14637055e-01 -4.05313283e-01 5.47457337e-01 -1.08599380e-01 -5.78063667e-01 5.56526124e-01 -4.77492988e-01 -3.65089625e-01 -1.89625069e-01 -1.25415993e+00 -7.87967086e-01 -1.13262028e-01 -1.69760510e-01 1.08481073e+00 7.68346250e-01 -7.32752085e-01 5.50720036e-01 7.11891830e-01 -1.46548986e+00 -3.89041781e-01 9.87257004e-01 9.98689651e-01 9.77515638e-01 8.98416042e-02 -5.47365785e-01 5.72716355e-01 -5.19308709e-02 1.44609973e-01 3.55423659e-01 -3.14185023e-02 -8.05642724e-01 -2.35099643e-01 2.12114919e-02 -2.83394344e-02 -5.81326425e-01 1.25336671e+00 -1.25074431e-01 2.34685421e-01 3.74508977e-01 -1.61909425e+00 -1.82866171e-01 1.30216271e-01 1.41199911e-02 -3.36765438e-01 4.25526738e-01 2.52828836e-01 3.96025091e-01 -3.32215309e-01 8.61627281e-01 1.47270179e+00 8.27989578e-01 7.02440068e-02 -2.05240980e-01 -6.04111969e-01 -1.57151490e-01 7.25645065e-01 -1.58147931e+00 1.21283114e-01 6.55746937e-01 -3.62221897e-01 1.11406946e+00 -3.43138546e-01 8.63248289e-01 5.26224017e-01 5.91677010e-01 4.00328279e-01 9.76188838e-01 1.88143611e-01 3.30967754e-01 -1.66791216e-01 8.15879926e-02 1.11844826e+00 3.41181755e-01 -9.84084830e-02 -2.61834413e-01 2.80038387e-01 8.45994771e-01 3.72397810e-01 2.45229438e-01 -8.53067517e-01 -1.29060256e+00 8.03593755e-01 1.14177692e+00 2.19435602e-01 -3.48878473e-01 4.80331689e-01 1.51393101e-01 1.16916135e-01 -2.39366859e-01 4.74134892e-01 -4.91163641e-01 -2.90887263e-02 -1.17191873e-01 4.15694267e-01 6.81628585e-01 1.57463706e+00 1.56907570e+00 -1.98097080e-02 5.60967267e-01 3.59834999e-01 4.80213314e-01 1.17314875e+00 4.73823488e-01 -1.42154193e+00 5.33712029e-01 9.34841096e-01 1.89384982e-01 -1.21820521e+00 -1.03958869e+00 -8.09624195e-02 -5.27130425e-01 3.03600222e-01 -1.48909941e-01 4.96041059e-01 -1.04346561e+00 1.26397538e+00 1.30987033e-01 2.56482631e-01 2.98599750e-01 1.22576785e+00 8.85050237e-01 3.89325142e-01 -8.91980752e-02 6.99448466e-01 1.49121594e+00 -9.28252637e-01 -5.87191343e-01 -7.43039012e-01 9.33638930e-01 -1.17514901e-01 7.36302078e-01 1.17315166e-01 -2.04185337e-01 -6.93636358e-01 -1.36548924e+00 -4.14467633e-01 -7.55715013e-01 -1.60030693e-01 8.84245932e-01 1.30331200e-02 -1.35059011e+00 4.86865938e-01 -7.23554790e-01 -9.12784815e-01 3.03281188e-01 2.44652510e-01 -7.97554731e-01 -1.97578475e-01 -1.50370026e+00 1.61942649e+00 9.58025634e-01 5.94173558e-02 -9.23825383e-01 -1.78667705e-03 -1.33456945e+00 -3.37783873e-01 1.74490273e-01 -7.87359834e-01 1.10484624e+00 -1.68728992e-01 -1.38209152e+00 1.12934697e+00 -1.18632510e-01 -8.19560826e-01 2.16837838e-01 -1.63384572e-01 -4.18170065e-01 8.54922757e-02 7.14243650e-01 9.39937770e-01 8.72040689e-02 -1.14381385e+00 -1.00424409e+00 -6.72708333e-01 4.60515134e-02 5.40625632e-01 8.45108390e-01 -4.91796762e-01 -4.81951028e-01 5.68201542e-01 1.07708764e+00 -1.11494970e+00 -6.54800951e-01 -2.08887026e-01 -1.00957707e-01 -3.81899565e-01 9.67844665e-01 -5.30141890e-01 4.47428897e-02 -2.19137740e+00 -1.11901678e-01 2.46841505e-01 2.26513475e-01 -6.58067465e-02 -3.08278017e-02 1.78529859e-01 5.22745848e-01 -1.40500292e-01 -2.44075999e-01 -1.65205915e-02 2.24745527e-01 7.86747038e-01 -2.28394493e-01 3.88619632e-01 -2.25151852e-01 1.22642803e+00 -1.15153289e+00 -4.34837073e-01 6.98241055e-01 -6.63935021e-03 -2.53776193e-01 -7.13289827e-02 -3.88351440e-01 6.39560759e-01 -9.49849665e-01 3.85225385e-01 8.11772168e-01 1.08269885e-01 -3.47899646e-01 3.85165155e-01 -5.97450733e-01 5.30573726e-01 -8.94743264e-01 2.54808760e+00 -5.33351243e-01 6.76034868e-01 -7.70327523e-02 -8.45713615e-01 1.54801309e+00 -2.74337918e-01 2.14864433e-01 -1.13458657e+00 3.93404126e-01 6.03698730e-01 -3.83446187e-01 -5.82886934e-01 8.07540238e-01 1.38410226e-01 -4.95688975e-01 -3.21635544e-01 -4.73018885e-02 -1.13985205e+00 -2.39258111e-01 -3.50764603e-03 1.06878328e+00 5.03911257e-01 1.00454368e-01 -2.99005747e-01 6.38599634e-01 9.22095239e-01 3.69064271e-01 7.63818204e-01 -2.40163222e-01 3.09568375e-01 7.55536556e-02 -9.08002138e-01 -1.09065425e+00 -1.16719294e+00 2.29382105e-02 1.96495265e-01 1.12539709e+00 -1.32947490e-02 -4.23263013e-01 -2.23426789e-01 4.29914556e-02 6.95911646e-01 -1.25118762e-01 -4.83470783e-02 -6.21986926e-01 -1.24206149e-03 3.66302937e-01 1.19754575e-01 1.27374756e+00 -1.14576542e+00 -8.13786805e-01 3.58952582e-01 -4.18890685e-01 -1.53383768e+00 4.68577027e-01 3.39026004e-01 -8.25896800e-01 -1.09819698e+00 4.84909005e-02 -1.11870503e+00 4.25537169e-01 6.96041286e-01 6.48612320e-01 -2.20153123e-01 2.02682614e-01 3.29938605e-02 -4.73243624e-01 -4.80477244e-01 -2.18641669e-01 3.08627814e-01 2.29107663e-01 -5.76999307e-01 6.96116865e-01 -4.93044645e-01 -3.32045972e-01 5.55265307e-01 -1.84198245e-01 4.05323386e-01 4.75778103e-01 1.78615332e-01 5.43083131e-01 1.95559934e-01 1.58245981e-01 -7.51529559e-02 -7.56292418e-03 -4.25569355e-01 -8.63255203e-01 -4.25330937e-01 -2.55385846e-01 5.73382005e-02 3.79394710e-01 4.60038960e-01 -5.73163629e-01 4.58641499e-01 -4.42764431e-01 -1.09707899e-01 -5.04595518e-01 3.86124372e-01 -3.90930891e-01 -3.11012745e-01 4.49306548e-01 3.64508003e-01 1.13286868e-01 -3.26136023e-01 5.63418567e-01 7.73400545e-01 1.06441689e+00 -2.81505316e-01 7.76680112e-01 7.82244086e-01 3.65205377e-01 -5.12997746e-01 -5.86073399e-01 -6.80608332e-01 -1.00108337e+00 -2.80518271e-02 1.22939432e+00 -1.14713252e+00 -8.65866899e-01 5.35300195e-01 -1.48492062e+00 -3.83338511e-01 1.87545702e-01 7.36165524e-01 -1.12291181e+00 2.30072156e-01 -3.10030818e-01 -2.71786004e-01 8.79357830e-02 -1.27814209e+00 9.83243585e-01 4.37279254e-01 1.03845984e-01 -5.89677632e-01 7.21961334e-02 1.81824774e-01 1.39596879e-01 1.53816521e-01 4.54942822e-01 -6.06779516e-01 -1.25624228e+00 -2.68039733e-01 -4.89209741e-01 -2.55193442e-01 -6.53553903e-02 -7.65890598e-01 -5.79286516e-01 2.29007214e-01 1.94973633e-01 1.63407341e-01 8.16969216e-01 7.79923499e-02 6.56573236e-01 3.94417316e-01 -7.98758328e-01 9.65908527e-01 1.42329586e+00 6.73999369e-01 7.60336041e-01 1.11433017e+00 7.64534175e-01 5.20195007e-01 1.00213337e+00 5.04006371e-02 9.46005642e-01 6.05543017e-01 9.51705337e-01 1.74191207e-01 1.20582648e-01 -7.14098930e-01 -1.35874704e-01 7.59625614e-01 6.86744526e-02 1.72772676e-01 -1.21464646e+00 4.97019053e-01 -2.19804358e+00 -5.47318697e-01 -7.28282809e-01 1.77156818e+00 -4.86758687e-02 2.90024430e-01 -6.20853245e-01 -2.06279293e-01 6.58104241e-01 -7.68678710e-02 -4.20407146e-01 -9.21812654e-02 -1.94296449e-01 -1.78076282e-01 1.12244296e+00 6.68254912e-01 -1.15421569e+00 1.67302358e+00 5.32438946e+00 5.91990173e-01 -1.07131231e+00 2.42801741e-01 -3.95752609e-01 7.98151016e-01 -4.24154475e-02 3.19746017e-01 -9.30201232e-01 3.23984623e-01 8.59631121e-01 1.62346795e-01 3.81517649e-01 1.22215128e+00 2.27554247e-01 -4.46950018e-01 -7.47541189e-01 1.18567789e+00 -1.62230939e-01 -1.38312566e+00 -2.93283045e-01 6.02002889e-02 5.45054197e-01 9.64684844e-01 -2.91710168e-01 3.27559561e-01 7.31361568e-01 -8.70455146e-01 9.11646247e-01 4.56539065e-01 4.99711722e-01 -5.40498972e-01 9.78041768e-01 6.99079394e-01 -1.30541503e+00 -2.39783898e-02 -7.36989319e-01 -5.97536385e-01 4.43903089e-01 1.36849090e-01 -1.28030980e+00 9.34921265e-01 7.97885180e-01 1.03579473e+00 -3.71112704e-01 1.32873011e+00 -7.36819148e-01 -2.53508717e-01 -5.56353748e-01 -3.47713768e-01 7.54483640e-01 -6.14070535e-01 5.60474753e-01 5.84283531e-01 7.45945156e-01 -5.34550101e-02 2.88581878e-01 8.02692592e-01 4.50942069e-01 -9.29834768e-02 -1.19141424e+00 5.82292616e-01 6.15403771e-01 9.91457641e-01 -1.00971532e+00 -3.25555831e-01 5.21492101e-02 8.04397941e-01 1.33582070e-01 1.06789142e-01 -7.49274194e-01 -8.23493898e-01 6.67852581e-01 -1.37563404e-02 -3.38115804e-02 -1.12051988e+00 -5.62751710e-01 -7.82837093e-01 -2.03721717e-01 3.26260142e-02 -3.30976009e-01 -1.57602930e+00 -4.90584731e-01 4.97949511e-01 -1.88144222e-01 -1.35129571e+00 -9.47068706e-02 -7.71521807e-01 -3.72670025e-01 9.55185771e-01 -1.77968299e+00 -1.08019352e+00 -8.59244168e-01 5.52679539e-01 2.83039510e-01 -3.24265599e-01 6.04566693e-01 5.62193953e-02 2.23452449e-01 -5.31119466e-01 1.26015678e-01 1.20719366e-01 2.65984356e-01 -8.64670634e-01 1.02682185e+00 5.20185232e-01 -4.79090326e-02 2.41891712e-01 6.99220300e-01 -9.20175374e-01 -1.42963588e+00 -1.15507627e+00 9.46557283e-01 -7.15039194e-01 7.71129072e-01 -3.31583858e-01 -7.51823723e-01 8.78967881e-01 -3.93257320e-01 -3.43863487e-01 -4.08446819e-01 -2.28886187e-01 -8.23665634e-02 -2.51462549e-01 -1.17311752e+00 5.83876789e-01 1.54073155e+00 -5.21518946e-01 -9.58189309e-01 3.59470874e-01 1.22717571e+00 -8.59150589e-01 -2.98409760e-01 6.31942868e-01 1.60446957e-01 -7.69027233e-01 8.13258886e-01 -1.43442629e-02 -7.82054961e-02 -9.35359478e-01 -3.16689134e-01 -1.43090713e+00 -3.17811638e-01 -1.56643882e-01 7.49586463e-01 4.38986719e-01 8.33858475e-02 -1.12973630e+00 8.80794883e-01 9.20039490e-02 -8.95200610e-01 2.23282769e-01 -1.25669694e+00 -9.20673907e-01 -3.65323007e-01 -8.36701751e-01 1.23041153e+00 6.67625248e-01 -1.97455361e-01 1.75094068e-01 3.88678424e-02 6.52912617e-01 5.23972452e-01 -5.92740998e-02 1.12479579e+00 -1.59282351e+00 8.14395428e-01 -4.43448424e-01 -1.36542940e+00 -1.37914777e+00 6.16668701e-01 -1.21627319e+00 5.56634426e-01 -2.21925569e+00 -4.37290549e-01 -1.01668072e+00 -9.77907106e-02 5.09158731e-01 6.72796726e-01 2.83327103e-01 1.62942205e-02 4.12545115e-01 -1.03281510e+00 5.39294302e-01 1.25580895e+00 -1.81733727e-01 -6.79443702e-02 -2.16095403e-01 2.09264029e-02 8.29950035e-01 1.03023517e+00 -4.63851839e-01 -1.32871807e-01 -6.82938576e-01 4.97906655e-01 2.63418525e-01 6.54457748e-01 -1.53087091e+00 5.95394969e-01 -2.58088470e-01 3.26814391e-02 -1.20440912e+00 6.28411233e-01 -1.11367321e+00 3.04598749e-01 9.23769116e-01 2.72445261e-01 -9.95259285e-02 1.35734556e-02 3.76338422e-01 -3.09638739e-01 -3.47903430e-01 4.47558403e-01 -4.66049403e-01 -1.73530757e+00 3.99092436e-01 -4.34287429e-01 -3.74332517e-01 1.05886972e+00 -3.93927813e-01 -4.56254870e-01 -1.13179728e-01 -3.45573932e-01 7.63604939e-01 8.42977703e-01 7.40327895e-01 6.98480427e-01 -1.32378197e+00 -3.03542614e-01 5.27970552e-01 3.64513695e-01 5.29747427e-01 2.35084891e-01 5.91647983e-01 -1.50243902e+00 9.59309161e-01 -7.73734152e-01 -9.68174040e-01 -5.10097921e-01 5.37331164e-01 4.09135312e-01 3.91750604e-01 -8.37238729e-01 4.93117779e-01 1.35103375e-01 -1.22587895e+00 -2.98619211e-01 -5.95255554e-01 -2.76191235e-01 -6.51727617e-01 1.22077003e-01 2.20901236e-01 6.82180300e-02 -1.10117376e+00 -6.05202138e-01 9.18530047e-01 7.86039114e-01 -9.46863145e-02 9.80978012e-01 -5.87120056e-01 -4.05045062e-01 4.96812046e-01 9.95607495e-01 -6.89212263e-01 -1.06950045e+00 -1.44479364e-01 3.20586860e-01 -3.16551358e-01 -2.71216124e-01 -3.17615479e-01 -5.13692915e-01 9.96977746e-01 6.48032427e-01 -1.09769300e-01 3.14891309e-01 1.70818925e-01 9.86623526e-01 1.09703434e+00 1.58035076e+00 -9.59057808e-01 -3.67525697e-01 1.37692785e+00 6.16320014e-01 -1.37496316e+00 -2.41481885e-01 -2.73070514e-01 -5.75096428e-01 9.23647761e-01 5.95165431e-01 -2.65702158e-01 4.57402050e-01 -2.67579675e-01 3.40289801e-01 -3.77739102e-01 -7.53373578e-02 -6.09706700e-01 -1.88262403e-01 1.23727763e+00 -7.13339031e-01 3.13885152e-01 8.72459188e-02 4.06545311e-01 -8.96228433e-01 4.29405645e-02 3.53091061e-01 7.83626914e-01 -1.37271118e+00 -6.90115333e-01 -4.46418852e-01 -4.63328026e-02 4.59118456e-01 1.92992374e-01 5.20776287e-02 8.26127589e-01 5.52749217e-01 8.78328681e-01 4.54338729e-01 -5.38798153e-01 3.23188573e-01 3.77743654e-02 2.63577789e-01 -3.96240830e-01 2.36919105e-01 -6.34060919e-01 1.49507876e-02 -7.59391308e-01 -1.32857755e-01 -2.41804510e-01 -2.15807772e+00 -1.97979107e-01 2.34546795e-01 3.27182084e-01 1.53391147e+00 1.28049362e+00 3.30743015e-01 2.00042516e-01 3.93946379e-01 -1.08352184e+00 2.90874213e-01 -7.55542040e-01 -7.82599688e-01 1.40117645e-01 -3.67207043e-02 -8.80365193e-01 -5.34702949e-02 -4.59256858e-01]
[7.50096321105957, -2.1524956226348877]
9561c25e-9038-4179-acb6-e26d5ea5a54d
character-n-gram-embeddings-to-improve-rnn
1906.05506
null
https://arxiv.org/abs/1906.05506v1
https://arxiv.org/pdf/1906.05506v1.pdf
Character n-gram Embeddings to Improve RNN Language Models
This paper proposes a novel Recurrent Neural Network (RNN) language model that takes advantage of character information. We focus on character n-grams based on research in the field of word embedding construction (Wieting et al. 2016). Our proposed method constructs word embeddings from character n-gram embeddings and combines them with ordinary word embeddings. We demonstrate that the proposed method achieves the best perplexities on the language modeling datasets: Penn Treebank, WikiText-2, and WikiText-103. Moreover, we conduct experiments on application tasks: machine translation and headline generation. The experimental results indicate that our proposed method also positively affects these tasks.
['Masaaki Nagata', 'Jun Suzuki', 'Sho Takase']
2019-06-13
null
null
null
null
['headline-generation']
['natural-language-processing']
[-5.56324646e-02 1.10278524e-01 -7.26443172e-01 2.83728391e-02 -4.79021043e-01 -1.10283382e-01 7.12585330e-01 2.11213976e-02 -8.85675848e-01 6.97924733e-01 8.12753618e-01 -1.00880194e+00 3.97941411e-01 -8.51376414e-01 -2.75379062e-01 -3.33556205e-01 4.92299274e-02 2.14183316e-01 -2.60791391e-01 -3.74615967e-01 3.25114638e-01 3.29193443e-01 -7.57389724e-01 -3.67852785e-02 8.39262903e-01 3.64342064e-01 2.69867241e-01 7.87008166e-01 -6.44781232e-01 5.64212143e-01 -4.31444466e-01 -6.41109407e-01 3.66771109e-02 -1.71566203e-01 -6.32129610e-01 -2.59360313e-01 -1.49230137e-01 -3.16194445e-01 -8.26584101e-01 1.02338576e+00 6.43976986e-01 1.87375844e-01 7.64630079e-01 -7.11128712e-01 -1.66294205e+00 1.04315758e+00 -4.71660078e-01 3.69917065e-01 -7.81072900e-02 -2.06293985e-02 1.44516361e+00 -1.27164519e+00 4.59398597e-01 1.29887140e+00 4.90518838e-01 7.34315634e-01 -1.00078392e+00 -4.72710609e-01 3.00386637e-01 1.16071321e-01 -1.09353936e+00 -2.24326774e-01 7.14574873e-01 -2.50101984e-01 1.51254022e+00 -9.28172693e-02 5.01898229e-01 1.51595259e+00 4.67426956e-01 9.46931362e-01 6.29032433e-01 -9.19150889e-01 -1.23953067e-01 1.92863807e-01 7.28614330e-01 5.97719669e-01 4.84968454e-01 -5.88694327e-02 -2.16659829e-01 -1.60436556e-01 7.89406240e-01 8.99760276e-02 -5.58768548e-02 1.86774939e-01 -1.30467522e+00 1.18405950e+00 -1.46330409e-02 5.48804879e-01 -5.06632745e-01 3.85608494e-01 5.99166989e-01 3.08766991e-01 8.10605347e-01 5.99305451e-01 -4.30106610e-01 -5.09599030e-01 -3.92306060e-01 -1.31318301e-01 6.96328163e-01 9.86991525e-01 4.77489740e-01 5.02837360e-01 -3.41195732e-01 1.25631332e+00 3.02758634e-01 3.66604716e-01 1.32405305e+00 -3.69733155e-01 5.97167373e-01 2.69468933e-01 7.33623207e-02 -9.41075802e-01 -2.38642320e-02 -3.44322026e-01 -6.67685032e-01 -5.56720018e-01 -1.25786617e-01 -5.85417569e-01 -8.12513530e-01 1.52896631e+00 -2.43990600e-01 1.43378869e-01 4.38832283e-01 4.91777509e-01 8.38494420e-01 1.18570352e+00 1.16343856e-01 -6.74469098e-02 1.30452538e+00 -1.54996383e+00 -1.13616729e+00 -3.43791008e-01 1.04606533e+00 -8.12564790e-01 1.24954629e+00 9.63238534e-03 -9.05385077e-01 -4.67319280e-01 -9.91262197e-01 -2.21748292e-01 -5.50408781e-01 5.03508508e-01 8.79230678e-01 4.90605772e-01 -9.30611551e-01 5.09408951e-01 -8.03653538e-01 -4.12556589e-01 -1.01124577e-01 -1.59104448e-02 -1.43843770e-01 7.14481547e-02 -1.58445871e+00 1.12544096e+00 5.67294836e-01 1.81859434e-01 -3.95799160e-01 -3.30327421e-01 -1.08205426e+00 8.10065642e-02 -2.19734952e-01 -5.31127691e-01 1.21855295e+00 -5.38605332e-01 -1.91908789e+00 4.20721024e-01 -3.22536200e-01 -6.18175328e-01 -9.17075649e-02 -7.26379037e-01 -5.76849520e-01 -2.48098940e-01 -2.35260174e-01 4.66250360e-01 3.96333337e-01 -5.87739229e-01 -4.18769091e-01 8.38182345e-02 4.56420407e-02 -2.12941375e-02 -1.21498954e+00 1.77444279e-01 -1.98433086e-01 -1.00708628e+00 -2.97218502e-01 -6.92546070e-01 -5.70154309e-01 -3.42365950e-01 -3.42425048e-01 -6.69406474e-01 3.54023665e-01 -8.29872131e-01 1.77697802e+00 -1.80984974e+00 5.56999482e-02 -2.85158873e-01 -4.36028354e-02 6.58261001e-01 -7.48320401e-01 9.77526903e-01 -7.64294788e-02 3.77863407e-01 -3.22437510e-02 -3.40025485e-01 1.79675519e-01 3.74839783e-01 -6.78900599e-01 1.61121249e-01 3.09225351e-01 1.27288795e+00 -9.21009123e-01 -2.34808773e-01 1.20045897e-02 5.00091314e-01 -3.73182893e-01 3.11638057e-01 -9.53687206e-02 -5.83283901e-01 -3.19121212e-01 3.37780952e-01 2.98219144e-01 1.23992808e-01 2.42196500e-01 1.82570696e-01 -2.13183045e-01 8.97754729e-01 -4.43281710e-01 1.61305332e+00 -1.03361356e+00 7.30908692e-01 -6.72661483e-01 -8.01662922e-01 1.17992640e+00 5.24654031e-01 4.57432121e-03 -4.16522652e-01 2.72933871e-01 1.54097170e-01 2.63286740e-01 -6.34329140e-01 9.68418777e-01 1.38744460e-02 1.12198487e-01 8.39302480e-01 3.03990304e-01 2.21061110e-01 2.93649882e-01 2.17879444e-01 8.73965919e-01 -5.56451194e-02 4.84064192e-01 -1.91881090e-01 3.68822694e-01 -3.48740637e-01 4.05603170e-01 4.35099006e-01 -1.08330198e-01 3.37819844e-01 5.61231792e-01 -2.91854203e-01 -1.40872324e+00 -9.42853451e-01 5.53993545e-02 1.29635549e+00 -4.23646152e-01 -7.86784887e-01 -4.94966418e-01 -6.37284219e-01 -1.21162899e-01 1.08946598e+00 -6.50156021e-01 -1.92688778e-01 -6.94138885e-01 -8.44049037e-01 9.34440136e-01 7.51741707e-01 -8.11668783e-02 -1.26863444e+00 1.10279799e-01 5.69126785e-01 -2.31824163e-02 -1.08917534e+00 -6.86621070e-01 -2.97478437e-02 -1.10297036e+00 -3.88721764e-01 -9.34317172e-01 -1.29580247e+00 6.45497203e-01 3.14848363e-01 9.09739494e-01 -2.43074730e-01 7.32096210e-02 -2.63616210e-03 -6.40833616e-01 -3.67982388e-01 -3.79306227e-01 5.92817485e-01 2.55027682e-01 -2.08227426e-01 9.63730752e-01 -6.47632778e-01 -1.44087315e-01 -3.68387908e-01 -9.02709961e-01 9.38324705e-02 8.19483340e-01 9.91242111e-01 1.50706872e-01 -7.47889280e-01 7.62144506e-01 -8.82510424e-01 1.51297915e+00 -6.32359505e-01 -3.73292595e-01 3.58439535e-01 -9.58269477e-01 3.87142688e-01 7.38679230e-01 -8.19110692e-01 -7.91056395e-01 -6.05364561e-01 -3.81132066e-01 -9.50741917e-02 2.30311722e-01 8.30332398e-01 2.71825999e-01 3.98014188e-01 3.13080192e-01 6.06939495e-01 -7.94834793e-02 -6.83046281e-01 8.47963274e-01 1.13191617e+00 1.37176603e-01 -6.09356701e-01 6.90734446e-01 -8.30621943e-02 -7.11126685e-01 -9.28825736e-01 -4.58247185e-01 -3.61242265e-01 -4.61027831e-01 3.21974844e-01 7.89262891e-01 -8.20472658e-01 -2.00321794e-01 1.33594215e-01 -1.75935388e+00 9.30011198e-02 -1.44515978e-02 9.30625439e-01 -2.13840142e-01 5.91723025e-01 -1.14171159e+00 -8.64079714e-01 -8.00462902e-01 -9.83155429e-01 5.85918188e-01 4.60739769e-02 -4.56524074e-01 -1.44528294e+00 6.43062294e-01 -1.95131496e-01 5.57923377e-01 -3.24185669e-01 1.24710572e+00 -9.28217530e-01 -4.86783460e-02 -2.48827547e-01 -2.44948208e-01 6.35780096e-01 2.89915860e-01 9.33697298e-02 -6.83772326e-01 -1.89673714e-02 -3.60674858e-01 -2.18737140e-01 1.05392003e+00 1.33186311e-01 1.10098422e+00 -5.33405721e-01 7.88069591e-02 4.42604750e-01 1.41702604e+00 4.06413265e-02 7.06849039e-01 4.85338062e-01 7.81823337e-01 5.17332971e-01 2.15092734e-01 3.14565003e-01 2.29909182e-01 3.14037204e-01 6.26363605e-02 4.08767313e-02 1.49662346e-01 -5.54826319e-01 8.45863998e-01 2.05252409e+00 4.23838571e-02 -3.61390829e-01 -7.64380455e-01 1.05293512e+00 -1.85371435e+00 -6.20481133e-01 -2.58038267e-02 1.64296019e+00 1.09249151e+00 1.75136790e-01 -2.95489877e-01 -2.42687449e-01 6.29208028e-01 5.87364078e-01 -7.13666230e-02 -1.28224778e+00 2.19363645e-02 6.90195084e-01 6.76161706e-01 6.79719448e-01 -6.68710709e-01 1.39438367e+00 6.32474232e+00 8.51335764e-01 -9.28327382e-01 1.93994835e-01 2.91668147e-01 3.46519314e-02 -8.03325653e-01 -1.99349418e-01 -9.04689610e-01 4.73611951e-01 1.32615411e+00 -6.20803833e-01 2.39514187e-01 9.11791325e-01 2.98736304e-01 5.52685738e-01 -9.15283680e-01 8.65981042e-01 2.55476862e-01 -1.43675280e+00 5.07598698e-01 -1.66071802e-02 6.80933475e-01 1.99696124e-01 1.55049905e-01 5.66364586e-01 6.71402752e-01 -1.16870368e+00 1.33299500e-01 3.03083181e-01 7.04338431e-01 -1.04999626e+00 9.48529482e-01 2.67840326e-01 -9.31532443e-01 1.32279530e-01 -8.68137598e-01 -3.62055987e-01 4.36763585e-01 7.55864978e-01 -1.00013268e+00 3.77472222e-01 -1.23638466e-01 9.41377282e-01 -2.15441719e-01 5.55613577e-01 -7.15486348e-01 1.10636246e+00 2.23081172e-01 -6.83692515e-01 7.11306214e-01 -4.51997966e-01 4.27723169e-01 1.64027143e+00 2.53568888e-01 -1.52345255e-01 -1.54163390e-01 8.20428193e-01 -4.35764968e-01 4.98655409e-01 -8.62208307e-01 -9.07031298e-01 5.45001328e-01 1.20918572e+00 -1.41945034e-01 -3.73336762e-01 -5.71381450e-01 9.62539911e-01 6.54851377e-01 5.15195131e-01 -6.83948100e-01 -1.01282799e+00 1.22608078e+00 -3.79028261e-01 3.41961861e-01 -7.45420098e-01 -1.94843635e-01 -1.35466468e+00 -1.59674119e-02 -6.97253883e-01 -1.99640244e-01 -4.52126294e-01 -1.59791112e+00 8.87755513e-01 -2.76638687e-01 -9.03334677e-01 -4.81162012e-01 -9.57527936e-01 -8.08175325e-01 1.08969665e+00 -1.77086830e+00 -9.73571777e-01 5.59917033e-01 -1.79704159e-01 8.85532796e-01 -4.67669517e-01 1.14643061e+00 4.27638292e-01 -8.80342066e-01 8.83194208e-01 5.68040609e-01 6.38851702e-01 5.62083900e-01 -1.00992870e+00 1.22786129e+00 8.34674060e-01 3.29657257e-01 1.20149529e+00 3.51339996e-01 -4.96367037e-01 -1.39625251e+00 -1.25334859e+00 1.75393140e+00 -1.65994838e-01 1.22205305e+00 -5.78445792e-01 -8.19163740e-01 9.02294934e-01 7.27428138e-01 -3.55614215e-01 1.00784528e+00 2.41980821e-01 -3.73593211e-01 3.42335433e-01 -4.36167359e-01 1.06771815e+00 9.27664280e-01 -6.50801718e-01 -1.04270136e+00 4.08744752e-01 1.48625839e+00 1.36343136e-01 -6.40745342e-01 7.10958242e-02 6.07277095e-01 -1.65120244e-01 7.92852700e-01 -1.24098587e+00 9.28417444e-01 3.00725996e-01 -1.46591842e-01 -1.74798548e+00 -5.52113891e-01 -5.38824081e-01 -3.58838081e-01 1.16954541e+00 7.29848385e-01 -8.65406692e-01 4.49021220e-01 2.10979655e-01 -1.78183824e-01 -1.16784275e+00 -5.73499084e-01 -1.05162477e+00 4.76751745e-01 -5.33028424e-01 5.10648251e-01 9.50107038e-01 2.17045397e-01 5.25310576e-01 -6.58479571e-01 -2.25357458e-01 2.53421012e-02 -3.58445734e-01 5.81252933e-01 -7.10049808e-01 -3.81906599e-01 -4.28443342e-01 -1.90237612e-01 -1.53161323e+00 6.33736908e-01 -1.04039633e+00 -8.62002149e-02 -1.70248306e+00 3.98245640e-02 -3.88762308e-03 -5.42743981e-01 2.43551195e-01 -4.82800663e-01 -5.30293174e-02 1.11312665e-01 -1.87133536e-01 -2.05191717e-01 1.01616907e+00 8.59572828e-01 -1.66417703e-01 -1.92552677e-03 -4.66172576e-01 -7.17565715e-01 5.27585208e-01 1.05316174e+00 -4.93668407e-01 -2.33360454e-01 -1.04276037e+00 1.60307586e-01 -4.22493160e-01 -3.34304750e-01 -4.39313740e-01 -1.33993924e-01 -2.61984020e-01 -1.72036663e-01 -4.62721258e-01 2.60570049e-01 -4.04940128e-01 -7.36500859e-01 5.17410636e-01 -7.50646114e-01 8.04904461e-01 1.10017903e-01 4.33980674e-01 -1.84969708e-01 -6.47805631e-01 2.77842343e-01 -7.19962195e-02 -5.42835951e-01 3.20878208e-01 -7.72158802e-01 5.51865511e-02 5.17487705e-01 2.23172107e-03 -9.41688120e-02 -3.47375184e-01 -2.52947956e-01 1.14291869e-01 -5.38017638e-02 8.56867671e-01 8.77717197e-01 -1.79829407e+00 -9.11714017e-01 1.51360556e-01 1.98827505e-01 -5.89069009e-01 -4.64990884e-01 4.04333383e-01 -6.08074486e-01 7.65372097e-01 -5.98947741e-02 1.96756557e-01 -9.73826528e-01 3.26045066e-01 -1.09717831e-01 -6.03904724e-01 -4.49248344e-01 7.95521200e-01 -6.61177337e-02 -7.11383224e-01 2.46157229e-01 -5.61868072e-01 -5.41656613e-01 -1.34406816e-02 7.23852813e-01 4.22503620e-01 -2.61505008e-01 -2.63510495e-01 4.65747118e-02 2.47036546e-01 -5.10149479e-01 -3.86742026e-01 1.59725797e+00 -4.03014235e-02 -4.17599440e-01 7.95168757e-01 1.54409587e+00 7.88810924e-02 -4.02430475e-01 -5.07579744e-01 2.44957432e-01 -2.95641989e-01 -1.09595217e-01 -1.83937714e-01 -7.63076961e-01 1.24889517e+00 1.53606042e-01 -9.32609290e-02 4.60513264e-01 -4.93549526e-01 1.64330590e+00 7.28777289e-01 2.80812616e-03 -1.31283426e+00 4.25156467e-02 1.18607712e+00 6.10206008e-01 -8.46727073e-01 -2.86913455e-01 1.26511753e-01 -3.50562662e-01 1.47278452e+00 6.76174581e-01 -4.48966026e-01 6.01331770e-01 4.24555577e-02 1.93550542e-01 2.58670509e-01 -1.18988717e+00 -5.23697399e-02 8.89326632e-02 4.48800117e-01 9.48443234e-01 1.66171670e-01 -1.12908030e+00 6.78109229e-01 -2.22440094e-01 -1.38628095e-01 7.80372798e-01 7.57057667e-01 -6.22892141e-01 -1.48270750e+00 -5.51113114e-02 4.77487922e-01 -5.09447694e-01 -8.16408932e-01 -3.95902902e-01 4.68666524e-01 -3.25773716e-01 7.00794756e-01 1.58989877e-01 -5.26164591e-01 2.26171296e-02 4.57247317e-01 1.38876870e-01 -1.02551150e+00 -4.09181744e-01 -3.09952021e-01 3.77219766e-01 -5.14610261e-02 -6.75469860e-02 -5.36963120e-02 -1.13780940e+00 -3.67046297e-01 -5.13084888e-01 4.60523695e-01 8.37932527e-01 7.03701794e-01 3.38875622e-01 6.75495267e-01 7.26725221e-01 -4.23646718e-01 -9.10363317e-01 -1.56804144e+00 -4.07620817e-01 5.54230325e-02 1.75485015e-01 -2.33700961e-01 -3.23359638e-01 -1.61500677e-01]
[10.920461654663086, 8.86082935333252]
0989d8fb-3fbf-4b90-baeb-2dfce6d79bb3
ju_cse-a-conditional-random-field-crf-based
null
null
https://aclanthology.org/S14-2063
https://aclanthology.org/S14-2063.pdf
JU\_CSE: A Conditional Random Field (CRF) Based Approach to Aspect Based Sentiment Analysis
null
['Sivaji yopadhyay', 'B', 'Soumik al', 'M', 'Dipankar Das', 'Braja Gopal Patra']
2014-08-01
null
null
null
semeval-2014-8
['subjectivity-analysis']
['natural-language-processing']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.2457170486450195, 3.8112142086029053]
c41a549f-645c-455a-9f24-8b5f66731c30
evaluation-of-latent-space-disentanglement-in
2110.05587
null
https://arxiv.org/abs/2110.05587v1
https://arxiv.org/pdf/2110.05587v1.pdf
Evaluation of Latent Space Disentanglement in the Presence of Interdependent Attributes
Controllable music generation with deep generative models has become increasingly reliant on disentanglement learning techniques. However, current disentanglement metrics, such as mutual information gap (MIG), are often inadequate and misleading when used for evaluating latent representations in the presence of interdependent semantic attributes often encountered in real-world music datasets. In this work, we propose a dependency-aware information metric as a drop-in replacement for MIG that accounts for the inherent relationship between semantic attributes.
['Alexander Lerch', 'Karn N. Watcharasupat']
2021-10-11
null
null
null
null
['music-generation', 'music-generation']
['audio', 'music']
[ 2.55976081e-01 -1.51843861e-01 -8.62689018e-02 -3.42667818e-01 -6.97142959e-01 -7.09621489e-01 8.02028596e-01 2.08285540e-01 -2.07625493e-01 8.68719518e-01 5.74620485e-01 7.17782825e-02 -6.66871369e-01 -7.12383032e-01 -1.77509665e-01 -5.77955842e-01 1.39200866e-01 5.54499865e-01 -4.60164517e-01 -1.26331717e-01 1.51988268e-01 1.54367536e-01 -1.68822527e+00 -2.55090464e-02 9.66078281e-01 6.20020747e-01 -1.91214874e-01 5.02366126e-01 -6.61463812e-02 7.14838326e-01 -8.31000328e-01 -4.96564895e-01 1.74929827e-01 -6.95061326e-01 -3.27957779e-01 -3.60182464e-01 2.42128912e-02 1.22848153e-01 -1.74847424e-01 9.65406716e-01 4.79339063e-01 6.09989353e-02 8.87828767e-01 -1.40753579e+00 -4.99907196e-01 9.08433259e-01 -2.65129149e-01 1.11428864e-01 3.09658319e-01 -2.15159096e-02 1.65010428e+00 -7.10911334e-01 3.91063988e-01 1.09468496e+00 5.26668608e-01 4.33466554e-01 -1.65208626e+00 -9.84138489e-01 -1.59271732e-01 2.77946323e-01 -1.31073844e+00 -3.65620166e-01 1.11927640e+00 -4.47856992e-01 6.60239875e-01 3.71382415e-01 5.10736763e-01 1.39244485e+00 -1.79953709e-01 6.49999201e-01 1.07261801e+00 -4.12054569e-01 1.54445589e-01 -3.87951769e-02 -3.41723152e-02 2.56261081e-01 7.43277252e-01 2.55171984e-01 -9.56779242e-01 -2.03345358e-01 8.82540047e-01 -2.03597873e-01 -4.32191007e-02 -6.27462268e-01 -1.29765737e+00 1.00331748e+00 2.60116041e-01 2.67481476e-01 -8.62613469e-02 4.58803475e-02 3.61091852e-01 5.15303135e-01 5.15585124e-01 1.11613381e+00 -2.73535043e-01 -6.63694680e-01 -8.91996086e-01 4.11701441e-01 7.53353953e-01 5.32781482e-01 6.33374333e-01 2.12917626e-01 -1.24946274e-02 7.73410797e-01 2.86183387e-01 1.43401578e-01 5.57213068e-01 -7.50086546e-01 2.59797066e-01 9.00116146e-01 4.27825414e-02 -9.58730280e-01 -1.80848271e-01 -7.59503663e-01 -7.35138893e-01 1.67025492e-01 2.11390883e-01 -6.59202505e-03 -7.16349483e-01 2.08368754e+00 7.54056918e-03 1.81346744e-01 2.36271679e-01 8.20742726e-01 6.27889216e-01 -7.17591569e-02 6.99580833e-02 -8.96510854e-02 8.10638011e-01 -4.95090961e-01 -6.78401649e-01 -3.24432820e-01 3.43741983e-01 -6.86717987e-01 1.22624516e+00 2.74638414e-01 -8.07827055e-01 -4.96311188e-01 -1.30503380e+00 6.42859703e-03 -2.22693637e-01 -2.43704662e-01 8.75259221e-01 6.19799137e-01 -5.46105385e-01 7.22128868e-01 -6.01071060e-01 1.77577391e-01 1.72272861e-01 3.57448667e-01 -4.36472356e-01 1.48020461e-01 -1.29914486e+00 8.93152356e-01 3.98569316e-01 -3.25042754e-02 -6.87845170e-01 -7.72205293e-01 -7.49863625e-01 1.45492330e-01 4.76076722e-01 -8.97974372e-01 8.30248177e-01 -7.23478258e-01 -1.37610161e+00 5.83877861e-01 2.82119840e-01 -1.35846660e-01 4.97298181e-01 -4.98501450e-01 -5.39052904e-01 -5.09014428e-01 3.02464515e-02 2.39841610e-01 7.91055620e-01 -1.28854942e+00 5.11354916e-02 -5.27547956e-01 -1.90755911e-03 4.35230345e-01 -1.84647039e-01 -2.55304009e-01 1.97714552e-01 -9.48445320e-01 3.44575465e-01 -9.46003437e-01 -7.70805702e-02 -4.49235141e-01 -5.05768895e-01 4.81643863e-02 5.82048655e-01 -2.20252216e-01 1.32408702e+00 -2.04656410e+00 6.49396002e-01 2.31390372e-01 3.32881510e-01 1.69303074e-01 -3.97986695e-02 5.35491347e-01 -2.17560112e-01 1.20529691e-02 -2.00073287e-01 -3.92625540e-01 9.57414061e-02 3.31809014e-01 -2.43908778e-01 3.38818952e-02 6.36265874e-02 6.78284287e-01 -8.82827997e-01 -2.16374218e-01 4.27404605e-02 5.51566243e-01 -5.15742660e-01 2.02004105e-01 -1.62022859e-01 5.74553728e-01 -3.83257329e-01 3.00507069e-01 6.95549101e-02 -1.96922690e-01 3.06163579e-01 -2.61029333e-01 1.41950712e-01 6.27529085e-01 -9.72409308e-01 1.99963546e+00 -5.19946933e-01 6.28302217e-01 -4.99010444e-01 -5.96030772e-01 1.21785319e+00 3.82116199e-01 3.17278117e-01 -3.97900403e-01 1.50378928e-01 2.44960964e-01 4.03112054e-01 -6.69584349e-02 4.25085813e-01 -3.25962991e-01 -2.15672135e-01 4.59771186e-01 1.82764872e-03 -1.75059345e-02 2.02454954e-01 1.19152986e-01 1.14010632e+00 2.50812799e-01 4.06266183e-01 -2.27699041e-01 1.85286134e-01 -4.45094675e-01 6.71512544e-01 3.91289771e-01 1.30757660e-01 8.61590922e-01 6.28953815e-01 -2.61797428e-01 -1.01797009e+00 -1.37657475e+00 3.21209282e-02 5.47988355e-01 -4.09207717e-02 -7.45909095e-01 -3.79890442e-01 -5.88091373e-01 1.03872970e-01 8.62501860e-01 -5.46591938e-01 -5.81742942e-01 -2.18575537e-01 -9.00614202e-01 6.88390851e-01 7.00176656e-01 7.09292367e-02 -7.03981519e-01 -4.97961819e-01 2.91736454e-01 -2.76125580e-01 -8.12013328e-01 -5.00103123e-02 3.03108573e-01 -7.73185134e-01 -1.01982224e+00 -2.52801597e-01 4.35451157e-02 2.64773816e-01 -1.58209607e-01 1.45599401e+00 -3.65371943e-01 -3.06181341e-01 -8.64585564e-02 -2.88113922e-01 -2.68402696e-01 -3.98040384e-01 2.78382063e-01 4.49123494e-02 -8.34813789e-02 4.62124258e-01 -1.13114429e+00 -6.52628005e-01 2.35542253e-01 -8.34537446e-01 4.10258144e-01 6.62689984e-01 9.08916712e-01 3.83026868e-01 -1.35543197e-01 9.48480129e-01 -9.94886279e-01 1.01250279e+00 -4.37805593e-01 -7.60331377e-02 2.10153461e-01 -1.14888525e+00 6.02309108e-01 2.26758108e-01 -5.32595694e-01 -9.01418269e-01 -4.47717398e-01 2.45622560e-01 -5.33814847e-01 -1.37945965e-01 6.93622172e-01 -3.94668877e-01 5.78736722e-01 6.79534733e-01 -8.99824426e-02 -2.10491061e-01 -4.97142941e-01 4.74519432e-01 4.02508497e-01 4.25161570e-01 -5.80203474e-01 6.57664418e-01 2.21557431e-02 1.21867284e-01 -1.50077015e-01 -8.24826241e-01 -1.84107095e-01 -5.68519533e-01 1.61620993e-02 6.25777245e-01 -9.19326961e-01 -4.43460435e-01 1.87192291e-01 -7.44652748e-01 8.44121352e-02 -6.71930909e-01 6.89448297e-01 -6.68183386e-01 -6.71084970e-02 -1.93019092e-01 -6.53523088e-01 -1.83267951e-01 -9.68348444e-01 8.40453506e-01 3.52234878e-02 -7.98595428e-01 -1.01171017e+00 4.72234994e-01 4.76234436e-01 4.04660374e-01 5.47979414e-01 1.20020497e+00 -7.34830737e-01 -3.25875819e-01 -4.43450838e-01 -1.18700400e-01 2.11014956e-01 4.69113320e-01 -7.75301680e-02 -1.02207208e+00 -1.39483819e-02 -1.51165545e-01 -2.01762527e-01 7.38058567e-01 -3.38986218e-02 7.60090888e-01 -3.12459528e-01 5.31287938e-02 3.25153649e-01 1.30215394e+00 8.81822929e-02 4.37066227e-01 1.79002717e-01 9.63877797e-01 4.15984720e-01 1.69310078e-01 4.43268597e-01 2.21339092e-01 7.39531994e-01 3.07783216e-01 2.67632186e-01 -1.92134351e-01 -6.22367322e-01 1.37125909e-01 9.92904484e-01 -2.02042535e-01 -4.18254375e-01 -9.05378222e-01 2.99120486e-01 -1.95732057e+00 -6.33556724e-01 1.09747730e-01 2.30067587e+00 9.77960408e-01 2.56339937e-01 -6.03173375e-02 6.20121717e-01 4.00588781e-01 2.66399354e-01 -6.41993821e-01 -2.30908200e-01 -1.87468335e-01 2.35837221e-01 7.81645551e-02 2.30267733e-01 -7.32466578e-01 6.28453612e-01 5.99570513e+00 5.62194645e-01 -7.31348693e-01 -2.43082605e-02 1.47401541e-01 -1.95978031e-01 -8.29677820e-01 2.86097050e-01 -2.19795331e-01 3.68189365e-01 6.53350770e-01 -2.41150990e-01 2.56593674e-01 5.89087129e-01 -2.37145662e-01 1.26713142e-01 -1.49689484e+00 1.16186070e+00 1.25656873e-01 -7.90187359e-01 2.88877755e-01 4.23681401e-02 7.10621595e-01 -3.68106544e-01 5.05261660e-01 1.28614195e-02 4.66275424e-01 -1.13466620e+00 5.41697502e-01 4.41335976e-01 5.96746027e-01 -8.28202784e-01 5.76581776e-01 5.92126958e-02 -8.24072242e-01 1.26736850e-01 1.72379598e-01 -2.69690424e-01 -3.57906371e-02 6.96834743e-01 -7.60007560e-01 5.55697739e-01 1.83226600e-01 7.13012099e-01 -4.73200858e-01 8.76256764e-01 -5.32192051e-01 4.33970511e-01 4.62536924e-02 1.42814562e-01 -7.87225291e-02 -3.06013256e-01 7.20516264e-01 5.59526026e-01 2.83520848e-01 -3.67057949e-01 -3.22225720e-01 1.06944680e+00 2.47183722e-02 -2.82101333e-02 -6.62803710e-01 -4.81385648e-01 4.87070352e-01 8.81168425e-01 -5.80472469e-01 -9.53147039e-02 -1.56907886e-01 1.01067007e+00 2.15815723e-01 2.10940599e-01 -7.11068928e-01 -1.38944879e-01 1.20122361e+00 -5.41736633e-02 -1.38561293e-01 -5.45111299e-01 -4.44136649e-01 -1.21349907e+00 -8.15225318e-02 -9.47804749e-01 2.50628322e-01 -7.51546264e-01 -1.40784419e+00 5.90124965e-01 1.51319176e-01 -1.25744009e+00 -6.28743947e-01 -1.23554155e-01 -3.63363266e-01 7.62323081e-01 -9.22663629e-01 -1.08628345e+00 -1.44173652e-01 3.62670004e-01 4.70743924e-01 -2.14660451e-01 1.27649534e+00 3.22994232e-01 -5.70418239e-01 6.51383042e-01 -5.13772294e-02 -2.40322538e-02 6.20200098e-01 -1.38024867e+00 2.21636459e-01 5.55074871e-01 8.55481505e-01 6.65321112e-01 1.00935614e+00 -3.73635978e-01 -9.57208574e-01 -5.80500603e-01 7.97373772e-01 -5.99760234e-01 5.71996808e-01 -3.97716612e-01 -7.74723291e-01 5.31532407e-01 -9.80156511e-02 -3.39003712e-01 1.15063643e+00 6.39359295e-01 -9.79569495e-01 -3.88155431e-02 -7.06289649e-01 6.95913196e-01 1.18577123e+00 -7.74230182e-01 -4.21692431e-01 -1.36245206e-01 6.46114826e-01 -1.91948488e-02 -1.00836277e+00 6.82270229e-01 8.79801750e-01 -8.46972942e-01 9.18089390e-01 -7.42535055e-01 4.59409326e-01 -1.84546992e-01 -2.23546237e-01 -1.58427846e+00 -2.49270394e-01 -5.23012102e-01 -1.05245627e-01 1.26411855e+00 6.19749963e-01 -2.25642309e-01 8.58651638e-01 7.00402021e-01 2.25937352e-01 -4.36410159e-01 -9.06854570e-01 -7.73243725e-01 -2.27943301e-01 -3.67641360e-01 7.89606452e-01 1.22114849e+00 1.90224916e-01 8.78644764e-01 -4.57770437e-01 -2.48496443e-01 6.26338899e-01 2.40330398e-01 6.04467154e-01 -1.55082989e+00 -5.16713977e-01 -5.06584585e-01 -7.91497231e-01 -1.84590206e-01 8.80754739e-02 -7.56363928e-01 -2.16655999e-01 -1.30445385e+00 1.95132807e-01 -6.00183785e-01 -6.44010127e-01 2.13870987e-01 -3.78695309e-01 2.29360968e-01 1.82331234e-01 3.46890658e-01 -2.02373385e-01 8.43338311e-01 9.55471992e-01 -5.27688973e-02 -1.99297637e-01 2.56976020e-02 -7.82998145e-01 7.67961144e-01 1.00847244e+00 -6.03002131e-01 -8.67702484e-01 -4.46874470e-01 6.98000371e-01 -5.14800800e-03 3.05845737e-01 -1.09026015e+00 -1.90779418e-02 -2.23826431e-02 4.07718532e-02 -1.24570921e-01 5.89699030e-01 -6.79975450e-01 7.50604451e-01 1.50461182e-01 -7.02175856e-01 2.82405913e-01 -7.50241950e-02 6.63564742e-01 -5.32370806e-01 9.66034830e-02 2.10499197e-01 2.00023465e-02 -3.83641720e-01 6.00494854e-02 1.04633987e-01 1.72374740e-01 6.24896049e-01 -2.11924657e-01 -1.89022332e-01 -3.59378666e-01 -8.78283978e-01 -4.79559362e-01 2.14927718e-01 7.72201061e-01 5.77050030e-01 -1.40204573e+00 -8.12030256e-01 2.50629723e-01 3.12209129e-01 -2.78284222e-01 1.11652508e-01 4.90749270e-01 -5.72159179e-02 1.93055511e-01 -3.29225123e-01 -9.50875580e-02 -1.27609420e+00 2.72842139e-01 -6.71470687e-02 -4.40823466e-01 -3.03050548e-01 7.48306453e-01 1.82859913e-01 -1.78886242e-02 -6.69499859e-02 -1.52092785e-01 -1.07779324e-01 1.71770245e-01 4.44034152e-02 4.76646811e-01 6.86379075e-02 -5.21430790e-01 -2.12674081e-01 1.72709137e-01 2.92223040e-02 -4.08217490e-01 1.15963840e+00 3.99647951e-02 1.51990131e-01 9.51463759e-01 9.34578896e-01 7.73318671e-03 -1.11627555e+00 -1.59051716e-01 1.35610461e-01 -6.92311347e-01 2.50826459e-02 -9.35338020e-01 -8.29420149e-01 7.70303667e-01 5.36939502e-01 2.56750882e-01 9.77735341e-01 -7.43096545e-02 4.33153749e-01 2.90853977e-02 1.94662392e-01 -8.07391703e-01 1.99744985e-01 4.85820249e-02 8.41077685e-01 -1.00511992e+00 2.03154944e-02 -3.48384708e-01 -6.84242487e-01 6.75359309e-01 5.06154358e-01 -6.84347674e-02 5.91327786e-01 2.78978139e-01 -9.85195786e-02 -1.00532934e-01 -8.82725954e-01 -2.14147568e-01 5.51561773e-01 4.85898703e-01 7.49438703e-01 4.62806076e-01 -3.31724197e-01 5.07691920e-01 -7.19087780e-01 -2.42654949e-01 1.82614118e-01 4.65409964e-01 5.64134791e-02 -1.48082232e+00 2.01952577e-01 3.06970894e-01 -3.23265821e-01 -2.53431022e-01 -5.21164238e-01 6.85092747e-01 2.05786467e-01 7.30288446e-01 1.15687370e-01 -6.88186944e-01 2.61175901e-01 9.06039551e-02 7.11795032e-01 -5.83543897e-01 -3.61888796e-01 -2.68270701e-01 2.19791174e-01 -2.60280371e-01 -4.99754548e-01 -6.07478857e-01 -8.37366998e-01 -8.39221403e-02 -5.69159627e-01 1.30396545e-01 7.98407674e-01 7.87282467e-01 4.56994116e-01 5.97336709e-01 4.50265110e-01 -3.07078332e-01 -5.31065464e-01 -1.09002006e+00 -5.48832893e-01 6.39850080e-01 1.34282112e-01 -1.01123846e+00 -3.71898115e-01 -1.51313275e-01]
[9.313465118408203, 4.8592963218688965]
664974ee-7393-4c99-bc8b-32d24be6506d
band-biomedical-alert-news-dataset
2305.1448
null
https://arxiv.org/abs/2305.14480v1
https://arxiv.org/pdf/2305.14480v1.pdf
BAND: Biomedical Alert News Dataset
Infectious disease outbreaks continue to pose a significant threat to human health and well-being. To improve disease surveillance and understanding of disease spread, several surveillance systems have been developed to monitor daily news alerts and social media. However, existing systems lack thorough epidemiological analysis in relation to corresponding alerts or news, largely due to the scarcity of well-annotated reports data. To address this gap, we introduce the Biomedical Alert News Dataset (BAND), which includes 1,508 samples from existing reported news articles, open emails, and alerts, as well as 30 epidemiology-related questions. These questions necessitate the model's expert reasoning abilities, thereby offering valuable insights into the outbreak of the disease. The BAND dataset brings new challenges to the NLP world, requiring better disguise capability of the content and the ability to infer important information. We provide several benchmark tasks, including Named Entity Recognition (NER), Question Answering (QA), and Event Extraction (EE), to show how existing models are capable of handling these tasks in the epidemiology domain. To the best of our knowledge, the BAND corpus is the largest corpus of well-annotated biomedical outbreak alert news with elaborately designed questions, making it a valuable resource for epidemiologists and NLP researchers alike.
['Nigel Collier', 'David Buckeridge', 'Anya Okhmatovskaia', 'Yannan Shen', 'Zaiqiao Meng', 'Meiru Zhang', 'Zihao Fu']
2023-05-23
null
null
null
null
['epidemiology', 'event-extraction', 'named-entity-recognition-ner']
['medical', 'natural-language-processing', 'natural-language-processing']
[ 1.84879616e-01 1.65180787e-01 -2.46814683e-01 -2.03737170e-01 -7.70247161e-01 -5.22354007e-01 6.20401204e-01 1.05004573e+00 -5.92127800e-01 9.71494079e-01 7.15250194e-01 -3.63394588e-01 -2.40366757e-01 -8.58004928e-01 -5.29883206e-01 -3.81150037e-01 -3.57644707e-01 4.86230940e-01 1.51872307e-01 -1.69187531e-01 5.88776842e-02 3.39368463e-01 -1.13302243e+00 5.66118419e-01 9.28546846e-01 7.59467304e-01 1.00696698e-01 6.34641051e-01 -2.73845851e-01 1.01088655e+00 -9.01059628e-01 -4.75803852e-01 -5.67558646e-01 -4.36336368e-01 -8.99017930e-01 -4.27566290e-01 -3.88328940e-01 -1.68939516e-01 -1.48922369e-01 7.53527701e-01 5.17055809e-01 -9.55827311e-02 5.65741479e-01 -1.07271409e+00 -6.22503579e-01 5.19897699e-01 -1.56578556e-01 8.94208610e-01 8.18610787e-01 -8.00548345e-02 9.33695138e-01 -4.28994209e-01 1.13887751e+00 8.25030625e-01 7.58189380e-01 3.73310477e-01 -6.27491415e-01 -3.47739667e-01 -9.21963230e-02 2.53065228e-01 -9.60651457e-01 -2.88844407e-01 2.97657758e-01 -6.51156962e-01 1.19253695e+00 4.95361447e-01 4.30647969e-01 1.57784998e+00 2.99995929e-01 5.53528488e-01 7.57516801e-01 2.16532610e-02 1.83463931e-01 1.40671879e-01 2.75878876e-01 4.74448591e-01 4.10613418e-01 -1.41868055e-01 -3.66056144e-01 -8.39653850e-01 2.64018238e-01 3.83043766e-01 -6.73873127e-01 6.29863203e-01 -1.50894272e+00 8.01442683e-01 1.76025391e-01 4.92785245e-01 -9.13492978e-01 -5.47290981e-01 7.07849443e-01 3.15277904e-01 9.29100871e-01 7.39741802e-01 -9.48499024e-01 -5.32692969e-02 -5.23875415e-01 2.23153040e-01 1.35268188e+00 5.52170694e-01 8.04055948e-03 -5.28439522e-01 -2.93646991e-01 6.94084644e-01 1.03215314e-01 6.69359028e-01 2.66768932e-01 -2.57349879e-01 5.36865950e-01 7.70645082e-01 1.31677300e-01 -1.48514223e+00 -7.71404922e-01 -3.82797360e-01 -1.01823664e+00 -9.95130122e-01 3.02880019e-01 -4.18291688e-01 -5.86954534e-01 1.61232543e+00 6.49928749e-01 3.51330340e-01 2.15580091e-01 4.85300153e-01 1.34842455e+00 1.05384994e+00 3.23454887e-01 -5.40108502e-01 2.11786723e+00 -2.80034572e-01 -1.25875461e+00 -1.78996786e-01 4.35451925e-01 -8.91430676e-01 3.82080883e-01 1.68770194e-01 -6.28291249e-01 7.62524828e-02 -4.11708772e-01 2.56831437e-01 -9.34804499e-01 -1.13724217e-01 3.74517262e-01 5.73319532e-02 -4.95241106e-01 4.48618941e-02 -7.05964208e-01 -9.05193508e-01 4.25820082e-01 -2.78059930e-01 -3.77886236e-01 -1.27917454e-01 -1.82041526e+00 1.19227552e+00 4.58161592e-01 -8.21863785e-02 -5.13307452e-01 -1.16943729e+00 -8.48725915e-01 -1.27019078e-01 6.39357746e-01 -8.51458967e-01 1.05240011e+00 2.77496427e-01 -7.16547847e-01 8.34082127e-01 -2.28932500e-01 -5.45316815e-01 4.12922204e-02 -1.63438648e-01 -9.91499126e-01 4.05846089e-01 3.79381388e-01 4.37260084e-02 2.09581599e-01 -6.65649056e-01 -6.67081833e-01 -2.62498319e-01 -1.70541838e-01 -3.84625912e-01 -3.20722789e-01 7.61112809e-01 -1.39047980e-01 -9.68730509e-01 -5.45152187e-01 -5.10358632e-01 -4.05433059e-01 -2.69014060e-01 -5.86707830e-01 -2.97166586e-01 4.82801437e-01 -1.09163511e+00 1.54439580e+00 -1.97029960e+00 -3.35666865e-01 -2.25347772e-01 4.94544029e-01 1.72402501e-01 8.36824179e-02 9.42205846e-01 5.25252894e-02 2.12132961e-01 -2.80899882e-01 2.64269084e-01 -3.51036578e-01 2.98401445e-01 -5.72348714e-01 3.28811705e-01 7.01074600e-01 9.43753123e-01 -1.15828741e+00 -5.65406144e-01 -2.85162479e-01 5.46750307e-01 -4.29546177e-01 4.91304457e-01 -4.63245809e-01 5.82625866e-01 -8.33021820e-01 4.54083204e-01 1.43722028e-01 -6.61068916e-01 -3.14148329e-02 -1.31773949e-01 -3.49675566e-02 6.86839104e-01 -6.84397936e-01 9.49135482e-01 -1.25000313e-01 4.42135632e-01 1.33049130e-01 -9.53927338e-01 5.48551738e-01 7.85385966e-01 7.81188250e-01 -3.77166033e-01 1.52263753e-02 -1.89581454e-01 -3.64962280e-01 -1.10550976e+00 1.42355680e-01 4.22655344e-02 -2.94652611e-01 6.72683537e-01 -2.30164051e-01 4.28655028e-01 3.38665158e-01 2.20001712e-01 1.55116224e+00 -6.53972149e-01 8.46028268e-01 -4.07075547e-02 2.14799464e-01 4.52592939e-01 6.82097316e-01 5.24920583e-01 -7.69575909e-02 1.68408424e-01 6.44323468e-01 -6.32780075e-01 -6.61259353e-01 -8.15750837e-01 -4.89782780e-01 8.70872438e-01 -5.64131856e-01 -5.50751984e-01 -5.40620744e-01 -5.28836727e-01 -2.65330791e-01 5.84713817e-01 -8.03535879e-01 1.49675220e-01 -5.62476516e-01 -1.42071021e+00 7.78800786e-01 7.59863257e-02 3.55011404e-01 -1.26811135e+00 -8.05804551e-01 5.22696316e-01 -9.43204463e-01 -1.34181440e+00 -3.48082721e-01 -1.15552329e-01 -3.47680718e-01 -1.69665349e+00 -6.84014082e-01 -5.98044991e-01 5.76399446e-01 -2.50253946e-01 1.23799646e+00 -1.80215746e-01 -5.72950780e-01 3.80835861e-01 -5.59798777e-01 -8.74136925e-01 -7.07629383e-01 3.65699157e-02 -1.43809929e-01 -2.67379612e-01 5.06508887e-01 -3.91641445e-02 -5.64221203e-01 2.33993530e-01 -1.26396167e+00 -2.72343338e-01 5.08613706e-01 5.71829319e-01 3.96358222e-01 2.91183218e-02 1.15359604e+00 -1.13159585e+00 9.72904086e-01 -1.48552275e+00 -1.90384984e-01 4.00042057e-01 -3.38054091e-01 -1.50665686e-01 6.03424430e-01 -3.13867360e-01 -1.14529634e+00 -4.68010902e-01 -4.73072290e-01 6.43899858e-01 -6.21421874e-01 1.14690876e+00 3.63830209e-01 6.48612559e-01 8.01864147e-01 8.44121501e-02 -1.45543993e-01 -6.31837845e-01 3.87031548e-02 9.77969646e-01 4.16581035e-01 -6.42379448e-02 2.09683016e-01 3.61554682e-01 -3.86506915e-01 -1.01440656e+00 -1.32582664e+00 -8.18829775e-01 1.29137263e-01 2.41740588e-02 1.14034116e+00 -7.04988301e-01 -7.08160639e-01 1.93076864e-01 -1.60429668e+00 2.75356412e-01 -1.02802902e-01 5.66676974e-01 6.28872961e-02 1.51969478e-01 -1.00735760e+00 -5.18701971e-01 -5.62214315e-01 -6.94475293e-01 9.20428276e-01 -2.16170415e-01 -5.26220679e-01 -1.06983900e+00 4.36871827e-01 4.10664529e-01 5.64954937e-01 8.45888138e-01 1.02145159e+00 -1.22196865e+00 -2.95963837e-03 -3.21190745e-01 -1.84632778e-01 -1.98923513e-01 2.78042734e-01 -1.84465960e-01 -7.55165279e-01 1.43707305e-01 3.81808996e-01 -1.57096654e-01 9.70562518e-01 2.16865510e-01 6.99246466e-01 -1.01999080e+00 -6.07975483e-01 -8.55949358e-04 7.98279226e-01 3.09236884e-01 3.26291800e-01 1.74455434e-01 2.04577282e-01 8.79538178e-01 2.92063981e-01 7.31161714e-01 6.41969323e-01 1.28006399e-01 7.40329549e-02 -1.12810127e-01 3.30116689e-01 -5.34432791e-02 -4.08049487e-03 9.45828676e-01 4.71787937e-02 -5.69573820e-01 -1.33558559e+00 6.00692213e-01 -1.64168274e+00 -9.85513508e-01 -1.78559452e-01 1.60971749e+00 1.25001431e+00 -1.50395095e-01 -1.50546744e-01 -2.04525068e-01 6.77416265e-01 2.45275244e-01 -3.25870723e-01 -6.58210833e-03 -1.18490510e-01 -7.69528225e-02 9.63701606e-02 -1.02401022e-02 -1.28592193e+00 3.05209696e-01 6.47341108e+00 5.46222508e-01 -5.98988235e-01 2.79887408e-01 6.07118785e-01 2.84598202e-01 -8.96132439e-02 -5.06439090e-01 -8.20881426e-01 5.98577321e-01 1.33972812e+00 -1.64681897e-01 5.64004034e-02 5.49314260e-01 3.41914386e-01 1.74685623e-02 -8.73732269e-01 5.11879563e-01 1.75978094e-01 -1.58930123e+00 -6.57801181e-02 1.19420150e-02 4.84571666e-01 2.86125630e-01 -3.55362236e-01 5.81812449e-02 2.91590601e-01 -7.08800435e-01 -8.79158638e-03 9.00680721e-01 4.99799967e-01 -3.80711079e-01 1.09944832e+00 6.03851795e-01 -9.77321625e-01 2.13289917e-01 -2.26800978e-01 2.06144661e-01 5.00849366e-01 1.11396050e+00 -1.21602094e+00 5.86131096e-01 8.55279803e-01 6.48821414e-01 -3.03067386e-01 9.55906749e-01 -1.96390614e-01 9.30010140e-01 -3.31140846e-01 -2.10833296e-01 5.02745323e-02 3.51773024e-01 6.90661371e-01 1.55310881e+00 1.40212506e-01 5.63005149e-01 1.47995919e-01 6.21292412e-01 -2.14046091e-01 2.68598258e-01 -7.73407996e-01 -5.26626945e-01 3.11752766e-01 1.05366158e+00 -6.97010696e-01 -4.64457870e-01 -3.69687259e-01 4.85262454e-01 2.09934056e-01 3.28860164e-01 -8.10221255e-01 -5.06291330e-01 5.11739075e-01 1.05804369e-01 -2.79129893e-02 1.20544732e-01 4.53409463e-01 -1.11903477e+00 -2.19944745e-01 -9.37319279e-01 1.00000131e+00 -4.49038386e-01 -1.80573475e+00 8.89220357e-01 7.35354722e-02 -7.75231004e-01 -3.34109306e-01 -2.93569088e-01 -2.99938619e-01 4.10814583e-01 -1.47887778e+00 -7.19470441e-01 1.43102938e-02 3.83030444e-01 4.11135197e-01 7.96523243e-02 9.80170131e-01 6.00763857e-01 -7.14349687e-01 9.92285237e-02 3.67525257e-02 3.37264270e-01 9.06758606e-01 -7.76171148e-01 3.84332687e-01 3.98706704e-01 -2.68822700e-01 7.98601687e-01 7.07953215e-01 -9.30671155e-01 -1.20300889e+00 -1.48053110e+00 1.60060608e+00 -9.17486072e-01 1.00917876e+00 -1.69688478e-01 -1.09425282e+00 5.71459830e-01 3.25664192e-01 -3.62891704e-01 1.21672130e+00 -1.07940264e-01 -2.62400001e-01 3.17449898e-01 -1.26776910e+00 3.94738287e-01 6.53450847e-01 -3.84418249e-01 -1.20151472e+00 8.96450281e-01 1.05540216e+00 -1.94548786e-01 -1.30205798e+00 4.97548699e-01 1.11812249e-01 -3.47205967e-01 1.06983173e+00 -1.16551578e+00 5.81140518e-01 -2.24287715e-02 3.06324095e-01 -1.26175439e+00 -3.91293652e-02 -3.98683786e-01 -1.64355189e-01 1.15451992e+00 6.65801406e-01 -9.89370346e-01 -2.49191299e-02 3.38602632e-01 8.63103569e-03 -9.20343578e-01 -7.38038599e-01 -3.40687066e-01 -5.56163132e-01 -5.30076444e-01 6.32303596e-01 1.43266118e+00 2.16138929e-01 4.26119089e-01 -2.81577975e-01 4.10172075e-01 1.88438535e-01 1.24951161e-01 1.46165550e-01 -1.36776567e+00 -4.35869135e-02 -1.99583456e-01 3.78759727e-02 -4.76660341e-01 -1.55763388e-01 -6.60401702e-01 1.31441131e-01 -1.89219332e+00 4.10453260e-01 -8.41085762e-02 -1.43773898e-01 4.81419355e-01 -4.18040752e-01 2.84848381e-02 -4.49469298e-01 2.22286478e-01 -8.13279450e-01 8.97342339e-02 9.70398486e-01 -1.14125140e-01 -4.26183045e-02 4.68402058e-02 -6.21016085e-01 8.96211147e-01 8.81194115e-01 -9.40014303e-01 4.58250158e-02 -2.26365060e-01 8.04590940e-01 1.81793958e-01 3.52350414e-01 -3.35729599e-01 4.44969773e-01 -3.26634556e-01 1.90273061e-01 -6.66024864e-01 -1.62648067e-01 -5.34518600e-01 2.50336319e-01 7.85822332e-01 -5.16881227e-01 1.30822852e-01 9.72886309e-02 9.65206027e-01 -3.93128216e-01 7.07954690e-02 2.69141942e-01 -1.34592727e-01 -2.99592435e-01 3.05597812e-01 -8.97782862e-01 6.39796317e-01 1.00621939e+00 4.90851134e-01 -9.35920894e-01 -2.15656772e-01 -5.50613225e-01 4.31027770e-01 -2.19498962e-01 5.16675234e-01 5.79054296e-01 -7.42753446e-01 -1.20597327e+00 -8.30906555e-02 4.70389932e-01 -9.09195542e-02 3.75630677e-01 1.09370184e+00 -4.86182749e-01 7.59137034e-01 2.80618817e-01 -7.76655674e-02 -7.38360703e-01 9.46827471e-01 -3.15840244e-01 -7.03581154e-01 -5.64989746e-01 5.20718336e-01 5.87718422e-03 -4.56754863e-01 2.90170640e-01 -6.11714959e-01 -6.69135332e-01 5.56482852e-01 9.72773433e-01 4.18951303e-01 1.75837986e-02 -4.31135088e-01 -6.82108521e-01 -1.36219822e-02 -1.11168884e-01 3.75857979e-01 1.60430574e+00 3.88909876e-02 -4.67322916e-01 2.77366519e-01 1.03615510e+00 -7.25923805e-03 -1.93563148e-01 -2.69426703e-01 4.06169146e-01 2.64410824e-01 -3.19522142e-01 -9.99205351e-01 -6.12311840e-01 5.24940848e-01 -1.49844021e-01 8.00596595e-01 9.22518432e-01 5.66118121e-01 1.17127502e+00 6.56499982e-01 1.28343692e-02 -7.21861124e-01 -9.00277048e-02 5.94714940e-01 1.17347574e+00 -1.16943395e+00 -1.88153848e-01 -3.94787967e-01 -6.34288073e-01 7.51762509e-01 5.84180951e-02 4.92875159e-01 9.52473342e-01 4.75383282e-01 4.23925892e-02 -6.24156952e-01 -9.79062319e-01 -1.01792388e-01 4.33682203e-01 3.38956237e-01 4.04625595e-01 -1.37093350e-01 -3.93187523e-01 1.03240633e+00 1.20737433e-01 4.43391781e-03 9.94227231e-02 1.00021422e+00 -4.52126741e-01 -6.03851736e-01 -4.94838208e-01 9.57849860e-01 -1.14046419e+00 -4.55193609e-01 -5.46208024e-01 5.02075195e-01 9.12320167e-02 1.18333673e+00 -1.27555579e-01 1.18738793e-01 5.15867531e-01 3.55540328e-02 -4.07562673e-01 -6.24220073e-01 -7.00751007e-01 -2.26562902e-01 4.62836832e-01 -4.06177908e-01 -5.68105578e-01 -2.92195737e-01 -1.17649019e+00 -2.03152314e-01 -1.49787426e-01 5.38489997e-01 5.17857790e-01 1.00146210e+00 7.42352903e-01 7.25484669e-01 2.73953617e-01 1.70577586e-01 -2.19605595e-01 -9.09320593e-01 -1.98283702e-01 3.59527588e-01 3.93806070e-01 -3.49569947e-01 -4.17108834e-01 2.76909411e-01]
[8.51162338256836, 9.226598739624023]
e15cf729-cc61-4d88-b6e9-0c8cb5d44a85
dc-mbr-distributional-cooling-for-minimum
2212.04205
null
https://arxiv.org/abs/2212.04205v2
https://arxiv.org/pdf/2212.04205v2.pdf
DC-MBR: Distributional Cooling for Minimum Bayesian Risk Decoding
Minimum Bayesian Risk Decoding (MBR) emerges as a promising decoding algorithm in Neural Machine Translation. However, MBR performs poorly with label smoothing, which is surprising as label smoothing provides decent improvement with beam search and improves generality in various tasks. In this work, we show that the issue arises from the un-consistency of label smoothing on the token-level and sequence-level distributions. We demonstrate that even though label smoothing only causes a slight change in the token-level, the sequence-level distribution is highly skewed. We coin the issue \emph{autoregressive over-smoothness}. To address this issue, we propose a simple and effective method, Distributional Cooling MBR (DC-MBR), which manipulates the entropy of output distributions by tuning down the Softmax temperature. We theoretically prove the equivalence between pre-tuning label smoothing factor and distributional cooling. Extensive experiments on NMT benchmarks validate that distributional cooling improves MBR in various settings.
['Yue Zhang', 'Jie zhou', 'Fandong Meng', 'Jin Xu', 'Jianhao Yan']
2022-12-08
null
null
null
null
['nmt']
['computer-code']
[ 5.95108926e-01 1.87954307e-01 -4.31288034e-01 -6.32345438e-01 -1.27002203e+00 -5.65339983e-01 5.00642061e-01 -1.30870447e-01 -5.26282728e-01 8.65555108e-01 2.20487610e-01 -7.46776164e-01 2.93441325e-01 -2.50726551e-01 -8.59602988e-01 -9.10455704e-01 2.29249731e-01 3.79817605e-01 -9.62094963e-02 2.31514759e-02 2.04441935e-01 3.73390429e-02 -9.83061016e-01 2.76207685e-01 1.16318703e+00 4.34003443e-01 2.13830322e-01 5.67601681e-01 8.48590862e-03 3.72755826e-01 -6.24281883e-01 -6.25860691e-01 2.10534051e-01 -7.78846323e-01 -6.49823129e-01 -4.88344640e-01 3.51948172e-01 -5.73651828e-02 7.38033354e-02 1.43771648e+00 4.85871792e-01 5.90858310e-02 8.86221170e-01 -9.91261125e-01 -7.10653245e-01 1.18614483e+00 -8.96417379e-01 1.82806700e-01 -1.82712853e-01 -8.11412781e-02 1.16548789e+00 -6.36253059e-01 3.98545891e-01 1.44894373e+00 5.83184481e-01 9.05655384e-01 -1.41373765e+00 -7.58352697e-01 2.20039412e-01 -7.50743821e-02 -1.18684518e+00 -3.85353327e-01 5.24012685e-01 -3.21290314e-01 9.12106454e-01 3.90095025e-01 1.19653307e-01 1.08252895e+00 3.05865526e-01 7.80938387e-01 1.36934888e+00 -6.92485511e-01 3.46360743e-01 6.94398582e-02 2.59357452e-01 5.74364424e-01 5.08963540e-02 1.61492214e-01 -6.47980750e-01 -2.60810643e-01 4.19790655e-01 -5.79377532e-01 -1.89785272e-01 8.59651491e-02 -1.09767735e+00 9.11986232e-01 -1.23708703e-01 2.07577065e-01 1.00948643e-02 5.93117893e-01 2.79261172e-01 3.90727252e-01 7.91528583e-01 4.41998661e-01 -7.50595272e-01 -5.60331702e-01 -1.04416966e+00 1.32515316e-03 6.27173722e-01 8.97617221e-01 5.47722578e-01 1.19331166e-01 -4.81104672e-01 9.76574719e-01 4.07448024e-01 7.15843916e-01 4.64883089e-01 -9.86458242e-01 6.21749103e-01 -1.21865653e-01 -5.37943356e-02 -3.43625993e-01 -2.12896913e-01 -5.01165867e-01 -7.65612781e-01 -7.12173432e-02 7.17859149e-01 -4.08543140e-01 -1.17363167e+00 2.42514539e+00 1.53143227e-01 9.15435776e-02 -1.13882475e-01 9.32085514e-01 2.49400049e-01 8.01280200e-01 3.29921454e-01 -5.48514366e-01 1.19863987e+00 -9.29718971e-01 -1.02244627e+00 -2.14703962e-01 9.05145884e-01 -7.48117805e-01 1.41385174e+00 2.99004912e-01 -1.21282530e+00 -8.78761336e-02 -8.61960411e-01 -1.13124579e-01 -1.30808344e-02 6.27798811e-02 4.50330436e-01 1.03320038e+00 -8.61348271e-01 7.22615480e-01 -9.53092933e-01 -2.77617257e-02 2.02235028e-01 3.94887000e-01 -5.13591096e-02 1.93845093e-01 -1.28824329e+00 1.03776240e+00 2.73850799e-01 3.14130001e-02 -6.05578601e-01 -7.54946232e-01 -6.54997945e-01 1.01359099e-01 1.88069493e-01 -4.43951100e-01 1.41633010e+00 -1.04050767e+00 -1.95354962e+00 7.64760613e-01 -6.23633504e-01 -4.68475193e-01 4.87404972e-01 -2.79549450e-01 -1.45361107e-02 -5.65812528e-01 -4.63308245e-01 6.06234729e-01 1.02085197e+00 -1.06000042e+00 -4.93896008e-01 -3.57238770e-01 -4.04102117e-01 1.88538045e-01 -2.34261617e-01 2.30776921e-01 -2.17915177e-01 -8.46087575e-01 4.18784879e-02 -1.07902217e+00 -1.21986963e-01 -9.04168487e-01 -3.01232815e-01 -4.17799801e-01 1.90040022e-01 -7.48056412e-01 1.39873862e+00 -2.10969567e+00 2.23376676e-01 1.89920843e-01 -1.24906838e-01 1.51649237e-01 -1.67027891e-01 -1.91087276e-02 -6.51884452e-02 5.07476211e-01 -4.93146926e-01 -4.39077467e-01 1.76437378e-01 2.58812577e-01 -5.34091353e-01 5.97036242e-01 2.57630825e-01 1.02098441e+00 -8.14531147e-01 -3.69904310e-01 -2.73469269e-01 4.38322961e-01 -7.34898984e-01 -5.85332606e-03 -4.12512392e-01 4.80541587e-01 -7.32985511e-02 3.91152382e-01 8.35351765e-01 1.03951339e-02 3.03373516e-01 1.10782281e-01 7.82904550e-02 7.27721930e-01 -6.63640499e-01 1.77948689e+00 -3.23705077e-01 6.08290136e-01 -8.68612751e-02 -9.38100100e-01 7.43586719e-01 2.01795280e-01 -5.59946634e-02 -5.88198900e-01 1.21387564e-01 3.10213566e-01 1.85781851e-01 -2.73147970e-01 6.83767557e-01 -5.93418300e-01 -1.83047071e-01 6.52398825e-01 -1.16434716e-01 -3.97221558e-02 -3.86437804e-01 -9.37117934e-02 9.45903659e-01 2.99113899e-01 6.04035053e-03 -4.74735260e-01 -2.32787337e-02 -2.20677108e-01 8.99302065e-01 9.01504934e-01 -1.89523458e-01 5.89912295e-01 7.81078756e-01 8.17460045e-02 -1.15535629e+00 -1.05332994e+00 -2.61841774e-01 1.67098784e+00 6.90380707e-02 -4.91543084e-01 -1.23147392e+00 -8.33464146e-01 -1.52707785e-01 1.09308219e+00 -5.27621865e-01 -2.44335249e-01 -9.46897209e-01 -1.29588330e+00 7.40073740e-01 3.32460225e-01 -1.09382004e-01 -7.15354383e-01 -2.13257983e-01 -3.31475139e-02 -3.24252278e-01 -7.37993062e-01 -6.87303841e-01 5.68417490e-01 -9.20219183e-01 -1.57230854e-01 -6.63936734e-01 -5.83202302e-01 5.12128830e-01 -2.40989432e-01 1.14949882e+00 -1.56660736e-01 1.82840973e-01 -4.38656926e-01 -2.10624442e-01 -2.94946939e-01 -7.97061265e-01 5.44065952e-01 1.19491771e-01 -4.20598954e-01 4.04401809e-01 -5.70094645e-01 -2.28396058e-01 1.94176629e-01 -7.43164361e-01 5.12024164e-02 4.41409916e-01 9.55811083e-01 4.88256007e-01 -2.47763529e-01 7.86942720e-01 -1.17627811e+00 8.63162458e-01 -3.63900363e-01 -5.85790873e-01 5.32980800e-01 -8.04271817e-01 5.57158709e-01 4.50758010e-01 -5.78828931e-01 -9.86868083e-01 -9.89041775e-02 -3.51564795e-01 -7.89297223e-02 1.12853535e-01 3.64607304e-01 -1.13845140e-01 4.37710881e-01 4.39416856e-01 -1.61747560e-02 -1.77126810e-01 -5.10857522e-01 6.14284039e-01 8.96429300e-01 4.73429263e-01 -8.98704946e-01 5.51157117e-01 1.34101719e-01 -1.97143823e-01 -4.46921170e-01 -8.69677544e-01 -2.25079376e-02 -2.93194950e-01 1.40767202e-01 8.02764237e-01 -6.94167435e-01 -6.09037399e-01 3.58113110e-01 -1.18620682e+00 -7.09033668e-01 -7.34341741e-02 5.27583122e-01 -6.10661447e-01 2.23376080e-01 -7.92331994e-01 -9.03886437e-01 -4.20057982e-01 -1.26351297e+00 1.00180650e+00 -8.27682018e-03 -3.92152101e-01 -8.03723991e-01 1.13211602e-01 2.75603086e-01 4.33720708e-01 -1.91683233e-01 1.33772218e+00 -6.70612752e-01 -2.28459314e-01 2.96667069e-01 -1.05806038e-01 3.84160280e-01 -8.90623871e-03 -1.07747696e-01 -1.07106400e+00 -3.90399426e-01 8.38928223e-02 -1.18525669e-01 1.06471801e+00 5.64148664e-01 1.15364027e+00 -2.27675050e-01 -6.54042661e-02 7.04875827e-01 1.17016816e+00 7.13685825e-02 5.23186743e-01 1.34082034e-01 7.46370912e-01 4.03369069e-01 4.33138490e-01 1.21773727e-01 8.80944952e-02 8.37500274e-01 2.09002271e-02 3.40164937e-02 -6.01501279e-02 -3.56704950e-01 7.12175965e-01 1.16342556e+00 2.04601008e-02 -3.02442521e-01 -8.46024990e-01 2.26139784e-01 -1.88670456e+00 -6.98868513e-01 -8.04868862e-02 2.27205801e+00 1.49664164e+00 1.55396909e-01 1.47824019e-01 -2.09138125e-01 8.84547591e-01 -1.90205518e-02 -5.29287100e-01 -9.06892419e-01 -9.23377052e-02 3.15880239e-01 9.05285418e-01 9.71611738e-01 -6.67626977e-01 1.29924393e+00 6.98691940e+00 1.33289552e+00 -1.03314960e+00 5.62366843e-01 7.56831229e-01 -2.43057519e-01 -7.64543355e-01 1.82859655e-02 -1.03975832e+00 8.24257970e-01 1.18235171e+00 -3.08697182e-03 6.55369282e-01 5.81225812e-01 2.13154957e-01 7.01571628e-02 -1.21058667e+00 7.80789793e-01 -8.62546936e-02 -9.97769713e-01 -2.16717750e-01 7.27514625e-02 7.35560775e-01 2.54717439e-01 2.93126911e-01 3.54876846e-01 6.45919383e-01 -1.22244334e+00 8.77728939e-01 1.22627877e-01 9.57040071e-01 -9.86966074e-01 5.43515325e-01 5.23343742e-01 -4.76204753e-01 1.96249858e-01 -5.18558919e-01 1.09443910e-01 3.21460187e-01 7.56113768e-01 -7.55028665e-01 1.49190426e-01 3.97538006e-01 2.81208575e-01 -3.42445225e-01 4.66267526e-01 -5.09218454e-01 1.25634158e+00 -3.54568243e-01 3.12151536e-02 9.87844095e-02 -3.88315648e-01 4.91196901e-01 1.61609197e+00 2.52974361e-01 -7.97736868e-02 -2.63028294e-01 8.74249935e-01 -3.23945910e-01 6.86635152e-02 -1.77254885e-01 2.66303476e-02 5.17375827e-01 9.32298481e-01 -5.69383502e-01 -2.20624819e-01 -7.40675405e-02 8.81110907e-01 5.74353456e-01 3.80719602e-01 -1.14020908e+00 -7.24747777e-02 6.79476917e-01 -1.08524971e-01 2.04016551e-01 -2.31283918e-01 -8.01471889e-01 -1.22460973e+00 -1.07672242e-02 -1.05730879e+00 1.58347651e-01 -3.86290550e-01 -1.09425390e+00 3.63887340e-01 -1.55155137e-01 -5.88179469e-01 -2.33302400e-01 -2.84337372e-01 -2.07356527e-01 1.03126979e+00 -1.59462059e+00 -7.78205454e-01 5.07804334e-01 5.41207306e-02 5.38633943e-01 9.18676853e-02 6.81768537e-01 2.26023912e-01 -6.85662806e-01 1.30117583e+00 3.86151850e-01 -1.73399284e-01 8.42376709e-01 -1.43693686e+00 6.69296920e-01 7.90557325e-01 7.60932565e-02 6.91595316e-01 9.25808072e-01 -6.35381758e-01 -1.35110712e+00 -9.53658581e-01 1.08061457e+00 -4.65669155e-01 5.51899493e-01 -6.73280537e-01 -1.10652769e+00 6.02491856e-01 2.84041524e-01 -3.95166785e-01 6.07617080e-01 3.23126376e-01 -5.09142399e-01 1.99315697e-01 -9.89329696e-01 7.04913914e-01 1.05017245e+00 -5.95548034e-01 -6.33385956e-01 2.79603928e-01 1.14157343e+00 -4.79841828e-01 -7.26144612e-01 5.80789566e-01 4.44038868e-01 -5.36140978e-01 5.71006358e-01 -7.94564009e-01 4.18398201e-01 -8.59770924e-02 -3.66865188e-01 -1.48705053e+00 -1.74366757e-01 -1.11716127e+00 -6.01521879e-02 1.42810559e+00 8.86088908e-01 -6.55324578e-01 6.48711503e-01 7.27080345e-01 -1.45998716e-01 -8.00623953e-01 -1.07682109e+00 -9.46027756e-01 7.39506066e-01 -5.55636227e-01 5.70775628e-01 9.47200835e-01 2.67336726e-01 3.61727476e-01 -6.25914574e-01 -1.66861489e-02 4.73192990e-01 -1.83321968e-01 3.11331242e-01 -7.16937840e-01 -5.31806529e-01 -4.94339526e-01 2.10428923e-01 -1.24291003e+00 5.57991982e-01 -1.16575170e+00 4.98399258e-01 -9.78514731e-01 3.79538268e-01 -5.94182730e-01 -4.36267108e-01 4.95859593e-01 -5.74783027e-01 9.34268162e-02 1.15633324e-01 1.61899462e-01 -4.16522473e-01 4.29967493e-01 9.80960906e-01 1.32670626e-01 -2.29675055e-01 -1.16549313e-01 -5.97324967e-01 5.61893225e-01 9.73785877e-01 -8.86448205e-01 -1.69266865e-01 -7.09326923e-01 6.74202621e-01 -1.72197402e-01 -2.26016089e-01 -3.45756859e-01 -1.62055284e-01 -2.09155917e-01 -1.96246564e-01 -4.86304998e-01 1.78353325e-01 -4.06181931e-01 1.07740350e-02 3.58109057e-01 -9.79617417e-01 1.55059278e-01 3.24940495e-02 4.82265830e-01 1.73984140e-01 -4.37686950e-01 1.04427183e+00 1.97505221e-01 -1.59945432e-02 -1.36722803e-01 -5.08911192e-01 3.93937171e-01 5.18416762e-01 1.63940698e-01 -4.56715703e-01 -1.51077822e-01 -4.10826296e-01 1.04899414e-01 5.01565993e-01 3.63898635e-01 1.28423080e-01 -1.23509574e+00 -9.02182817e-01 2.49780208e-01 -2.31416777e-01 -7.96934813e-02 -1.72418788e-01 9.48711038e-01 -9.95931998e-02 2.76926070e-01 4.82643366e-01 -6.43256187e-01 -1.26663411e+00 2.58096516e-01 2.16674581e-01 -2.46662393e-01 -4.67322707e-01 1.10228670e+00 1.45415768e-01 -5.19971371e-01 4.78115529e-01 -4.77378219e-01 3.18588823e-01 -1.34660915e-01 4.00121212e-01 2.36778215e-01 1.85231715e-01 -3.38672280e-01 -2.59276003e-01 3.68454099e-01 -3.31245333e-01 -3.93787593e-01 1.15682793e+00 -2.43362188e-01 -3.18402529e-01 5.19905508e-01 1.09611082e+00 6.69678673e-02 -1.33420300e+00 -2.13690251e-01 3.19866925e-01 -2.94005126e-01 1.04693264e-01 -8.63329053e-01 -7.44723320e-01 9.28031147e-01 4.44033235e-01 1.45854071e-01 8.47756922e-01 -1.55279960e-03 9.61941302e-01 2.25359410e-01 1.50356188e-01 -1.22766304e+00 -4.12055194e-01 8.01511228e-01 3.84402007e-01 -1.02847791e+00 -2.12571383e-01 -2.49435231e-01 -6.77607656e-01 6.01560593e-01 2.66580254e-01 1.23817988e-01 4.81580645e-01 5.48112631e-01 1.47028103e-01 3.08300585e-01 -9.09792244e-01 2.15056196e-01 1.39730260e-01 3.37849855e-01 6.98579311e-01 4.29098785e-01 -6.68272376e-01 6.09759986e-01 -5.42459190e-01 -2.59845972e-01 2.87712634e-01 5.84272802e-01 -5.54539204e-01 -1.31278300e+00 -2.44616672e-01 2.23980606e-01 -9.57786143e-01 -6.04423702e-01 -2.27428868e-01 3.03483337e-01 -2.40730252e-02 9.03473198e-01 3.10181398e-02 -3.35285515e-01 -2.14683548e-01 3.75892401e-01 7.08398163e-01 -5.07727802e-01 -6.14787757e-01 2.09905609e-01 1.12837054e-01 -2.18070611e-01 -1.55905500e-01 -6.10305548e-01 -1.26715767e+00 -3.37962896e-01 -6.82577789e-01 3.38208884e-01 1.05487978e+00 1.10282683e+00 2.56633848e-01 3.33360493e-01 4.12084728e-01 -4.06720698e-01 -1.09524715e+00 -1.15055251e+00 -4.33709651e-01 3.59635830e-01 3.61790210e-01 -2.88452655e-01 -5.93317688e-01 1.28817707e-01]
[11.536667823791504, 9.800569534301758]
81cb83fa-847f-4616-b2ef-71058cbdf10a
accurate-learning-of-graph-representations-1
2102.11533
null
https://arxiv.org/abs/2102.11533v4
https://arxiv.org/pdf/2102.11533v4.pdf
Accurate Learning of Graph Representations with Graph Multiset Pooling
Graph neural networks have been widely used on modeling graph data, achieving impressive results on node classification and link prediction tasks. Yet, obtaining an accurate representation for a graph further requires a pooling function that maps a set of node representations into a compact form. A simple sum or average over all node representations considers all node features equally without consideration of their task relevance, and any structural dependencies among them. Recently proposed hierarchical graph pooling methods, on the other hand, may yield the same representation for two different graphs that are distinguished by the Weisfeiler-Lehman test, as they suboptimally preserve information from the node features. To tackle these limitations of existing graph pooling methods, we first formulate the graph pooling problem as a multiset encoding problem with auxiliary information about the graph structure, and propose a Graph Multiset Transformer (GMT) which is a multi-head attention based global pooling layer that captures the interaction between nodes according to their structural dependencies. We show that GMT satisfies both injectiveness and permutation invariance, such that it is at most as powerful as the Weisfeiler-Lehman graph isomorphism test. Moreover, our methods can be easily extended to the previous node clustering approaches for hierarchical graph pooling. Our experimental results show that GMT significantly outperforms state-of-the-art graph pooling methods on graph classification benchmarks with high memory and time efficiency, and obtains even larger performance gain on graph reconstruction and generation tasks.
['Sung Ju Hwang', 'Minki Kang', 'Jinheon Baek']
2021-02-23
accurate-learning-of-graph-representations
https://openreview.net/forum?id=JHcqXGaqiGn
https://openreview.net/pdf?id=JHcqXGaqiGn
iclr-2021-1
['graph-reconstruction']
['graphs']
[ 1.61746770e-01 4.61204916e-01 -3.76901358e-01 -1.89791739e-01 -3.80858600e-01 -6.11623704e-01 4.93761182e-01 5.24625123e-01 -9.39195454e-02 4.89564866e-01 1.03031568e-01 -3.15254539e-01 -2.16026515e-01 -1.38990688e+00 -7.13165581e-01 -7.92780757e-01 -4.17086840e-01 3.64709258e-01 3.92510593e-01 -1.10143736e-01 -2.67158300e-01 4.91052747e-01 -1.24952447e+00 2.00029969e-01 6.84331954e-01 8.73773217e-01 1.79311827e-01 5.87651193e-01 -5.53638563e-02 6.86986923e-01 -4.51923132e-01 -7.53339767e-01 1.59026176e-01 -1.19438544e-01 -9.34164941e-01 -8.66588131e-02 5.43265700e-01 -6.91406131e-02 -9.10023391e-01 1.28414178e+00 2.54478186e-01 9.81104001e-02 5.62426448e-01 -1.45012069e+00 -1.19285595e+00 1.06287742e+00 -4.72241640e-01 1.20980316e-03 2.73090720e-01 -1.79617524e-01 1.76684356e+00 -6.82910860e-01 6.53195500e-01 1.27483892e+00 6.98020995e-01 1.98768660e-01 -1.31093061e+00 -4.63515073e-01 4.87032712e-01 2.55231410e-01 -1.53599691e+00 1.01800628e-01 7.81687498e-01 -1.53229713e-01 1.06544697e+00 4.59527284e-01 6.47400439e-01 7.76039302e-01 2.71461368e-01 6.19758546e-01 5.48876762e-01 -2.00806912e-02 -1.56707406e-01 -2.26357728e-01 5.16475558e-01 1.21813619e+00 6.05627537e-01 -4.15756971e-01 -1.71516955e-01 -1.28223255e-01 8.09377193e-01 8.00662413e-02 -5.27767241e-01 -3.96397322e-01 -1.11046100e+00 8.37244034e-01 1.21314096e+00 4.49528456e-01 -2.60412306e-01 3.41181606e-01 3.84892404e-01 4.32120740e-01 4.21097726e-01 3.70124817e-01 -1.48816869e-01 7.36799538e-01 -4.68622983e-01 1.39654368e-01 6.80192232e-01 1.12389815e+00 9.13603246e-01 -1.63262531e-01 -5.39068818e-01 5.45260251e-01 2.70983398e-01 1.51603734e-02 2.60328025e-01 -2.70341516e-01 5.80819845e-01 1.14126837e+00 -4.53012437e-01 -1.57830107e+00 -6.24430120e-01 -5.08849144e-01 -1.25163662e+00 -3.58540684e-01 1.43016890e-01 3.39290112e-01 -9.53402936e-01 2.05095720e+00 3.10909469e-02 2.76272655e-01 -9.94242057e-02 6.10408604e-01 1.46346772e+00 5.33010423e-01 3.22133787e-02 4.71181385e-02 1.35248005e+00 -1.02385068e+00 -5.53644836e-01 -3.18233669e-01 7.21742213e-01 -1.11602224e-01 9.67768073e-01 -1.75044075e-01 -9.96965766e-01 -4.34772551e-01 -1.10843349e+00 -1.85108289e-01 -6.28992975e-01 -3.06774914e-01 1.04025459e+00 6.44567370e-01 -1.37288713e+00 8.62173736e-01 -6.83441937e-01 -5.15055418e-01 5.05758405e-01 5.01321197e-01 -8.08301389e-01 -2.87973464e-01 -1.20463157e+00 6.51432872e-01 5.54787636e-01 2.14976951e-01 -5.11452854e-01 -5.77493310e-01 -1.30965233e+00 7.39042640e-01 4.16019559e-01 -6.68860674e-01 5.40648818e-01 -5.11597931e-01 -1.00747883e+00 8.04278791e-01 -1.25776649e-01 -4.48151857e-01 -1.03920780e-01 5.23341238e-01 -3.02325904e-01 2.55632579e-01 -7.25020394e-02 5.35120845e-01 4.63008583e-01 -9.41026568e-01 -1.10950083e-01 -4.00585771e-01 3.40238988e-01 1.24974154e-01 -6.11765087e-01 -1.28489241e-01 -7.44555533e-01 -6.20091498e-01 3.62631708e-01 -7.74400175e-01 -2.00719297e-01 -2.56180882e-01 -7.95871198e-01 -3.81577939e-01 5.32782257e-01 -5.06710052e-01 1.39133620e+00 -1.90639389e+00 4.07595098e-01 4.81057018e-01 7.87546992e-01 1.43152028e-01 -5.82544684e-01 5.72374463e-01 -2.01734826e-01 5.58074236e-01 -2.64121771e-01 -2.19237000e-01 1.12380207e-01 3.24935794e-01 -1.51578784e-01 4.47725564e-01 4.02260512e-01 1.47487438e+00 -8.60248029e-01 -4.72975284e-01 -1.60990264e-02 4.79558438e-01 -6.51073813e-01 9.72324833e-02 9.56060067e-02 -7.23100547e-03 -3.37767124e-01 4.87542629e-01 6.70218647e-01 -8.22760284e-01 6.30073786e-01 -3.12340409e-01 5.86271584e-01 3.13608170e-01 -9.34922636e-01 1.46266937e+00 -1.15353778e-01 3.38993162e-01 -4.95149521e-03 -1.20804930e+00 8.87034357e-01 9.64258760e-02 4.13276613e-01 -3.59449446e-01 2.58709472e-02 -1.20688096e-01 1.54294625e-01 -5.59350178e-02 4.11423266e-01 8.81992206e-02 -2.45710686e-01 2.50265181e-01 4.55934376e-01 1.24881864e-01 3.84932041e-01 6.46452427e-01 1.58892858e+00 -3.46124589e-01 4.18895692e-01 -4.25306290e-01 4.35415685e-01 -5.56675494e-01 4.89095241e-01 8.59730661e-01 -3.66193894e-03 7.05207765e-01 1.05321980e+00 -4.26935583e-01 -4.79883730e-01 -1.08488822e+00 2.29524836e-01 1.02261341e+00 2.98921764e-01 -8.37426782e-01 -5.62723637e-01 -9.82502103e-01 1.07581966e-01 1.09714188e-01 -8.77632737e-01 -5.26185811e-01 -6.06280446e-01 -9.11247432e-01 5.41707516e-01 5.80024481e-01 3.16581726e-01 -1.03663635e+00 2.79505197e-02 1.57732621e-01 -3.13255675e-02 -1.25605226e+00 -6.08376324e-01 1.29853770e-01 -5.87267816e-01 -1.28746760e+00 -5.10701299e-01 -1.11007011e+00 8.98396075e-01 4.65812683e-01 1.33017373e+00 6.72321796e-01 -9.25691649e-02 9.71392393e-02 -2.56497175e-01 1.65639341e-01 -1.82053655e-01 3.62813175e-01 -2.91227072e-01 7.59744719e-02 6.01749541e-03 -7.96708226e-01 -2.81332016e-01 2.14318544e-01 -8.83020759e-01 1.73990540e-02 5.48943341e-01 9.26740766e-01 6.14162445e-01 3.52300465e-01 5.39920390e-01 -1.14622605e+00 7.11780608e-01 -4.37874138e-01 -4.25807983e-01 5.39394200e-01 -3.89373213e-01 2.56958544e-01 8.25435221e-01 -1.38178870e-01 -5.68979084e-01 -2.13824764e-01 1.69970635e-02 -3.79375458e-01 2.53267705e-01 9.57117915e-01 -5.78659475e-01 -3.04471076e-01 2.55264312e-01 5.05063869e-02 -1.05957270e-01 -2.44666755e-01 4.28581655e-01 1.64879672e-02 4.57180113e-01 -4.13201034e-01 9.07111347e-01 3.58042300e-01 4.10728335e-01 -6.60680890e-01 -6.94227397e-01 -2.54133314e-01 -6.70402765e-01 8.52344483e-02 7.61005044e-01 -6.21531904e-01 -7.36831784e-01 3.25713575e-01 -1.15509629e+00 -1.93162531e-01 9.41376239e-02 5.54486327e-02 -1.49730802e-01 7.18368530e-01 -9.17284667e-01 -3.34698081e-01 -3.14535767e-01 -1.19305694e+00 9.23548043e-01 -6.72176406e-02 6.60062507e-02 -1.11521244e+00 -3.13616186e-01 -8.14862847e-02 2.09997267e-01 3.68299961e-01 1.44519889e+00 -7.76334763e-01 -8.72486353e-01 -4.74625856e-01 -6.72576249e-01 5.17710224e-02 -2.87995581e-02 -1.29998282e-01 -5.66525221e-01 -6.26962066e-01 -6.51664913e-01 -1.00504085e-01 1.34827960e+00 2.38028407e-01 1.33228672e+00 -5.11617124e-01 -5.56422353e-01 6.59931481e-01 1.46646297e+00 -3.51400524e-01 5.91700494e-01 -2.60425925e-01 1.27192378e+00 5.51095903e-01 -2.36198157e-01 -1.15149148e-01 6.29284263e-01 5.70539057e-01 6.89951718e-01 -8.47839117e-02 -3.13553810e-01 -5.38139164e-01 2.41831452e-01 8.31970453e-01 -2.78766342e-02 -6.46796167e-01 -5.46411157e-01 4.85866725e-01 -1.99062407e+00 -8.16663027e-01 -2.69110233e-01 2.14450121e+00 3.77018720e-01 1.15422800e-01 1.19281046e-01 5.11086248e-02 8.80942523e-01 5.74447334e-01 -2.71000117e-01 -2.22344518e-01 -2.87597597e-01 4.89557981e-01 6.97811007e-01 4.60943967e-01 -1.08580101e+00 1.00370860e+00 6.14557600e+00 6.34500802e-01 -7.50012577e-01 3.29230502e-02 4.67830747e-01 3.65504593e-01 -6.13296509e-01 -8.95643681e-02 -3.83699149e-01 1.89033926e-01 6.25735343e-01 -3.00686270e-01 5.79374433e-01 5.78070343e-01 -5.84086835e-01 4.28166479e-01 -1.14939225e+00 7.27390289e-01 1.83582842e-01 -1.34375715e+00 3.53010505e-01 1.58362389e-01 6.59556150e-01 9.69880670e-02 -1.70522496e-01 4.12758887e-01 4.81560349e-01 -1.35458338e+00 3.48580271e-01 2.14921162e-01 4.48419482e-01 -6.93121970e-01 7.83283412e-01 1.45861320e-02 -1.83274829e+00 -7.79804513e-02 -5.76362252e-01 -4.15470563e-02 -7.14700595e-02 5.29933035e-01 -5.57652712e-01 1.10739625e+00 5.10220826e-01 7.68684030e-01 -8.55742872e-01 8.92781079e-01 -4.16278929e-01 2.94198722e-01 -2.35928461e-01 -1.24722943e-01 2.76837379e-01 -1.65176690e-01 5.04713774e-01 1.00106359e+00 3.20552945e-01 -1.60952732e-01 3.67142797e-01 1.03642011e+00 -5.87950110e-01 1.67099044e-01 -7.93740213e-01 -3.36828321e-01 4.24164563e-01 1.50612032e+00 -1.03290617e+00 -1.70855969e-01 -5.49067318e-01 8.39718759e-01 9.27629471e-01 4.50415134e-01 -6.74639106e-01 -5.79201996e-01 7.06913471e-01 1.25546958e-02 5.26407301e-01 -1.59160137e-01 -3.37589979e-02 -1.29702866e+00 1.82429388e-01 -3.95605832e-01 7.59466529e-01 -5.14062405e-01 -1.42696941e+00 8.48784268e-01 -8.86748508e-02 -6.82843268e-01 2.12476283e-01 -8.95195723e-01 -8.46160948e-01 6.98975921e-01 -1.35460258e+00 -1.56643724e+00 -2.95630127e-01 5.43570161e-01 -1.90309718e-01 6.60546422e-02 9.24327075e-01 2.46690571e-01 -7.18582988e-01 8.61218393e-01 -4.93951529e-01 3.95934641e-01 2.16516301e-01 -1.41341460e+00 6.98805690e-01 9.77518916e-01 4.06440735e-01 6.92232549e-01 2.24070787e-01 -6.44510150e-01 -1.62411249e+00 -1.35642695e+00 1.00180638e+00 -2.38731652e-01 7.54267871e-01 -7.41123080e-01 -1.25993598e+00 1.00596631e+00 5.28929867e-02 4.20061678e-01 3.90601814e-01 3.73958260e-01 -7.61401713e-01 -5.88860661e-02 -9.26910698e-01 6.68148577e-01 1.46349967e+00 -5.98200440e-01 -1.70929119e-01 3.73123020e-01 1.04388547e+00 -2.42065325e-01 -1.19851041e+00 5.30604601e-01 3.62017214e-01 -8.70468199e-01 1.00697744e+00 -8.45231414e-01 3.43970031e-01 -1.67740807e-01 -1.79790482e-01 -1.27455616e+00 -8.21753681e-01 -6.53875887e-01 -4.43582460e-02 1.12267601e+00 5.00607550e-01 -9.07413602e-01 7.91708648e-01 3.15686762e-01 -1.31182998e-01 -8.08173954e-01 -9.36449289e-01 -7.45931506e-01 1.89171657e-02 -2.36435011e-01 1.02330887e+00 1.00521088e+00 2.59045541e-01 5.95091343e-01 -2.47644067e-01 5.15713394e-01 5.50480545e-01 2.74970680e-01 6.29014373e-01 -1.48400664e+00 -3.38500679e-01 -8.83581638e-01 -8.94604921e-01 -8.18522513e-01 6.40494525e-01 -1.70266974e+00 -2.27275878e-01 -1.93664896e+00 4.62138683e-01 -9.82433259e-02 -5.88234723e-01 8.07081580e-01 -3.59296441e-01 3.72451633e-01 2.30762988e-01 -8.39299187e-02 -6.46443963e-01 4.93013382e-01 1.28568065e+00 -4.34649259e-01 -4.48185652e-02 -2.57010251e-01 -1.01879561e+00 3.98638159e-01 5.36475539e-01 -1.69369817e-01 -3.66871774e-01 -4.78424191e-01 3.08945596e-01 -3.11869811e-02 6.29534662e-01 -7.17171550e-01 2.06330970e-01 2.22686574e-01 8.05691928e-02 -3.02765638e-01 1.43092155e-01 -5.72743893e-01 1.26620576e-01 4.32576627e-01 -3.50498885e-01 1.10915951e-01 -6.13678358e-02 8.01939130e-01 -1.73729464e-01 5.63859344e-02 5.18572330e-01 -1.64069176e-01 -5.19651055e-01 9.37146366e-01 7.46745020e-02 -1.97946742e-01 7.48539090e-01 -1.14114664e-01 -6.80786312e-01 -3.61636668e-01 -6.93548858e-01 3.34167033e-01 2.96845287e-01 5.24799705e-01 5.55836380e-01 -1.62620425e+00 -7.17573702e-01 4.28182751e-01 3.23139846e-01 -2.69116610e-02 3.20303321e-01 7.83030570e-01 -5.53266406e-02 3.02121013e-01 -1.64611772e-01 -2.83304870e-01 -1.21573269e+00 7.47828186e-01 2.69962221e-01 -7.67098308e-01 -7.20887721e-01 9.28285599e-01 5.85269451e-01 -5.65878928e-01 4.78333794e-02 -5.47892094e-01 -1.81318313e-01 -1.03006996e-01 3.10706139e-01 9.08918679e-02 2.36796886e-01 -8.32431197e-01 -4.25143570e-01 4.03847694e-01 -7.27608278e-02 5.91436505e-01 1.26259124e+00 2.43170246e-01 -5.91521561e-01 5.47832949e-03 1.37663710e+00 -1.24438040e-01 -7.61682868e-01 -3.36353064e-01 -8.15272331e-02 -3.19165587e-01 -1.33717537e-01 -3.14909548e-01 -1.40115905e+00 7.67737985e-01 -2.15107024e-01 8.19570184e-01 9.98631835e-01 4.60260510e-01 6.08854473e-01 3.53262424e-01 2.94073522e-01 -3.61184239e-01 -1.45956725e-01 3.55512500e-01 9.83891428e-01 -1.04891491e+00 4.20629643e-02 -9.79887843e-01 -3.02062541e-01 9.98539448e-01 5.93421161e-01 -3.94055396e-01 8.06190848e-01 2.65132473e-03 -7.79956639e-01 -4.27455276e-01 -6.27111912e-01 -4.63106602e-01 7.53063679e-01 7.25234985e-01 2.94759870e-01 3.95825237e-01 -1.52975470e-01 9.27634954e-01 -1.09310612e-01 -5.72680116e-01 2.77209967e-01 3.16069931e-01 -1.85557887e-01 -1.09825850e+00 2.47495964e-01 7.52193272e-01 -2.38297150e-01 -1.69436380e-01 -7.14809477e-01 9.45427120e-01 -2.87139773e-01 7.21406937e-01 1.22329980e-01 -6.98348582e-01 2.43926123e-01 -2.11211368e-01 5.49442947e-01 -8.24985623e-01 -5.96274972e-01 -2.77468592e-01 1.06566340e-01 -6.14508688e-01 -1.64305180e-01 -1.74624592e-01 -1.16615796e+00 -4.35270935e-01 -4.15602833e-01 -2.27797534e-02 1.22656018e-01 7.09257007e-01 5.72319329e-01 6.78708196e-01 4.38612819e-01 -8.66340280e-01 -1.76901713e-01 -8.87997627e-01 -8.11811566e-01 5.80460668e-01 1.00146383e-01 -6.33081794e-01 -2.68092543e-01 -6.42480791e-01]
[7.092013835906982, 6.343496799468994]
398bce5c-c699-467b-86ff-a0eb54d2b70a
a-latent-space-model-for-hla-compatibility
2211.02234
null
https://arxiv.org/abs/2211.02234v1
https://arxiv.org/pdf/2211.02234v1.pdf
A Latent Space Model for HLA Compatibility Networks in Kidney Transplantation
Kidney transplantation is the preferred treatment for people suffering from end-stage renal disease. Successful kidney transplants still fail over time, known as graft failure; however, the time to graft failure, or graft survival time, can vary significantly between different recipients. A significant biological factor affecting graft survival times is the compatibility between the human leukocyte antigens (HLAs) of the donor and recipient. We propose to model HLA compatibility using a network, where the nodes denote different HLAs of the donor and recipient, and edge weights denote compatibilities of the HLAs, which can be positive or negative. The network is indirectly observed, as the edge weights are estimated from transplant outcomes rather than directly observed. We propose a latent space model for such indirectly-observed weighted and signed networks. We demonstrate that our latent space model can not only result in more accurate estimates of HLA compatibilities, but can also be incorporated into survival analysis models to improve accuracy for the downstream task of predicting graft survival times.
['Kevin S. Xu', 'Zhipeng Huang']
2022-11-04
null
null
null
null
['survival-analysis']
['miscellaneous']
[ 2.93384250e-02 -2.18960002e-01 -6.07280910e-01 -6.42103553e-01 -1.22546539e-01 -6.97667956e-01 2.95864493e-01 5.63676357e-01 -2.85057425e-01 8.29628468e-01 3.88993084e-01 -4.32430267e-01 -2.53089309e-01 -1.31671762e+00 -1.73201188e-01 -6.10386431e-01 -5.74511051e-01 7.56893992e-01 -3.66745621e-01 2.20506623e-01 -1.08425118e-01 7.99119890e-01 -7.23956108e-01 -1.01108491e-01 9.77159023e-01 5.51389813e-01 -3.69066983e-01 6.38383389e-01 -4.36710894e-01 4.17440146e-01 -2.17725903e-01 -4.51109022e-01 2.76296288e-01 -5.72462738e-01 -3.63606811e-01 -3.14794630e-01 1.10961989e-01 -6.81179091e-02 -5.89751899e-01 9.70417559e-01 2.65205383e-01 -4.90539461e-01 9.46007848e-01 -1.47886050e+00 -5.46589613e-01 6.19117439e-01 -1.22050814e-01 -4.31242377e-01 6.42601401e-02 2.68283184e-03 1.06366408e+00 -5.77168107e-01 7.02969790e-01 1.19487178e+00 6.10843599e-01 4.37255144e-01 -1.77141964e+00 -7.65155017e-01 -1.00730650e-01 7.41174072e-02 -1.28433478e+00 -2.90338397e-01 6.79365456e-01 -8.08336973e-01 2.49205261e-01 2.88194746e-01 1.09649718e+00 7.18753099e-01 4.54468071e-01 3.97529751e-01 1.16751075e+00 -4.08612818e-01 2.50020236e-01 -2.14589164e-01 4.13546741e-01 6.11466408e-01 2.81232595e-01 4.68815476e-01 -5.34670293e-01 -6.02438092e-01 1.05324554e+00 4.49063748e-01 -4.14676130e-01 -7.76899219e-01 -1.37602484e+00 9.77446973e-01 4.65033829e-01 1.34254560e-01 -5.61666965e-01 1.77270249e-01 8.59073550e-02 5.01506269e-01 1.36014864e-01 -1.34096161e-01 -2.87945658e-01 2.86818475e-01 -4.55413610e-01 -2.02113912e-01 1.02762973e+00 6.85936987e-01 6.71490014e-01 -2.94530302e-01 -2.28049293e-01 6.83184087e-01 4.58652347e-01 3.94675434e-01 -2.08270714e-01 -6.04059279e-01 8.83899163e-03 7.58098304e-01 8.15083385e-02 -5.78876793e-01 -3.75051498e-01 -5.11924505e-01 -1.15192628e+00 3.98308158e-01 8.85481298e-01 8.83215070e-02 -1.07973564e+00 2.14150524e+00 3.63633223e-03 3.40014011e-01 -5.85348569e-02 6.91310823e-01 4.38057572e-01 2.07532957e-01 5.35381079e-01 -3.29030007e-01 1.26082981e+00 -5.15391290e-01 -6.06676042e-01 1.08225793e-02 6.55855894e-01 -6.24059260e-01 2.11702928e-01 -1.44268528e-01 -8.98172855e-01 2.44883224e-01 -8.09261382e-01 2.53253847e-01 -3.42606194e-02 -2.00363740e-01 9.34485555e-01 4.90917116e-01 -9.65591908e-01 7.94949591e-01 -7.75240660e-01 -4.53651279e-01 4.37716246e-01 4.73740250e-01 -5.54310620e-01 -4.16185588e-01 -1.28673136e+00 1.13217127e+00 -2.30347738e-01 2.99495399e-01 -3.99995387e-01 -8.14911485e-01 -6.73155010e-01 1.81857660e-01 -2.18862608e-01 -1.30450284e+00 3.99030924e-01 -6.23002708e-01 -1.15106201e+00 1.05330038e+00 -1.73003823e-01 -1.64040178e-01 5.81910133e-01 5.21390855e-01 -3.22059005e-01 -1.87696561e-01 -2.17689380e-01 3.25533181e-01 2.16380924e-01 -8.89184296e-01 -1.00937963e-01 -5.46747863e-01 -5.15765965e-01 -4.15928476e-02 2.39288732e-02 -9.12391692e-02 -6.34613931e-02 -5.43664515e-01 5.23680687e-01 -1.05901051e+00 -4.86882955e-01 7.99264729e-01 -3.86958688e-01 -2.65390966e-02 5.21937385e-02 -4.70684826e-01 7.12064624e-01 -2.16540384e+00 2.71895766e-01 4.82667536e-01 5.34478307e-01 -1.19414076e-01 -3.85315478e-01 5.97741961e-01 -3.85541499e-01 1.53632402e-01 -1.41793028e-01 2.27199197e-01 -1.86244726e-01 3.08466554e-01 -1.62206769e-01 7.49624729e-01 -5.45955375e-02 1.16402137e+00 -1.00188994e+00 -3.21151465e-01 2.15838224e-01 7.79099941e-01 -1.87977813e-02 7.95846879e-02 2.83547074e-01 4.64840859e-01 -4.48341459e-01 5.58541059e-01 7.76721478e-01 -3.90151352e-01 7.28787959e-01 -7.99296498e-02 -5.97057864e-02 5.06354384e-02 -6.07067287e-01 1.35370123e+00 6.56775683e-02 4.42076981e-01 -1.36169037e-02 -7.74244010e-01 8.89028609e-01 3.10296565e-01 7.02671468e-01 -4.45435554e-01 -1.59533415e-02 1.73730135e-01 2.20959708e-01 3.61696854e-02 -4.74565297e-01 -7.18494356e-01 8.08648579e-03 3.98340702e-01 -2.01027349e-01 3.60953659e-01 -1.12611048e-01 1.82794571e-01 1.19692039e+00 -4.14497733e-01 3.04025739e-01 -3.78377914e-01 3.06146860e-01 -1.30650207e-01 1.10646164e+00 5.73304832e-01 -4.20307279e-01 3.27953815e-01 9.98193860e-01 -4.21986878e-01 -1.04798663e+00 -1.68778694e+00 -5.14642358e-01 2.71517515e-01 2.54566789e-01 2.21924067e-01 1.83899730e-01 -5.19286513e-01 9.52884912e-01 3.50046366e-01 -5.28409839e-01 -2.23183647e-01 -3.68962586e-01 -5.68370759e-01 5.51688850e-01 5.94658434e-01 -1.93394169e-01 -5.38196504e-01 4.91470963e-01 3.73732388e-01 6.26173094e-02 -4.83609170e-01 -4.52596784e-01 2.11782791e-02 -1.27569699e+00 -1.12988555e+00 -9.59899426e-01 -8.15237105e-01 1.03036642e+00 -7.13564157e-02 1.05682528e+00 5.55713952e-01 -5.79678178e-01 9.22739282e-02 2.91506201e-01 5.85246459e-02 -2.96323121e-01 -3.82002234e-01 2.18509018e-01 1.29100144e-01 1.53499216e-01 -8.92181098e-01 -1.01891446e+00 4.64637041e-01 -6.21562123e-01 6.52160943e-02 5.46460450e-01 9.43345129e-01 3.64307404e-01 -3.82610202e-01 7.81645656e-01 -1.05324674e+00 2.98080206e-01 -4.87686157e-01 -5.63854516e-01 8.41761529e-01 -8.90859187e-01 1.59141734e-01 5.35325520e-02 -3.95002633e-01 -5.27213156e-01 4.64920476e-02 3.17835331e-01 -1.32981792e-01 1.49772629e-01 7.81943679e-01 -1.63088799e-01 -2.58734673e-01 -1.68735050e-02 -1.35291843e-02 2.79140502e-01 -3.41362298e-01 2.69830465e-01 2.78495759e-01 2.56154329e-01 -4.13057476e-01 6.50872469e-01 4.49308187e-01 6.46831572e-01 -4.33560729e-01 -1.66188374e-01 -2.95075625e-01 -5.53161025e-01 -2.61611015e-01 5.86822748e-01 -9.19877708e-01 -1.08309078e+00 8.70489478e-01 -8.86607409e-01 -2.49544099e-01 5.90453343e-03 1.02866685e+00 -1.39162198e-01 3.20610404e-01 -1.19573653e+00 -6.47607625e-01 -6.92127869e-02 -6.29004478e-01 3.21249306e-01 1.14002444e-01 -2.95216650e-01 -1.60571086e+00 1.82761326e-01 9.80565995e-02 5.23912609e-01 3.20278049e-01 1.87465620e+00 -4.94768918e-01 -1.05611587e+00 -5.39242864e-01 -4.12901312e-01 -1.17665887e-01 2.04107657e-01 1.58729970e-01 -2.92466372e-01 -3.34106743e-01 -3.42941582e-01 1.36654332e-01 9.10084605e-01 6.50602281e-01 3.46015930e-01 -2.57230606e-02 -6.42800450e-01 6.29484653e-01 1.37635148e+00 -2.00487045e-03 8.06386292e-01 -1.33960277e-01 3.48692358e-01 7.86235392e-01 2.60080814e-01 2.94391394e-01 2.54457861e-01 4.47839707e-01 1.43448278e-01 -2.23402664e-01 -1.14216715e-01 -2.92723894e-01 5.33300601e-02 9.72254872e-01 1.91048875e-01 -3.24230403e-01 -1.13182235e+00 4.77494061e-01 -1.88156009e+00 -7.26688445e-01 -6.80666149e-01 2.63351321e+00 7.96122551e-01 -1.02081560e-01 -1.73435599e-01 -4.70897764e-01 1.05526185e+00 -2.49374464e-01 -7.04700649e-01 2.03645319e-01 -3.49639118e-01 2.41884276e-01 7.18482673e-01 9.17672276e-01 -2.68287629e-01 2.75777608e-01 7.38052034e+00 2.79824436e-02 -1.11253953e+00 -2.79359221e-01 6.25155687e-01 -1.51253775e-01 -7.60699630e-01 8.77652645e-01 -2.26460248e-01 3.65025818e-01 4.15241748e-01 -7.32220292e-01 2.89130539e-01 2.24942327e-01 -8.83066207e-02 2.72770096e-02 -1.47653484e+00 4.96640265e-01 -3.72884721e-01 -1.23900414e+00 -1.70097813e-01 2.89275706e-01 4.95438784e-01 -1.38266057e-01 -5.72209656e-02 3.97139601e-02 6.03255749e-01 -1.02207577e+00 7.14108348e-02 1.02353477e+00 8.46049726e-01 -5.69778681e-01 7.31141865e-01 2.34372526e-01 -9.31339085e-01 3.44655335e-01 -1.48432016e-01 -6.70986697e-02 5.16084194e-01 1.10298169e+00 -7.35009789e-01 2.31144100e-01 -4.71480899e-02 6.50148094e-01 -9.81998518e-02 1.33804882e+00 -3.29226404e-01 3.15826952e-01 -9.60842669e-02 2.09605202e-01 -5.28132141e-01 -6.73330605e-01 4.23044562e-01 5.57393909e-01 4.77821887e-01 2.64347404e-01 2.04100430e-01 1.02125454e+00 -1.18226014e-01 2.87479963e-02 -4.68499541e-01 -3.65881562e-01 6.04680836e-01 1.07807982e+00 -6.22324526e-01 -2.44435713e-01 -3.56940508e-01 6.98180854e-01 2.95812964e-01 8.13020647e-01 -2.45507106e-01 -1.31990835e-01 1.06652665e+00 4.82189320e-02 -2.24988192e-01 -2.00394735e-01 -5.39395630e-01 -1.30076396e+00 -2.11194396e-01 -1.11634165e-01 5.10650158e-01 -5.53445637e-01 -2.02005196e+00 1.39301509e-01 -6.24517679e-01 -1.19476128e+00 2.01923013e-01 -6.04140282e-01 -7.51236796e-01 1.55785692e+00 -1.52602172e+00 -1.11463189e+00 -1.08951084e-01 2.15995744e-01 -4.69846070e-01 2.88714040e-02 1.09005451e+00 3.11356962e-01 -5.57407379e-01 2.29047820e-01 1.15504228e-01 3.62824112e-01 9.34125066e-01 -1.00927436e+00 2.37093747e-01 2.85485476e-01 -3.99372488e-01 1.04278934e+00 5.20723522e-01 -1.09241569e+00 -1.48278654e+00 -8.43603849e-01 1.49943697e+00 -2.41185740e-01 8.21818471e-01 -2.91130871e-01 -7.80507147e-01 7.09201574e-01 -2.08092496e-01 2.90208787e-01 1.21022570e+00 5.76905131e-01 -4.77805316e-01 -8.39529112e-02 -1.01313329e+00 6.65012300e-01 9.13721979e-01 -8.02965581e-01 -9.27291811e-02 2.23509744e-01 1.68630019e-01 8.92742872e-02 -1.23091853e+00 4.39076126e-01 9.29752290e-01 -8.01403463e-01 1.08720458e+00 -9.37055647e-01 1.18788294e-01 -3.38772833e-01 1.17255084e-01 -1.27202380e+00 -5.17632067e-01 -6.99294508e-02 1.69762164e-01 1.15077877e+00 4.26641434e-01 -1.11246133e+00 9.98694420e-01 1.24108803e+00 4.62017924e-01 -4.27920640e-01 -1.16204524e+00 -8.60864699e-01 1.86860099e-01 8.78674239e-02 5.71578503e-01 1.36329460e+00 3.06086749e-01 1.99850976e-01 -2.42351934e-01 2.03662366e-01 1.10376561e+00 5.45343101e-01 3.01263273e-01 -1.88125551e+00 -6.07386641e-02 -7.15835035e-01 -8.09044957e-01 -5.17627537e-01 2.36086279e-01 -1.41119349e+00 -2.44341597e-01 -1.53422105e+00 6.07578814e-01 -1.13929546e+00 -8.39546382e-01 4.26977456e-01 -1.85542986e-01 -1.23225681e-01 1.85836051e-02 3.79217356e-01 2.25616962e-01 3.64095956e-01 1.14670086e+00 1.28540918e-02 -9.56036672e-02 2.24407971e-01 -3.79577786e-01 4.51014161e-01 4.38692510e-01 -8.17982912e-01 -1.20695680e-01 -2.04111174e-01 3.09698582e-01 1.17451882e+00 4.97445941e-01 -1.92534506e-01 3.69033933e-01 -5.61717927e-01 4.99053687e-01 -3.74968439e-01 8.11420083e-02 -7.38823354e-01 8.15877318e-01 1.00198913e+00 -5.33209264e-01 -2.43285060e-01 -4.46415097e-01 7.70504832e-01 -1.56444103e-01 -1.97751261e-02 5.28388202e-01 1.92995891e-01 1.07493915e-01 8.08799446e-01 -3.37343097e-01 -3.80016804e-01 9.91448879e-01 1.14736497e-01 -5.60305595e-01 -5.11597276e-01 -1.10478568e+00 4.16049361e-01 8.80220771e-01 1.37621522e-01 6.55718923e-01 -1.74223685e+00 -9.05183911e-01 3.11912775e-01 2.29509398e-01 -6.04888558e-01 3.15237105e-01 9.72417116e-01 -5.37254155e-01 2.28227094e-01 -3.74420404e-01 -6.64229393e-01 -1.37734950e+00 4.18704063e-01 2.08953455e-01 -2.62578547e-01 -4.76163894e-01 7.34133959e-01 3.74218017e-01 -4.32485729e-01 1.47623509e-01 1.49079189e-01 1.00985870e-01 1.06174126e-01 7.34155700e-02 2.64589131e-01 -4.99102384e-01 -2.46862561e-01 -4.32194352e-01 4.86721188e-01 -1.85523674e-01 -7.22706690e-02 1.48895931e+00 -9.43278335e-03 -6.65927291e-01 5.00443459e-01 1.10614955e+00 -2.99562197e-02 -8.68990541e-01 -4.53858674e-01 8.94444063e-02 -8.83171260e-01 2.34364271e-02 -7.73988783e-01 -1.16686594e+00 8.05191755e-01 5.74158311e-01 -1.69548765e-02 6.12849116e-01 -5.28836716e-03 7.11769402e-01 -1.87306345e-01 5.75893939e-01 -3.26947689e-01 -5.34456074e-01 1.62676379e-01 4.58542585e-01 -9.27908897e-01 -5.19930711e-03 -7.44655192e-01 -2.53232688e-01 1.05526793e+00 4.53709476e-02 1.35673676e-02 6.43311143e-01 3.83636206e-02 5.06251335e-01 -1.81929991e-01 -9.94715452e-01 1.70108676e-01 2.50523657e-01 5.34068823e-01 7.80584514e-01 5.17122090e-01 -4.96369302e-01 1.51410565e-01 3.39965940e-01 1.02110535e-01 2.10550696e-01 9.35058117e-01 -1.93891287e-01 -1.99328673e+00 -1.02998257e-01 6.01262331e-01 -1.03130424e-02 -3.18152383e-02 -5.62696755e-01 2.46717051e-01 -3.18003625e-01 3.67422462e-01 1.26855224e-01 -2.30191136e-03 2.19074070e-01 3.72287750e-01 7.28734255e-01 -3.80919993e-01 -2.47784361e-01 -9.17517766e-02 1.09931223e-01 -1.32725343e-01 -1.58307016e-01 -8.97034287e-01 -1.19423962e+00 -7.55776286e-01 -4.93688166e-01 6.91349283e-02 6.61651313e-01 6.94951594e-01 2.37809747e-01 2.71291077e-01 9.15364742e-01 -1.19758040e-01 -4.91016388e-01 -4.83551174e-01 -1.02140152e+00 4.59149808e-01 4.67742532e-01 -5.27512431e-01 -5.02670348e-01 -1.94032624e-01]
[7.1548237800598145, 5.251214504241943]
1ec4983b-3e43-4696-a3da-3baabbd301f2
user-localization-using-rf-sensing-a
2205.10321
null
https://arxiv.org/abs/2205.10321v1
https://arxiv.org/pdf/2205.10321v1.pdf
User Localization using RF Sensing: A Performance comparison between LIS and mmWave Radars
Since electromagnetic signals are omnipresent, Radio Frequency (RF)-sensing has the potential to become a universal sensing mechanism with applications in localization, smart-home, retail, gesture recognition, intrusion detection, etc. Two emerging technologies in RF-sensing, namely sensing through Large Intelligent Surfaces (LISs) and mmWave Frequency-Modulated Continuous-Wave (FMCW) radars, have been successfully applied to a wide range of applications. In this work, we compare LIS and mmWave radars for localization in real-world and simulated environments. In our experiments, the mmWave radar achieves 0.71 Intersection Over Union (IOU) and 3cm error for bounding boxes, while LIS has 0.56 IOU and 10cm distance error. Although the radar outperforms the LIS in terms of accuracy, LIS features additional applications in communication in addition to sensing scenarios.
['Stephan Sigg', 'Zheng-Hua Tan', 'Elisabeth de Carvalho', 'Petar Popovski', 'Dariush Salami', 'Cristian J. Vaca-Rubio']
2022-05-17
null
null
null
null
['gesture-recognition']
['computer-vision']
[ 5.29904306e-01 -1.88714772e-01 1.76605597e-01 -1.91099346e-01 -5.58987975e-01 -6.13801241e-01 5.52339435e-01 -2.40507647e-01 -4.06187683e-01 9.86789346e-01 -1.45567000e-01 -4.06626523e-01 -5.32869756e-01 -1.18121946e+00 -2.90814042e-01 -8.64513874e-01 -3.47719401e-01 2.78578430e-01 1.82430029e-01 -1.24334404e-02 2.01135144e-01 9.16004956e-01 -1.45977950e+00 -2.94098437e-01 6.41248703e-01 1.43627763e+00 7.22647533e-02 4.98555273e-01 -1.02418192e-01 6.10329024e-02 -9.22849953e-01 2.74392486e-01 3.07737947e-01 2.83350915e-01 -7.81405568e-02 -8.42438519e-01 2.49015048e-01 1.08589023e-01 -5.18473871e-02 7.39416718e-01 6.59509063e-01 -8.80008657e-03 8.14742088e-01 -7.57513165e-01 -3.17117214e-01 3.94006103e-01 -1.04086471e+00 5.61462231e-02 8.66763711e-01 -4.27016020e-01 1.66178927e-01 -6.80894971e-01 2.49921143e-01 1.02564561e+00 8.42894137e-01 2.07041770e-01 -6.69142306e-01 -1.22526753e+00 -5.44275761e-01 -2.29516327e-01 -1.75719821e+00 -2.33924955e-01 5.09348691e-01 -2.02256754e-01 6.49213195e-01 5.57672799e-01 2.72103876e-01 7.49516129e-01 7.10469723e-01 -4.80397716e-02 1.32367647e+00 -6.05160356e-01 4.51079369e-01 -1.51696473e-01 2.32127705e-03 3.89195770e-01 9.26397622e-01 4.94253933e-01 -4.19488430e-01 -2.45684221e-01 7.07979083e-01 3.21483463e-01 -2.02782944e-01 6.11936077e-02 -1.13485324e+00 4.24485445e-01 3.14255804e-01 1.00166106e+00 -4.33167458e-01 3.91487598e-01 -6.02645993e-01 1.40875518e-01 2.57445246e-01 6.22089744e-01 -2.80654341e-01 2.88613215e-02 -7.37179875e-01 6.08758181e-02 6.97461486e-01 7.85819352e-01 5.63364387e-01 1.55308649e-01 1.34355322e-01 4.28093076e-01 6.14813924e-01 1.90371549e+00 -3.95613760e-02 -3.73891443e-01 2.76915371e-01 1.84153914e-01 4.38361913e-01 -1.22329414e+00 -1.00017834e+00 -7.67145038e-01 -1.00213242e+00 4.32076901e-02 2.30112761e-01 -5.99991798e-01 -8.59573007e-01 1.36900675e+00 3.67824048e-01 5.18970132e-01 1.13561362e-01 5.95610023e-01 7.60671496e-01 4.78672236e-01 -1.18408455e-02 -1.77856401e-01 1.33643305e+00 2.18815058e-01 -5.26876092e-01 -3.01330388e-01 3.41472894e-01 -9.09925401e-01 2.64136553e-01 4.71751958e-01 -3.56783211e-01 -4.73199576e-01 -1.18422091e+00 1.10876143e+00 -3.68400097e-01 -1.70761988e-01 5.66525102e-01 1.19811058e+00 -5.78427672e-01 -9.27498788e-02 -7.58867502e-01 -5.26188195e-01 2.01733008e-01 2.64631182e-01 2.19021037e-01 -1.71467975e-01 -1.24511611e+00 6.78940535e-01 -3.33559006e-01 -1.14369668e-01 -3.47881876e-02 -6.83455825e-01 -6.03148103e-01 -3.04082721e-01 -1.19869962e-01 -3.05684894e-01 6.49662316e-01 1.68944493e-01 -1.44485962e+00 3.25690985e-01 -3.58265750e-02 -4.72532660e-01 -1.48070768e-01 -3.26934904e-01 -1.33989894e+00 8.71085152e-02 1.90894261e-01 -4.40700538e-02 4.89669055e-01 -9.68134642e-01 -6.90402031e-01 -7.24148333e-01 -2.37767383e-01 -4.36069757e-01 1.33134544e-01 -1.60699219e-01 7.36507535e-01 -4.27744329e-01 7.65217662e-01 -9.10685897e-01 -2.84632444e-01 -5.15278757e-01 -2.54772156e-01 -2.84476653e-02 9.71082509e-01 1.52997464e-01 9.28839564e-01 -2.00508451e+00 -8.49058509e-01 8.35892856e-01 -2.60649532e-01 1.58794850e-01 1.11181095e-01 4.43210185e-01 6.70739889e-01 -1.45545855e-01 -7.38800317e-02 2.81582534e-01 -7.96941370e-02 -2.58321673e-01 -3.97046894e-01 7.54144847e-01 -4.04098302e-01 7.12627769e-01 -6.98404551e-01 -1.03631586e-01 4.17052865e-01 6.17283821e-01 2.02863991e-01 -3.64194155e-01 3.22084129e-01 8.64118516e-01 -9.69706535e-01 1.07937765e+00 1.25352764e+00 1.42741486e-01 -2.72675067e-01 1.91255286e-02 -5.16767144e-01 -2.09750593e-01 -1.48617780e+00 1.42591774e+00 -6.66529655e-01 5.01096785e-01 2.00359240e-01 -6.90452933e-01 1.55946481e+00 1.39921203e-01 7.78735340e-01 -1.22562754e+00 -1.07437782e-02 2.92685598e-01 -5.19625187e-01 -5.52127898e-01 1.79261804e-01 -3.67425948e-01 -3.60903025e-01 6.53286397e-01 -5.32638550e-01 7.77898654e-02 -2.80690134e-01 -4.60653186e-01 1.61772573e+00 -2.11945698e-01 3.02182287e-01 -3.82628649e-01 6.09864533e-01 -9.91247147e-02 1.10359430e-01 9.12167192e-01 1.66777790e-01 7.04796389e-02 -1.03516233e+00 -3.87264222e-01 6.14616163e-02 -1.45548880e+00 -6.24443889e-01 6.82476640e-01 8.04537237e-01 3.16764444e-01 -1.71352446e-01 -1.09799042e-01 4.62501258e-01 6.67408764e-01 7.41495658e-03 1.95257366e-01 -5.57878196e-01 -8.40993941e-01 9.52851892e-01 4.20786142e-01 6.21914983e-01 -8.10670853e-01 -1.38572145e+00 2.60547429e-01 1.08973026e-01 -1.37493646e+00 3.66633952e-01 -4.38787006e-02 -5.83411455e-01 -8.67999732e-01 -5.04478633e-01 -3.64400476e-01 3.40702802e-01 6.83197081e-01 7.65771568e-01 -2.58228838e-01 -6.93931162e-01 6.57254219e-01 -3.67503464e-01 -9.29234743e-01 3.59351218e-01 -1.19136237e-01 5.49357772e-01 2.81599816e-03 7.70376503e-01 -8.42302024e-01 -7.31410146e-01 5.79612136e-01 -4.16790664e-01 -7.35049486e-01 1.00261652e+00 2.18099654e-01 3.49040478e-01 8.65153000e-02 7.49324501e-01 -6.05719984e-01 2.20561922e-01 -4.33009088e-01 -8.20742607e-01 5.61462119e-02 -4.41526949e-01 -2.16545507e-01 3.36567789e-01 -2.71532267e-01 -9.04897571e-01 -8.42810348e-02 -5.11604026e-02 3.04206342e-01 -6.90670073e-01 3.95710588e-01 -5.40078171e-02 -6.82339847e-01 9.67783928e-01 -4.94456552e-02 -6.97649598e-01 -3.68864059e-01 1.21754944e-01 9.47084606e-01 5.48922420e-01 -3.47087920e-01 1.33931327e+00 8.54494095e-01 4.74249810e-01 -1.43802869e+00 -5.42333126e-01 -7.46895850e-01 -2.89959550e-01 -2.25814730e-01 6.10434473e-01 -7.61185110e-01 -9.51003492e-01 4.25781533e-02 -1.06043553e+00 1.93789899e-01 8.57620910e-02 1.00329041e+00 -2.04015225e-01 -6.56110793e-02 1.74598284e-02 -1.47431278e+00 -4.45114344e-01 -1.21593453e-01 1.11646664e+00 6.36729836e-01 -2.97103912e-01 -6.72871709e-01 3.12865764e-01 1.40163124e-01 8.18558931e-01 7.34247088e-01 3.37787926e-01 -2.45391443e-01 -7.15524256e-01 -6.28030717e-01 -1.04676887e-01 -6.88616037e-01 1.97743952e-01 -8.39157462e-01 -9.97603059e-01 -2.00164586e-01 -7.09442198e-02 2.72455812e-01 5.70997179e-01 7.11561739e-01 5.23649812e-01 2.87861466e-01 -1.12125647e+00 7.83962786e-01 1.62828505e+00 8.48689377e-01 7.92559266e-01 2.34908029e-01 4.68266755e-02 1.70335755e-01 1.10152149e+00 5.09395599e-01 -2.78888106e-01 5.06230533e-01 2.46562526e-01 -6.24371544e-02 2.08694279e-01 7.31215999e-02 1.31166846e-01 6.49015978e-02 -2.98491895e-01 -1.79747671e-01 -9.21071351e-01 -3.61837484e-02 -1.43795609e+00 -1.19623852e+00 -3.60706747e-01 2.27930856e+00 -4.92922142e-02 1.22199636e-02 -4.43863988e-01 2.06033379e-01 6.42910600e-01 1.53804898e-01 -3.32809031e-01 -8.40807632e-02 1.71719138e-02 8.29598129e-01 1.14708114e+00 4.39186990e-01 -1.05003142e+00 3.59809220e-01 6.27983475e+00 4.61549044e-01 -1.27944243e+00 1.57528237e-01 -1.20941073e-01 2.08225071e-01 -2.76836425e-01 -3.90476733e-01 -8.84716511e-01 2.13558689e-01 7.05993176e-01 2.38892555e-01 -5.57406014e-03 4.19267803e-01 6.04137219e-02 -6.79588556e-01 -4.08220530e-01 1.17622459e+00 -2.02740118e-01 -1.09751928e+00 -3.04055095e-01 1.39288411e-01 4.99332011e-01 3.72325145e-02 1.06884003e-01 1.12401582e-01 1.06454566e-02 -1.08908951e+00 2.85052657e-01 6.32552803e-01 1.23398125e+00 -9.27965760e-01 9.98867750e-01 4.30668205e-01 -1.56874728e+00 1.01430885e-01 -4.64365929e-01 -4.58264887e-01 3.34026933e-01 1.11951387e+00 -8.20219994e-01 6.70783043e-01 7.51193821e-01 -5.77706285e-02 1.52614415e-01 9.23287630e-01 -9.12899524e-02 4.59214389e-01 -7.88183630e-01 -6.35631561e-01 1.11987116e-02 -3.63481849e-01 6.40243530e-01 1.29481637e+00 8.00806582e-01 5.48566759e-01 1.27992883e-01 6.38729751e-01 5.66876948e-01 -1.42434075e-01 -9.20098066e-01 3.91847610e-01 9.24964011e-01 1.19484198e+00 -7.38140881e-01 1.82040885e-01 -1.93288356e-01 1.76376790e-01 -9.16772723e-01 4.25733954e-01 -7.86418855e-01 -1.08916867e+00 7.32062876e-01 5.30782461e-01 1.83188707e-01 -7.22679853e-01 -5.14443874e-01 -5.32208920e-01 -2.29702950e-01 1.70284100e-02 -4.39472087e-02 -3.71204317e-01 -8.38443756e-01 4.10580993e-01 -1.44301534e-01 -1.30122125e+00 -1.10904604e-01 -5.23059487e-01 -5.10878682e-01 6.30969942e-01 -1.59074879e+00 -9.46128964e-01 -8.06384146e-01 5.12588680e-01 -1.75087780e-01 -1.67663783e-01 1.00180387e+00 3.15514952e-01 7.95162618e-02 4.21711802e-01 3.99438292e-01 7.10299015e-02 5.64810812e-01 -6.00775182e-01 2.69088179e-01 5.22930503e-01 3.24278682e-01 6.06801152e-01 8.42871368e-01 -6.91774845e-01 -1.73089528e+00 -1.08524394e+00 6.16316080e-01 -1.23285390e-01 4.04478937e-01 -2.47648254e-01 -6.28709495e-02 2.00584590e-01 -3.69744420e-01 3.95221375e-02 7.45240629e-01 5.50424755e-02 -1.88503370e-01 -5.28222919e-01 -1.66001284e+00 2.37790227e-01 1.28384984e+00 -1.01200581e-01 -3.81706208e-01 2.10087463e-01 2.82957375e-01 -1.78711072e-01 -8.17396700e-01 8.69930089e-01 1.18094218e+00 -7.99321651e-01 1.19985771e+00 3.53442162e-01 -5.49209297e-01 -3.34619939e-01 -5.48165083e-01 -9.14112687e-01 -3.13558519e-01 -5.61870694e-01 1.14638701e-01 1.05670404e+00 2.98888832e-01 -1.20875871e+00 8.97648931e-01 -2.25473538e-01 1.31716266e-01 -4.79848832e-01 -1.21813452e+00 -9.94085133e-01 -5.16964376e-01 -7.20731556e-01 9.38312232e-01 5.99383771e-01 5.29835150e-02 3.82170856e-01 -8.88630822e-02 6.79031193e-01 1.05491650e+00 4.44953263e-01 7.46183336e-01 -1.96887314e+00 -7.42936954e-02 6.25826120e-02 -6.19423807e-01 -1.24020565e+00 -2.50082165e-01 -5.58672309e-01 1.00286923e-01 -1.60200596e+00 -5.94080329e-01 -9.56929028e-01 -4.39166903e-01 6.49027824e-02 7.17614830e-01 6.68108761e-01 -2.25397319e-01 -5.78828901e-02 -5.22413611e-01 2.13398365e-03 7.44712353e-01 -2.14537352e-01 -3.03477645e-01 6.47970319e-01 -3.56207132e-01 7.27108181e-01 8.37957025e-01 -4.85194743e-01 -1.45193106e-02 -1.61274731e-01 1.31487519e-01 5.85537069e-02 5.63888997e-02 -1.71295178e+00 5.00235677e-01 -4.20650065e-01 8.17612827e-01 -7.57469714e-01 4.05510217e-01 -1.05233288e+00 4.58404124e-01 9.84062433e-01 5.69910645e-01 -3.23660374e-01 8.07777122e-02 7.47351527e-01 1.61971405e-01 1.19311146e-01 8.33116233e-01 4.23861668e-02 -6.79056704e-01 1.13446303e-01 -5.82279444e-01 -3.00474018e-01 1.19874942e+00 -6.21227264e-01 -5.25139213e-01 -3.92721951e-01 -2.34473541e-01 -1.35440499e-01 4.58599925e-02 6.45912290e-02 7.51355350e-01 -1.05918944e+00 -4.52861935e-01 4.03801769e-01 1.28968865e-01 -3.53276759e-01 -1.03659471e-02 8.42618704e-01 -3.00508618e-01 9.29507434e-01 1.28972642e-02 -8.76515090e-01 -1.07686460e+00 2.09732214e-03 7.79603561e-03 1.03336498e-01 -4.34962034e-01 6.16895616e-01 -1.45915434e-01 -3.02400976e-01 1.69145718e-01 -2.09490329e-01 -2.15681698e-02 -3.85605127e-01 8.57454956e-01 5.96473277e-01 2.44921613e-02 -3.77177179e-01 -9.24662590e-01 1.62327480e+00 6.85902178e-01 -2.77823299e-01 1.19592857e+00 -1.34941697e-01 1.64205030e-01 3.67514580e-01 4.10717040e-01 6.92099988e-01 -4.10099626e-01 -1.23958036e-01 3.99151742e-01 -3.66863489e-01 -1.21782646e-01 -9.13238943e-01 -6.00217581e-01 3.95366818e-01 1.03276706e+00 4.25182998e-01 1.08437288e+00 1.96620300e-01 9.39035356e-01 7.86577463e-01 1.49872684e+00 -7.56825566e-01 -3.88028234e-01 4.66688603e-01 3.33513290e-01 -6.51476264e-01 1.26900017e-01 -4.04629737e-01 3.16090107e-01 9.77063954e-01 1.21113166e-01 -3.08965772e-01 1.13911605e+00 1.02876651e+00 4.20198739e-01 -2.00547606e-01 3.84785831e-02 -3.56130481e-01 3.77691686e-02 1.12850344e+00 3.88650149e-01 4.04924154e-01 -4.03134137e-01 4.82162297e-01 -3.66853178e-01 -1.00059383e-01 9.49064791e-02 1.09926057e+00 -1.23992121e+00 -6.97360218e-01 -1.03145230e+00 6.87181473e-01 -3.39665949e-01 3.54919106e-01 -8.17668736e-02 7.24569261e-01 3.57745528e-01 1.49631500e+00 1.08769692e-01 -7.33112574e-01 4.98904675e-01 -1.04792044e-01 6.43744111e-01 -2.66741037e-01 -8.56976733e-02 8.27607233e-03 -1.54673472e-01 -6.44700408e-01 -6.75863266e-01 -4.06641513e-01 -1.66216660e+00 -9.61948112e-02 -5.25071979e-01 3.76907259e-01 1.07643580e+00 1.09101021e+00 4.30315435e-01 2.97490478e-01 7.81717598e-01 -5.71783364e-01 -2.50350147e-01 -1.05284131e+00 -1.11355817e+00 -3.00536901e-01 9.93133187e-02 -1.16211295e+00 -4.22314733e-01 -7.69836366e-01]
[6.557775497436523, 0.8021537661552429]
f7634584-e041-401d-98a7-99a3dae2fdf9
explanation-generation-for-multi-modal-multi
2008.03573
null
https://arxiv.org/abs/2008.03573v1
https://arxiv.org/pdf/2008.03573v1.pdf
Explanation Generation for Multi-Modal Multi-Agent Path Finding with Optimal Resource Utilization using Answer Set Programming
The multi-agent path finding (MAPF) problem is a combinatorial search problem that aims at finding paths for multiple agents (e.g., robots) in an environment (e.g., an autonomous warehouse) such that no two agents collide with each other, and subject to some constraints on the lengths of paths. We consider a general version of MAPF, called mMAPF, that involves multi-modal transportation modes (e.g., due to velocity constraints) and consumption of different types of resources (e.g., batteries). The real-world applications of mMAPF require flexibility (e.g., solving variations of mMAPF) as well as explainability. Our earlier studies on mMAPF have focused on the former challenge of flexibility. In this study, we focus on the latter challenge of explainability, and introduce a method for generating explanations for queries regarding the feasibility and optimality of solutions, the nonexistence of solutions, and the observations about solutions. Our method is based on answer set programming. This paper is under consideration for acceptance in TPLP.
['Esra Erdem', 'Aysu Bogatarkan']
2020-08-08
null
null
null
null
['multi-agent-path-finding']
['playing-games']
[-2.06446320e-01 4.63732362e-01 -1.58002123e-01 -2.15704471e-01 -2.35998318e-01 -1.08419073e+00 1.26536489e-01 5.30100524e-01 -4.23673615e-02 9.83544827e-01 -4.60817516e-01 -6.13259077e-01 -8.95841956e-01 -1.15174723e+00 -8.57894540e-01 -6.03154063e-01 -4.39559072e-01 1.08899009e+00 2.45932728e-01 -4.62738276e-01 3.66563499e-01 6.29842222e-01 -1.25706851e+00 -2.46737525e-01 8.89882982e-01 7.69992054e-01 7.50389278e-01 4.66973871e-01 -3.18482757e-01 1.10582367e-01 -5.73446095e-01 -2.21625879e-01 3.16710353e-01 -1.22045457e-01 -1.10414064e+00 3.80588979e-01 -5.88250279e-01 -1.20284438e-01 -6.59999298e-03 9.36951160e-01 -1.24131292e-01 2.59130895e-01 4.62038249e-01 -2.58760333e+00 -6.05056465e-01 7.24284232e-01 -3.91508520e-01 -1.70881391e-01 4.90740418e-01 2.96239816e-02 7.28481472e-01 -4.93335932e-01 5.93233526e-01 1.10481155e+00 5.23032472e-02 4.38547462e-01 -1.13074040e+00 7.52201229e-02 5.59804738e-01 3.80454749e-01 -1.31116915e+00 -1.70189023e-01 2.78044432e-01 -3.20657715e-02 1.03077292e+00 7.35922456e-01 3.61288130e-01 3.29314977e-01 5.27372181e-01 4.89223987e-01 6.31056786e-01 -3.26297939e-01 6.28139555e-01 1.81531116e-01 1.32136598e-01 3.90164614e-01 7.66908646e-01 -2.63932645e-01 -2.62591660e-01 -3.41726631e-01 4.92315620e-01 -6.32267222e-02 -2.37689078e-01 -6.88335061e-01 -1.24945152e+00 9.74142909e-01 1.15937628e-01 8.21920484e-03 -5.62940657e-01 1.08403981e-01 -3.98311019e-02 3.46070021e-01 -2.98188180e-01 8.28710616e-01 -6.68416560e-01 1.02956079e-01 2.97651906e-03 4.91842836e-01 1.05093467e+00 1.71295142e+00 8.44082713e-01 -3.94865245e-01 2.11783677e-01 1.72382757e-01 4.86196697e-01 6.00879550e-01 -3.37435722e-01 -1.31053197e+00 7.56994247e-01 5.01745462e-01 9.44425941e-01 -1.13978112e+00 -7.16983378e-01 1.71561122e-01 -3.30374330e-01 6.71441630e-02 3.60423058e-01 -1.58881649e-01 -5.16758621e-01 1.72869337e+00 3.66019517e-01 -2.31280267e-01 3.31902355e-01 9.14878666e-01 4.33790147e-01 1.02098358e+00 -3.42982501e-01 -6.13651395e-01 1.04251742e+00 -1.12342906e+00 -6.41801298e-01 -1.54353797e-01 5.44242978e-01 -3.93595994e-01 6.20427668e-01 3.89718652e-01 -1.44252646e+00 2.04615533e-01 -8.40927958e-01 3.69919509e-01 -5.41739941e-01 -5.29533327e-01 7.10711718e-01 4.09735680e-01 -1.23433161e+00 1.21023878e-01 -6.33883953e-01 -6.38502955e-01 -4.99143064e-01 6.82209253e-01 -3.60894203e-01 -4.47675765e-01 -9.61729586e-01 9.25102949e-01 3.66743773e-01 2.87025452e-01 -4.59941208e-01 -1.08049177e-01 -7.79167295e-01 2.15169281e-01 1.09030056e+00 -5.99965274e-01 1.27113235e+00 -4.27718878e-01 -1.21393096e+00 9.37439650e-02 -2.18857422e-01 -1.79368347e-01 3.59595239e-01 4.99892145e-01 -2.56401807e-01 8.57503638e-02 4.83371019e-01 5.25832891e-01 -1.31964013e-02 -1.50980413e+00 -8.64712596e-01 -2.88833618e-01 7.52367258e-01 3.29088002e-01 1.36220917e-01 -2.00268596e-01 -3.28265369e-01 1.96044579e-01 3.88932258e-01 -1.19953740e+00 -7.09307194e-01 -2.76449770e-01 -7.32274532e-01 -4.57704604e-01 4.84024316e-01 8.42995942e-02 8.85870755e-01 -1.65971982e+00 3.26241344e-01 6.86133385e-01 -3.74974944e-02 -5.42303383e-01 -4.41186160e-01 9.89411414e-01 3.09824586e-01 5.09847224e-01 -1.42280102e-01 -2.22090911e-03 3.84706408e-01 8.40562105e-01 -3.81350443e-02 3.66014928e-01 2.48157978e-02 5.64631879e-01 -1.00729752e+00 -1.81091338e-01 -1.15841180e-01 -5.08692205e-01 -3.55330825e-01 -1.36071354e-01 -6.80049062e-01 2.25786313e-01 -6.50044084e-01 7.04479933e-01 7.83409715e-01 -8.50021541e-02 3.73295546e-01 3.49090785e-01 -6.81904495e-01 -9.91213694e-02 -1.47147846e+00 1.36934566e+00 -3.04712623e-01 8.50004554e-02 2.88056642e-01 -6.48831487e-01 7.11451054e-01 -5.95989898e-02 6.23595655e-01 -5.63358128e-01 -1.43006846e-01 5.30658066e-01 7.01095015e-02 -5.47286391e-01 8.15410256e-01 3.26924294e-01 -4.47611034e-01 7.24009991e-01 -7.23421514e-01 -3.95007096e-02 5.57576418e-01 2.21155658e-01 1.12434328e+00 -3.84497166e-01 2.57534772e-01 -5.34976184e-01 2.48825043e-01 6.21151567e-01 8.43258619e-01 7.88227677e-01 -4.99613583e-02 9.13396478e-02 5.60566425e-01 -3.07486415e-01 -7.53591359e-01 -8.90789449e-01 3.89971942e-01 5.99719107e-01 1.04283166e+00 -2.87135363e-01 -3.84060413e-01 -4.06557411e-01 2.40804851e-01 9.85143125e-01 -2.87240386e-01 2.69500196e-01 -5.97366750e-01 -5.73399782e-01 -2.69866347e-01 1.98404163e-01 1.83633149e-01 -5.87266743e-01 -1.05295420e+00 5.00275373e-01 -4.69149083e-01 -9.88067031e-01 -4.09785569e-01 2.95037866e-01 -4.19293642e-01 -1.30175304e+00 -2.41985530e-01 -6.73795700e-01 1.10159731e+00 9.77721453e-01 8.12256396e-01 2.55321801e-01 9.18102115e-02 7.52481699e-01 -4.70241249e-01 -4.27543432e-01 -7.33978227e-02 -6.43908232e-02 1.34849325e-01 -4.04705018e-01 -1.32181302e-01 -9.73978862e-02 -4.11475122e-01 9.02595878e-01 -8.07137728e-01 -6.29767105e-02 4.30472404e-01 3.37950587e-01 1.04258025e+00 9.82098758e-01 7.53813386e-01 -4.28205431e-01 8.79404008e-01 -9.50852573e-01 -7.09211171e-01 7.28096664e-01 -6.78287745e-01 -7.81864300e-03 5.38474977e-01 -1.06720783e-01 -6.81244910e-01 -6.05373718e-02 6.46826565e-01 4.74132039e-02 4.28346507e-02 8.19775462e-01 -5.75527251e-01 -1.06144011e-01 -9.92875248e-02 1.25442874e-02 8.89567565e-03 3.90689932e-02 3.47502381e-01 1.97084218e-01 4.07367557e-01 -6.56575620e-01 6.01531863e-01 2.03485414e-01 5.61592698e-01 -3.79695386e-01 1.48545668e-01 -1.62158698e-01 -1.29931867e-01 -3.31414878e-01 5.78917921e-01 -1.35155424e-01 -1.19461739e+00 1.46527588e-02 -1.43978572e+00 -2.77541876e-01 -1.48753989e-02 2.53544599e-01 -8.71880114e-01 1.36834234e-01 -4.68197726e-02 -1.13247335e+00 2.82204747e-01 -1.20285833e+00 4.47888881e-01 3.07896763e-01 -1.52995557e-01 -7.41900682e-01 -1.64086252e-01 1.12331323e-01 4.01352078e-01 5.15553415e-01 1.24045312e+00 -6.51359081e-01 -1.17484760e+00 1.48694098e-01 2.30270438e-02 -8.19346547e-01 1.37331039e-01 -2.07709037e-02 1.24356665e-01 -3.40224415e-01 -1.74895734e-01 1.70163617e-01 -9.65060666e-02 4.96587366e-01 8.06040645e-01 -9.32610750e-01 -7.31306493e-01 -1.93136081e-01 1.59459507e+00 9.49356735e-01 3.86130750e-01 8.57564151e-01 -7.55658746e-02 1.14405763e+00 1.20663548e+00 6.27955496e-01 1.13025725e+00 8.16291749e-01 9.85054314e-01 4.15515602e-01 7.67558455e-01 1.71061501e-01 -7.61368349e-02 3.94513637e-01 1.71872720e-01 -1.14693868e+00 -9.69153643e-01 6.41403019e-01 -2.38989639e+00 -6.26086593e-01 -5.32053232e-01 2.03541303e+00 -6.98437961e-03 -3.61021422e-02 2.65373498e-01 1.26543455e-02 9.16157663e-01 -4.46979970e-01 -8.45459819e-01 -9.10416543e-01 -8.16523507e-02 -6.56288624e-01 8.41420293e-01 7.12280691e-01 -4.99983996e-01 2.52824724e-01 5.48121738e+00 2.44191810e-02 -5.13064802e-01 -8.09590146e-02 4.50304002e-01 3.76248397e-02 -8.29402804e-01 1.24993026e-01 -4.76810694e-01 4.16015267e-01 8.23500097e-01 -6.01327062e-01 9.69073117e-01 6.40879214e-01 3.98727477e-01 -4.65932280e-01 -1.17804372e+00 6.64185226e-01 -2.52280921e-01 -1.17134917e+00 -3.01929295e-01 2.38148719e-01 5.80952644e-01 -4.45414037e-01 -4.39973846e-02 -9.43282470e-02 3.59908044e-01 -7.62454033e-01 1.06069958e+00 2.73850918e-01 1.29263431e-01 -1.15728557e+00 7.29936779e-01 5.10456860e-01 -1.21627164e+00 -4.55637723e-01 -2.79886991e-01 -9.07403901e-02 6.75796449e-01 2.40987390e-01 -7.47752249e-01 1.08631575e+00 6.88207150e-01 -1.08271018e-01 2.81525940e-01 1.29049551e+00 -3.25287320e-02 -5.23210406e-01 -6.44448340e-01 -3.75940561e-01 5.02602220e-01 -4.23440188e-01 5.31873405e-01 5.61236560e-01 6.71669185e-01 6.18373632e-01 5.00343204e-01 9.38880026e-01 3.63033116e-01 -2.23382562e-01 -5.86831808e-01 7.64859319e-02 9.33487952e-01 1.06060445e+00 -1.09732401e+00 1.73604682e-01 -2.52808392e-01 5.15076041e-01 -7.58353099e-02 4.33581740e-01 -8.92002225e-01 -3.25334579e-01 6.77315652e-01 5.23075182e-03 -7.02122450e-02 -5.74007690e-01 -5.63693523e-01 -4.41232473e-01 2.18161628e-01 -4.68330741e-01 3.64469886e-01 -8.97437930e-01 -9.02127147e-01 5.48161983e-01 3.13221484e-01 -8.98706198e-01 -2.84530044e-01 -2.72616029e-01 -8.48099053e-01 8.70372295e-01 -1.68532026e+00 -6.91365480e-01 -1.89592719e-01 4.95681047e-01 4.70933229e-01 2.14412168e-01 7.18296766e-01 -1.45130485e-01 -6.37008250e-01 -6.37632385e-02 -5.92040680e-02 -9.53219235e-01 -1.51151866e-02 -1.11732674e+00 6.95937425e-02 8.12403560e-01 -6.49752140e-01 5.62582552e-01 1.01702142e+00 -6.74700737e-01 -2.24938679e+00 -1.08250964e+00 9.85304356e-01 -9.12197959e-03 5.17343879e-01 -9.62161347e-02 -4.21671450e-01 7.59788930e-01 1.58205926e-01 -6.38778567e-01 4.58163112e-01 -2.06928179e-01 4.91337359e-01 2.70474404e-02 -1.57426846e+00 7.10457742e-01 1.03289056e+00 4.83256727e-01 -1.03696972e-01 6.32008255e-01 8.68069112e-01 -3.78010899e-01 -5.18824339e-01 1.77918553e-01 1.71553671e-01 -5.11829078e-01 6.23644710e-01 -6.55961275e-01 3.48584615e-02 -7.06604898e-01 -4.91591036e-01 -1.50774205e+00 -5.17580211e-01 -7.50079453e-01 1.71575442e-01 1.06810892e+00 8.01864505e-01 -1.07108569e+00 6.80280089e-01 1.34150231e+00 -4.90041137e-01 -9.37139034e-01 -1.17086864e+00 -1.30348909e+00 -2.60724247e-01 -2.42717266e-02 1.08850884e+00 6.46400809e-01 3.77479315e-01 -6.56890273e-02 -1.87917799e-01 8.33751321e-01 5.62409520e-01 6.15852714e-01 5.79366684e-01 -1.03877676e+00 -8.89953077e-02 -2.47526854e-01 1.34402648e-01 -8.64187062e-01 5.39433733e-02 -5.28129935e-01 4.28715736e-01 -2.34198546e+00 -7.87131414e-02 -1.04261148e+00 1.75608069e-01 5.80287337e-01 5.50197661e-01 -6.09521866e-01 5.75094461e-01 3.40287626e-01 -8.08302760e-01 2.32761383e-01 1.17226601e+00 -2.05388919e-01 -3.66324723e-01 2.13304475e-01 -8.59047830e-01 2.63497412e-01 1.16219175e+00 -3.63791555e-01 -7.83733606e-01 -7.35476494e-01 5.03715158e-01 8.92412722e-01 -4.20040861e-02 -2.17156366e-01 6.38692260e-01 -1.08471406e+00 -4.72318172e-01 -6.86444283e-01 4.85175014e-01 -1.20846999e+00 7.17179537e-01 6.90708458e-01 -1.73133016e-02 9.55291629e-01 1.28250374e-02 6.30530179e-01 8.42874572e-02 -5.35723209e-01 1.19968124e-01 -2.67743707e-01 -8.52350891e-01 2.18721718e-01 -6.94484234e-01 -5.14513314e-01 1.69882214e+00 -2.91622698e-01 -9.96212721e-01 -5.40782928e-01 -5.47320366e-01 1.17098331e+00 4.13766831e-01 3.42744023e-01 8.44726861e-01 -1.15935886e+00 -4.06525016e-01 -2.32880667e-01 7.89797027e-03 2.15734392e-01 1.19771585e-01 8.40230882e-01 -3.21493775e-01 9.09775078e-01 -3.04146320e-01 -1.12816043e-01 -9.00914013e-01 1.02401412e+00 4.38870788e-02 -3.86315249e-02 -2.24514782e-01 3.85079533e-01 5.82968257e-02 -3.68594617e-01 8.86612460e-02 -3.78111511e-01 -2.58938335e-02 -2.13352636e-01 2.57711291e-01 7.52846897e-01 -1.39918268e-01 -1.69781029e-01 -8.61459792e-01 3.10708195e-01 1.99300438e-01 -2.43502975e-01 1.33169341e+00 -6.01048648e-01 -4.03520554e-01 5.43328524e-02 4.51178640e-01 -1.22000188e-01 -6.20895088e-01 1.53978601e-01 1.99092776e-01 -5.08419693e-01 -5.49468815e-01 -1.00923789e+00 -8.82556498e-01 -1.10680638e-02 -2.15773225e-01 8.93562317e-01 1.07710552e+00 2.48772249e-01 5.63777447e-01 6.35856688e-01 1.20154607e+00 -9.51382935e-01 -1.68199047e-01 4.54943299e-01 9.53963280e-01 -7.33441889e-01 -3.29780519e-01 -8.42681408e-01 -4.90448087e-01 1.23037255e+00 7.49680400e-01 3.89467776e-01 2.42219523e-01 2.58870542e-01 -3.31518710e-01 -1.25477746e-01 -8.69693339e-01 -8.15137923e-02 -6.08427644e-01 6.54080749e-01 -6.06251538e-01 4.63553518e-01 -6.34407640e-01 6.05180740e-01 -6.31222129e-02 -3.06684077e-01 1.22966719e+00 1.28914821e+00 -7.53169656e-01 -1.02514029e+00 -6.90313578e-01 1.68206811e-01 2.94277728e-01 4.48893666e-01 -4.71619010e-01 8.99965942e-01 1.57886252e-01 1.76454878e+00 1.80467233e-01 -2.07796782e-01 6.62877858e-01 -6.08964205e-01 4.16345388e-01 -4.11042124e-01 1.97321381e-02 -3.14231873e-01 4.22198653e-01 -4.21571344e-01 -4.46790814e-01 -8.14899027e-01 -1.74279761e+00 -5.71156681e-01 -2.99780101e-01 6.42158389e-01 9.46300924e-01 7.51156449e-01 3.36166710e-01 3.39791805e-01 8.13132763e-01 -5.03627956e-01 -3.28721702e-01 -1.35656372e-01 -6.99335515e-01 -3.72106761e-01 -3.37557984e-04 -7.23718464e-01 -1.94496378e-01 -5.17639816e-01]
[4.932080268859863, 1.7714234590530396]
1b5bb8af-5579-4be6-8dbe-ccd8dd2db950
blue-at-memotion-2-0-2022-you-have-my-image
2202.07543
null
https://arxiv.org/abs/2202.07543v3
https://arxiv.org/pdf/2202.07543v3.pdf
BLUE at Memotion 2.0 2022: You have my Image, my Text and my Transformer
Memes are prevalent on the internet and continue to grow and evolve alongside our culture. An automatic understanding of memes propagating on the internet can shed light on the general sentiment and cultural attitudes of people. In this work, we present team BLUE's solution for the second edition of the MEMOTION shared task. We showcase two approaches for meme classification (i.e. sentiment, humour, offensive, sarcasm and motivation levels) using a text-only method using BERT, and a Multi-Modal-Multi-Task transformer network that operates on both the meme image and its caption to output the final scores. In both approaches, we leverage state-of-the-art pretrained models for text (BERT, Sentence Transformer) and image processing (EfficientNetV4, CLIP). Through our efforts, we obtain first place in task A, second place in task B and third place in task C. In addition, our team obtained the highest average score for all three tasks.
['Ioan-Bogdan Iordache', 'Adrian Cosma', 'Ana-Maria Bucur']
2022-02-15
null
null
null
null
['meme-classification']
['natural-language-processing']
[-1.62655205e-01 -3.61831844e-01 3.45836997e-01 -1.81856185e-01 -6.52310133e-01 -5.81194520e-01 9.86377180e-01 4.36882943e-01 -6.80836320e-01 4.94906723e-01 6.14633918e-01 1.29039481e-01 6.21690989e-01 -6.50342643e-01 -3.81451786e-01 -1.02882415e-01 5.28181791e-01 3.29411566e-01 7.50428140e-02 -7.84927428e-01 8.09147239e-01 -3.86288613e-01 -1.26531005e+00 1.26251948e+00 5.39252043e-01 9.08326685e-01 -1.21692233e-01 1.11012042e+00 -4.67913270e-01 1.55166125e+00 -7.90847898e-01 -1.03947306e+00 -1.64923549e-01 -3.90815407e-01 -1.14838076e+00 1.19956449e-01 9.09602702e-01 -1.10229835e-01 -2.23915279e-01 8.93817723e-01 5.51535249e-01 3.74706127e-02 5.40591896e-01 -9.44160581e-01 -9.18814182e-01 7.88437724e-01 -5.17504334e-01 4.27262753e-01 5.69339037e-01 1.66859865e-01 1.00053155e+00 -1.20089638e+00 8.88971150e-01 1.32976580e+00 7.92313039e-01 4.03320312e-01 -1.02962351e+00 -4.29583073e-01 -1.96575329e-01 -2.11597141e-02 -9.26996887e-01 -3.70254099e-01 7.66999424e-01 -9.67701733e-01 1.01891077e+00 1.10701732e-01 6.91713929e-01 1.79328072e+00 3.14785570e-01 9.25917745e-01 1.67932725e+00 -1.94527492e-01 -2.33099908e-01 6.74094379e-01 1.68427929e-01 8.40649068e-01 -5.02993524e-01 -5.49608707e-01 -1.35675883e+00 4.35507037e-02 2.45071694e-01 -1.58291399e-01 8.08122978e-02 5.77625036e-01 -1.43647933e+00 8.75014007e-01 7.17036009e-01 3.81515294e-01 -2.04319015e-01 2.12989435e-01 8.28946829e-01 5.92926502e-01 1.17162824e+00 4.95615155e-01 -2.88504623e-02 -5.02510786e-01 -1.24937403e+00 3.98869365e-01 9.32294011e-01 7.13732660e-01 6.94748282e-01 -4.63138998e-01 -5.29332459e-01 1.24101353e+00 7.11077526e-02 5.45152307e-01 4.02300090e-01 -5.09215355e-01 8.41322362e-01 7.56130040e-01 3.23264189e-02 -1.44289494e+00 -5.94763458e-01 -6.01862073e-01 -5.77582300e-01 -1.63827866e-01 1.00393325e-01 -1.90211028e-01 -6.16805851e-01 1.44100726e+00 -1.29541010e-01 -2.59093106e-01 -3.55313003e-01 6.38842106e-01 1.32868350e+00 8.19046557e-01 5.72012030e-02 2.18935058e-01 1.48472106e+00 -1.36871481e+00 -5.89697599e-01 -5.11751711e-01 6.55423820e-01 -1.28490710e+00 1.32142925e+00 3.67229491e-01 -1.26779878e+00 -4.96020168e-01 -9.56222117e-01 -5.11723638e-01 -6.86925352e-01 1.18374094e-01 1.44678414e-01 3.87751073e-01 -1.37640381e+00 5.43822587e-01 -4.96243596e-01 -8.03449094e-01 4.90649551e-01 -2.02658512e-02 -2.16729343e-01 2.93926209e-01 -9.87461567e-01 1.19228423e+00 5.64793460e-02 -1.48948625e-01 -7.48042405e-01 -4.61458057e-01 -4.77297962e-01 -3.63265693e-01 -6.82551563e-02 -7.55834937e-01 1.21504140e+00 -1.29149091e+00 -1.31730914e+00 1.69914913e+00 8.37537693e-04 -1.98625624e-01 7.64384627e-01 -4.74666625e-01 -3.61376047e-01 4.39364538e-02 2.43024319e-01 8.41867566e-01 8.83646429e-01 -1.01535952e+00 -5.75865686e-01 -2.15914041e-01 1.49877250e-01 7.58989006e-02 -1.07923138e+00 6.05365157e-01 -1.66385189e-01 -5.65498114e-01 -2.96655446e-01 -7.92294264e-01 2.31273219e-01 -3.14443529e-01 -6.24799490e-01 -1.44542783e-01 5.55599332e-01 -1.01347613e+00 1.38699174e+00 -2.05483651e+00 3.89038295e-01 -1.53689161e-01 5.31442761e-01 -2.65886366e-01 -2.60731816e-01 7.64292777e-01 4.36177880e-01 3.64016533e-01 5.51314000e-03 -9.08786952e-01 5.32836393e-02 -2.43079692e-01 -1.37391120e-01 3.67615223e-01 1.25926539e-01 9.12219703e-01 -8.68306220e-01 -6.71387911e-01 -2.58488417e-01 3.84499609e-01 -6.65912986e-01 1.29047319e-01 -3.19157600e-01 3.16175878e-01 -3.16908419e-01 4.52656746e-01 4.90840346e-01 -5.89696944e-01 -1.28882885e-01 -3.22390571e-02 -5.60901940e-01 5.31743288e-01 -5.36498308e-01 1.55230451e+00 -7.90217280e-01 1.12053788e+00 2.53938317e-01 -2.08052248e-01 9.83755171e-01 2.55795389e-01 3.22379291e-01 -9.44089174e-01 4.23269331e-01 2.86404133e-01 -4.71947134e-01 -6.69694126e-01 1.09682322e+00 -2.17371419e-01 -3.96916091e-01 7.76190460e-01 1.67992756e-01 1.24603473e-02 4.85476792e-01 6.41515076e-01 1.01164341e+00 1.24123737e-01 4.76523228e-02 -4.97121215e-01 5.20351410e-01 1.61926359e-01 -2.51301497e-01 7.25581348e-01 -2.92849571e-01 5.93618751e-01 7.23611057e-01 -6.23888075e-01 -1.29451132e+00 -7.11824417e-01 1.71059638e-01 1.77061450e+00 -1.66858390e-01 -6.31520629e-01 -8.04516137e-01 -4.17190611e-01 -2.18886733e-01 4.03794140e-01 -9.41185772e-01 1.91082314e-01 -5.91717601e-01 -7.13997662e-01 3.86027366e-01 1.12517811e-01 7.47522652e-01 -1.30994081e+00 -5.05750835e-01 2.96597242e-01 -8.16196203e-01 -1.13350976e+00 -4.51649845e-01 -1.61862910e-01 -4.58211690e-01 -6.60936356e-01 -7.10006952e-01 -7.25040436e-01 2.68340498e-01 1.81861803e-01 1.67009342e+00 3.90516400e-01 1.42819270e-01 3.57595563e-01 -6.05197906e-01 -3.92839164e-01 -5.94950676e-01 5.01911044e-01 -3.31978202e-01 1.46982595e-01 4.34591532e-01 -3.56757700e-01 -4.35836941e-01 8.44764262e-02 -7.25508690e-01 4.70871180e-01 1.52875930e-01 7.00546145e-01 -1.98198646e-01 -5.49533963e-01 2.38699794e-01 -9.44420636e-01 9.32364464e-01 -7.94497967e-01 2.52053767e-01 -2.69513205e-02 -1.16572045e-02 -4.90421057e-01 4.79305208e-01 -1.63484260e-01 -9.09798682e-01 -3.80609304e-01 -1.37830123e-01 1.87404171e-01 2.33393863e-01 5.95916510e-01 7.89772153e-01 2.27455214e-01 1.03165162e+00 1.19787753e-01 -2.30162129e-01 -3.16518456e-01 2.84967363e-01 1.09780896e+00 3.99495572e-01 -4.75306600e-01 6.65907681e-01 6.69237018e-01 -5.39884448e-01 -9.22855139e-01 -1.59590471e+00 -6.33140326e-01 -6.65706754e-01 -6.99085653e-01 1.09223580e+00 -1.15364051e+00 -8.01530659e-01 7.22714901e-01 -1.47035956e+00 -2.94649184e-01 2.62318939e-01 -1.26589403e-01 -3.91459495e-01 -1.86367277e-02 -1.05773735e+00 -6.36221826e-01 -6.79839969e-01 -7.29693472e-01 9.99797106e-01 7.40798609e-03 -6.44701242e-01 -1.24606550e+00 2.30254650e-01 8.84958625e-01 6.54476404e-01 3.18655968e-01 5.70808291e-01 -6.47905350e-01 -2.96112835e-01 1.51964545e-01 -3.98958892e-01 2.15619266e-01 -4.13343400e-01 9.85450000e-02 -1.03739178e+00 -3.28934759e-01 1.56398453e-02 -9.99561369e-01 1.30730009e+00 4.90627848e-02 6.50807321e-01 -5.39354622e-01 3.04315500e-02 2.32883275e-01 1.40480208e+00 -5.92615843e-01 6.01182103e-01 8.14325571e-01 7.53725946e-01 8.33509207e-01 1.26768053e-01 7.75795817e-01 6.12878025e-01 5.36138296e-01 2.39925325e-01 -1.08412266e-01 -1.18320726e-01 -2.24600375e-01 8.73302877e-01 1.21494293e+00 -1.08678617e-01 -4.06224459e-01 -1.01644909e+00 5.76543808e-01 -1.84038413e+00 -1.31792545e+00 -6.85021758e-01 1.54130292e+00 8.52904975e-01 2.57888764e-01 5.80395699e-01 -3.01553220e-01 7.08862603e-01 5.08697510e-01 5.20740822e-02 -9.03944492e-01 -3.98690194e-01 -1.31573632e-01 1.05478838e-01 5.34362912e-01 -1.18319166e+00 1.04709160e+00 6.33104897e+00 5.32004535e-01 -1.22259319e+00 6.72066808e-01 6.81874275e-01 -5.30712426e-01 -2.57450312e-01 -3.19210291e-01 -7.13318586e-01 6.44958615e-01 9.90138054e-01 1.31142497e-01 4.61954474e-01 5.51883638e-01 1.06941000e-01 -6.26948401e-02 -1.04839027e+00 9.70241189e-01 6.30438149e-01 -1.53610349e+00 -7.39041641e-02 -1.45772994e-01 1.01899970e+00 3.97394180e-01 3.16231340e-01 5.06305397e-01 6.20739758e-02 -8.24786544e-01 1.33358991e+00 4.72963333e-01 4.42508757e-01 -5.15087247e-01 7.43713260e-01 1.92857474e-01 -6.78400993e-01 -6.58560619e-02 -1.37386411e-01 -4.03420985e-01 2.34499186e-01 5.78463793e-01 -5.26777327e-01 -1.29261175e-02 7.79486954e-01 1.12215137e+00 -1.04074478e+00 5.30381501e-01 -2.75997072e-01 5.40319502e-01 -1.22427724e-01 -6.21879220e-01 5.11197746e-01 2.02710088e-02 6.78554952e-01 1.91364133e+00 8.67076889e-02 -3.82199496e-01 3.47116381e-01 1.00430942e+00 -5.99361420e-01 4.49232399e-01 -2.92720318e-01 -3.70835572e-01 5.03107347e-02 1.62989914e+00 -8.40813160e-01 -5.32859802e-01 -4.55962986e-01 1.34509718e+00 6.70303702e-01 8.87049735e-02 -7.91134119e-01 -3.49337086e-02 6.79276884e-02 1.94460168e-01 -7.37310052e-02 -1.99619666e-01 -6.88093960e-01 -1.23365235e+00 -6.00454845e-02 -8.70493054e-01 3.57356071e-01 -8.29109669e-01 -1.66952538e+00 7.97039568e-01 -6.31796181e-01 -8.52692604e-01 3.45436603e-01 -7.67409861e-01 -7.99246907e-01 6.92367196e-01 -1.19438314e+00 -1.54412544e+00 -5.51382363e-01 5.29834747e-01 7.08661795e-01 -1.57222688e-01 4.60971743e-01 5.12677610e-01 -5.27371764e-01 3.35821748e-01 -1.10551432e-01 2.65539825e-01 1.06167650e+00 -1.18596196e+00 3.89725208e-01 5.30714095e-01 5.91981560e-02 4.69976246e-01 8.27578664e-01 -6.36656404e-01 -7.98559546e-01 -8.60214353e-01 1.30651832e+00 -9.71206307e-01 1.27334702e+00 -6.37994707e-01 -4.64834750e-01 6.82051361e-01 8.12950313e-01 -7.37517893e-01 8.00456703e-01 6.03303432e-01 -5.50794661e-01 4.36029494e-01 -8.12762439e-01 6.58879042e-01 8.05330396e-01 -7.39048362e-01 -5.80878139e-01 8.38942766e-01 4.27317500e-01 -3.40720862e-01 -8.56100321e-01 -4.80861902e-01 4.52357829e-01 -1.21067715e+00 4.89267200e-01 -6.73369348e-01 1.83537781e+00 1.35063767e-01 -5.54211028e-02 -1.30966496e+00 -4.22420532e-01 -5.61612964e-01 1.91116646e-01 1.34309137e+00 7.56946385e-01 -2.17650712e-01 4.91212338e-01 3.26601341e-02 -2.95708209e-01 -5.82872570e-01 -6.16014242e-01 -1.71485513e-01 2.69530833e-01 -2.43047595e-01 -1.74707219e-01 1.16567588e+00 1.87773243e-01 9.08792257e-01 -6.96454465e-01 -5.77236772e-01 4.32175696e-01 4.63720597e-03 7.75560379e-01 -1.12538648e+00 -4.53265011e-02 -8.26441348e-01 -1.37876570e-01 -6.61304772e-01 -8.98903608e-03 -1.13719440e+00 -1.44166142e-01 -1.72335088e+00 1.03041458e+00 1.03888005e-01 -2.29605407e-01 3.42297673e-01 -3.96258943e-02 8.63508940e-01 4.33756530e-01 3.71402830e-01 -1.01951396e+00 2.16076642e-01 1.29947162e+00 -2.88232386e-01 -6.61390647e-02 -4.70668763e-01 -8.30583692e-01 7.55256295e-01 8.03669274e-01 -5.50441027e-01 2.72698522e-01 -6.09205484e-01 1.34734011e+00 -4.63160217e-01 6.05914354e-01 -9.29550529e-01 2.04393119e-01 3.64948772e-02 3.26977491e-01 -7.04837978e-01 5.39063632e-01 -6.30539432e-02 -5.19868970e-01 3.53999078e-01 -6.37710631e-01 1.42475739e-01 -5.39534464e-02 7.89316520e-02 -2.39113063e-01 -9.21596810e-02 1.09821093e+00 -4.37655479e-01 -3.59568179e-01 1.07313199e-02 -5.43071747e-01 3.11561376e-01 5.46216071e-01 -1.48727730e-01 -1.06281233e+00 -5.36835432e-01 -8.10085595e-01 1.74812347e-01 6.01776659e-01 6.42118156e-01 4.51112479e-01 -1.15589154e+00 -1.40032339e+00 -3.64543259e-01 1.94862366e-01 -8.65092635e-01 4.49597418e-01 1.14092195e+00 -7.00025022e-01 9.74568054e-02 -5.39850295e-01 -4.29399699e-01 -1.16588867e+00 2.35499233e-01 2.54351377e-01 -3.98464590e-01 -3.90933931e-01 1.10349226e+00 -2.03965127e-01 -4.97103065e-01 -2.74485081e-01 3.45940530e-01 -4.38145816e-01 5.76188207e-01 7.67790556e-01 6.09426856e-01 -5.49797714e-02 -9.69139159e-01 -2.72473603e-01 4.50878531e-01 -2.58621424e-01 -3.50909710e-01 1.51993716e+00 -2.88964957e-01 -7.15746939e-01 8.61603737e-01 1.43694043e+00 9.79978815e-02 -6.71912968e-01 -4.48834077e-02 -1.37957186e-02 -4.32913512e-01 1.59761116e-01 -1.05418313e+00 -7.35404909e-01 1.13669312e+00 2.30764747e-01 6.00555360e-01 6.36818349e-01 7.07159936e-02 1.11608291e+00 3.58382970e-01 4.15913574e-02 -1.82279360e+00 9.02411342e-01 1.05144966e+00 1.17202878e+00 -1.33803475e+00 1.66337237e-01 -1.35990620e-01 -8.72599840e-01 1.17895341e+00 7.44964957e-01 -2.21216138e-02 2.94504672e-01 2.24334877e-02 3.14260274e-01 -6.11168265e-01 -9.54830885e-01 6.24651089e-02 3.06923479e-01 -9.69147682e-03 1.10942876e+00 -7.04717115e-02 -4.30251896e-01 2.38455549e-01 -6.27284408e-01 -2.58194715e-01 1.00621319e+00 7.90861607e-01 -7.54837453e-01 -6.02843642e-01 -3.30649853e-01 4.01440799e-01 -7.32959449e-01 -3.20210457e-01 -9.56966996e-01 3.29655826e-01 1.41266584e-01 1.19288301e+00 2.47785091e-01 -7.07865655e-01 1.58697531e-01 3.70287262e-02 3.76158655e-01 -7.90559113e-01 -1.51354086e+00 -1.07363567e-01 4.96190935e-01 -4.04962629e-01 -5.64653218e-01 -4.31068689e-01 -8.59416187e-01 -7.89764106e-01 8.70787352e-02 -2.33044744e-01 9.66377676e-01 8.92516851e-01 2.23491162e-01 4.87337798e-01 6.00983918e-01 -8.28361571e-01 -3.04341376e-01 -1.40996838e+00 -3.46324056e-01 7.51306474e-01 8.74041319e-02 -2.54912436e-01 -3.76062393e-01 2.96679378e-01]
[8.525493621826172, 10.725275993347168]
55859447-726b-4225-ac62-1e35a751873a
structure-aware-dropedge-towards-deep-graph
2306.12091
null
https://arxiv.org/abs/2306.12091v1
https://arxiv.org/pdf/2306.12091v1.pdf
Structure-Aware DropEdge Towards Deep Graph Convolutional Networks
It has been discovered that Graph Convolutional Networks (GCNs) encounter a remarkable drop in performance when multiple layers are piled up. The main factor that accounts for why deep GCNs fail lies in over-smoothing, which isolates the network output from the input with the increase of network depth, weakening expressivity and trainability. In this paper, we start by investigating refined measures upon DropEdge -- an existing simple yet effective technique to relieve over-smoothing. We term our method as DropEdge++ for its two structure-aware samplers in contrast to DropEdge: layer-dependent sampler and feature-dependent sampler. Regarding the layer-dependent sampler, we interestingly find that increasingly sampling edges from the bottom layer yields superior performance than the decreasing counterpart as well as DropEdge. We theoretically reveal this phenomenon with Mean-Edge-Number (MEN), a metric closely related to over-smoothing. For the feature-dependent sampler, we associate the edge sampling probability with the feature similarity of node pairs, and prove that it further correlates the convergence subspace of the output layer with the input features. Extensive experiments on several node classification benchmarks, including both full- and semi- supervised tasks, illustrate the efficacy of DropEdge++ and its compatibility with a variety of backbones by achieving generally better performance over DropEdge and the no-drop version.
['Junzhou Huang', 'Fuchun Sun', 'Tingyang Xu', 'Yu Rong', 'Wenbing Huang', 'Jiaqi Han']
2023-06-21
null
null
null
null
['node-classification']
['graphs']
[-3.1570829e-02 4.4081894e-01 -8.9243777e-02 -2.3301847e-01 -9.9037908e-02 -6.2252462e-01 6.1180198e-01 1.6949734e-01 -3.9973611e-01 4.9051017e-01 1.2813629e-01 -2.9419506e-01 -3.5186791e-01 -8.9158416e-01 -8.4195995e-01 -7.8836560e-01 -5.4271913e-01 1.5636626e-01 5.3099465e-01 -2.2896917e-01 -5.4473970e-02 5.8589023e-01 -1.2719381e+00 1.7566077e-01 8.4678262e-01 9.6791953e-01 -8.4612429e-02 5.4027754e-01 -2.3056051e-01 4.9091783e-01 -3.8255960e-01 -6.8781286e-01 5.0104487e-01 -3.8030180e-01 -6.8369508e-01 -1.7621104e-01 6.7243207e-01 4.9369764e-02 -4.4696182e-01 1.1638610e+00 3.3655593e-01 -1.3666335e-01 4.6171188e-01 -1.4311253e+00 -4.0839815e-01 9.7181606e-01 -5.7831377e-01 3.5084215e-01 -1.3621025e-01 4.5206130e-02 1.2773440e+00 -9.7867996e-01 7.1444839e-01 1.2208277e+00 1.2503707e+00 4.2667705e-01 -1.4365038e+00 -5.3103405e-01 3.3494616e-01 -9.5142648e-02 -1.4698318e+00 -2.6654541e-01 8.2286739e-01 -2.2949186e-01 8.3338141e-01 6.2386341e-02 6.0632467e-01 8.3754712e-01 9.5084809e-02 9.0811086e-01 6.8954915e-01 -1.7725006e-01 1.8767175e-01 5.6061286e-02 5.0039220e-01 9.8396230e-01 6.6321820e-01 -2.3660203e-02 -5.7861513e-01 -1.0837347e-01 7.9440713e-01 -2.1745615e-02 -4.3931523e-01 -7.5448138e-01 -6.9625103e-01 6.2394893e-01 8.5280210e-01 3.3990014e-01 -3.0440727e-01 3.2045192e-01 5.2829200e-01 5.9524673e-01 6.5764886e-01 2.7648064e-01 -5.5607611e-01 1.1988843e-01 -9.8977113e-01 5.6885727e-02 1.1283853e+00 1.0162617e+00 9.8232788e-01 3.0622259e-02 -2.3355794e-01 5.4148126e-01 -6.4434917e-03 -4.4777423e-02 9.8683015e-02 -5.4021931e-01 2.7613008e-01 9.8223084e-01 -6.0840303e-01 -9.2938280e-01 -5.7836163e-01 -1.1018214e+00 -1.2390981e+00 2.2095242e-01 7.0643568e-01 -1.8046987e-01 -7.5867683e-01 2.0516415e+00 1.3673341e-01 3.8634098e-01 -2.9085124e-01 6.0690194e-01 8.6945307e-01 1.3995376e-01 2.9789111e-02 -4.9673982e-02 1.1935468e+00 -9.1311765e-01 -3.5023183e-01 -9.0161093e-02 8.6240798e-01 -1.3891907e-01 1.1006094e+00 3.7689579e-01 -1.0067555e+00 -4.0824872e-01 -1.1637665e+00 1.2667288e-01 -5.2196556e-01 -6.0288265e-02 7.7607143e-01 6.8338138e-01 -1.4910687e+00 1.2225398e+00 -8.4372431e-01 -3.1070083e-01 7.7814168e-01 4.0783894e-01 -5.1909351e-01 9.9247538e-02 -9.1687256e-01 4.8968539e-01 2.6436406e-01 -4.8777377e-03 -6.1109751e-01 -1.0874554e+00 -6.7285383e-01 4.6324876e-01 4.5301825e-01 -6.7737842e-01 7.7528667e-01 -8.9825624e-01 -1.2478331e+00 7.1743125e-01 -1.5602365e-02 -6.7274791e-01 6.9805151e-01 -1.7322305e-01 -1.3732749e-01 -1.8690313e-03 -2.9991302e-01 4.8422503e-01 9.5381403e-01 -1.1218076e+00 -3.7800482e-01 -4.3575847e-01 7.9508133e-02 -1.2254440e-02 -7.3483938e-01 -4.9178150e-01 -5.1926082e-01 -3.8415286e-01 1.1614368e-01 -7.0589501e-01 -1.4108053e-01 4.1567195e-02 -5.0269485e-01 -4.8627180e-01 8.5608357e-01 -6.9376498e-02 1.3197860e+00 -2.2472832e+00 1.0873597e-01 3.3136502e-01 9.4019473e-01 4.1385576e-01 -3.6300412e-01 5.0052571e-01 -2.8900662e-01 2.6159629e-01 -2.6852763e-01 -6.3072354e-01 -8.7987736e-02 3.2629612e-01 -8.7209269e-02 6.9420725e-01 4.5875266e-01 1.0638580e+00 -1.1180673e+00 -3.4994683e-01 -2.6935220e-03 5.2343369e-01 -7.1802264e-01 -2.8675598e-01 -5.8173048e-03 -5.9015296e-02 -2.6480758e-01 3.7716481e-01 9.1292214e-01 -5.6979567e-01 2.0773843e-01 -2.2220229e-01 1.2161413e-01 4.2781094e-01 -1.3215486e+00 1.4424052e+00 -2.5216630e-01 7.5686401e-01 4.0320912e-01 -8.7406534e-01 7.5890273e-01 -1.5250335e-02 2.5559267e-01 -4.8171294e-01 1.4223823e-01 1.6573192e-01 1.6200875e-01 -1.9760045e-01 5.8611578e-01 1.5347126e-01 2.5109309e-01 2.4578774e-01 4.0268695e-01 4.5080832e-01 4.2920709e-01 5.6268287e-01 1.3584367e+00 5.0463822e-02 1.2582433e-01 -9.6808308e-01 3.7729448e-01 -5.7551557e-01 4.1153336e-01 9.6715724e-01 -1.6407718e-01 4.5331028e-01 1.1300758e+00 -3.5960260e-01 -7.3316640e-01 -1.0410885e+00 -1.7907313e-01 1.1785935e+00 7.3165327e-02 -8.3395022e-01 -8.8414061e-01 -9.6835738e-01 3.3375496e-01 2.6134154e-01 -8.1330508e-01 -3.4213281e-01 -5.8148682e-01 -7.8574473e-01 7.8798038e-01 6.6556245e-01 4.2503211e-01 -8.3414990e-01 -2.2828655e-01 9.5439553e-02 3.4810686e-01 -9.2807955e-01 -3.5780764e-01 5.3689808e-01 -1.1210457e+00 -1.1338166e+00 -5.4957533e-01 -8.2454193e-01 8.0829108e-01 2.1889144e-01 1.4570920e+00 6.3193613e-01 -1.1665115e-01 -8.5326219e-03 -2.4894622e-01 -1.2973015e-01 -2.7404472e-01 7.1843690e-01 -9.8275922e-02 7.6777324e-02 2.4749303e-01 -9.6129698e-01 -6.7702037e-01 2.8442100e-02 -9.9947035e-01 -3.7389569e-02 6.1103714e-01 6.5111756e-01 1.3432169e-01 -1.2296921e-01 4.8507366e-01 -1.2639278e+00 8.0884510e-01 -3.3728653e-01 -3.5809836e-01 1.4084227e-01 -7.6673311e-01 4.0051416e-01 9.2610556e-01 -2.8715229e-01 -2.8085974e-01 -1.7867337e-01 -9.4808094e-02 -4.7064215e-01 1.4667551e-01 3.7788236e-01 -6.4669698e-02 -2.3217866e-01 6.9775021e-01 1.3142970e-02 -3.9149936e-02 -5.1347625e-01 3.5547379e-01 1.2811916e-01 5.4831809e-01 -4.1525126e-01 1.0072838e+00 5.0832546e-01 2.1664201e-01 -9.4022399e-01 -7.9662800e-01 -3.6550048e-01 -5.7629251e-01 -2.1607959e-01 4.1454384e-01 -5.8449763e-01 -8.5532677e-01 5.6607527e-01 -9.8375481e-01 -4.4779605e-01 -4.7550020e-01 1.1192828e-01 -1.3533781e-02 2.5147250e-01 -6.6924816e-01 -6.7348760e-01 -3.0869597e-01 -8.9094317e-01 9.6018296e-01 2.7646551e-01 7.0299767e-02 -1.3461753e+00 -1.2122744e-01 -5.9843355e-01 5.3758258e-01 2.6672620e-01 9.5401627e-01 -9.2086375e-01 -4.5215145e-01 -1.5956657e-01 -4.5196158e-01 4.3358773e-01 -2.7123739e-03 -5.4705705e-02 -9.7166741e-01 -6.0708809e-01 -2.5090206e-01 2.6265392e-02 1.3572928e+00 3.1266892e-01 1.1691419e+00 -1.5582642e-01 -3.9192432e-01 9.9411905e-01 1.6332047e+00 -6.7837179e-01 6.5156132e-01 3.8124286e-02 8.2996881e-01 3.6230353e-01 -2.7986291e-01 2.3618956e-01 5.7028022e-02 2.7190280e-01 6.8721986e-01 -4.1922852e-01 -2.9611099e-01 -4.6706817e-01 2.0182069e-01 7.6604718e-01 -7.2276771e-02 -4.1345814e-01 -6.0663980e-01 4.6885887e-01 -1.7307341e+00 -6.4156115e-01 -4.4413781e-01 2.1859081e+00 7.3400640e-01 5.5909479e-01 3.2243383e-01 2.3277442e-01 5.5587482e-01 4.3665120e-01 -4.5342234e-01 -1.7825893e-01 -2.6173809e-01 4.0325737e-01 7.9157853e-01 5.0760782e-01 -9.0724778e-01 9.0289181e-01 6.2091374e+00 8.8667971e-01 -9.7831082e-01 -9.7359613e-02 5.6430912e-01 7.6903433e-02 -3.9207640e-01 1.3073576e-02 -9.0781945e-01 2.8107598e-01 7.1026522e-01 -1.9115035e-01 3.8183287e-01 8.5166293e-01 -1.3588923e-01 2.2559190e-01 -1.5109494e+00 5.5820793e-01 -1.2848125e-01 -1.4925551e+00 4.4809196e-02 1.8409096e-01 6.3050050e-01 3.5404778e-01 -1.0098182e-01 3.0557147e-01 4.3551859e-01 -8.8744354e-01 6.2943077e-01 3.2781646e-01 7.5151974e-01 -7.5832242e-01 6.0710472e-01 2.9774079e-01 -1.4985608e+00 1.7954028e-01 -3.4315255e-01 -1.5240803e-01 -1.3743502e-01 1.0059822e+00 -6.9428098e-01 7.3058599e-01 6.1367798e-01 8.6712050e-01 -8.1148016e-01 1.0661315e+00 -1.3218382e-01 6.7670059e-01 -4.5756555e-01 -1.3498260e-01 3.0102170e-01 -2.7472633e-01 7.3736268e-01 1.4234874e+00 -1.7388187e-02 -3.5508829e-01 4.4434588e-02 1.0283338e+00 -4.4277078e-01 -1.4787999e-01 -6.7884380e-01 4.5988124e-02 5.3503281e-01 1.4916348e+00 -1.0511880e+00 -2.7601153e-01 -2.8020900e-01 8.0789018e-01 7.3711038e-01 2.4576466e-01 -6.5748107e-01 -5.3510439e-01 6.3865143e-01 2.9963425e-01 5.2220213e-01 -2.4931808e-01 -3.4056225e-01 -8.0135697e-01 7.4716277e-02 -6.3211900e-01 2.0790844e-01 -8.2119897e-02 -1.3498652e+00 7.2165656e-01 -1.9678178e-01 -8.5812467e-01 1.4235722e-01 -7.7811700e-01 -7.4260551e-01 6.8222558e-01 -1.6067699e+00 -7.3807663e-01 -4.5076632e-01 4.5943633e-01 1.8576506e-01 1.7309654e-01 4.5951906e-01 3.8070449e-01 -7.3355502e-01 1.0655272e+00 -2.1492125e-02 2.5406697e-01 3.0775839e-01 -1.3929946e+00 7.8445065e-01 8.9289463e-01 2.7000758e-01 7.4978817e-01 6.2084591e-01 -5.0720137e-01 -1.4318559e+00 -1.0961174e+00 7.8034782e-01 -4.2105556e-01 8.1334692e-01 -6.9338739e-01 -1.1027467e+00 4.8304811e-01 1.0677182e-01 2.8695711e-01 2.8443837e-01 3.7757522e-01 -5.3997099e-01 -2.1776509e-01 -9.1327310e-01 6.5390646e-01 1.6915272e+00 -5.2070856e-01 -3.0301002e-01 6.6001751e-02 7.8853261e-01 -1.2410834e-01 -6.9794124e-01 4.2517096e-01 4.2409280e-01 -1.2931261e+00 7.7699667e-01 -6.5209419e-01 2.7139869e-01 -1.0896788e-02 2.5167984e-01 -1.3497519e+00 -5.6683779e-01 -1.0158210e+00 -1.6980048e-01 1.2353472e+00 6.0552406e-01 -8.3065695e-01 1.2645465e+00 8.3592348e-02 -3.5021064e-01 -9.9904042e-01 -8.5339689e-01 -9.4810784e-01 2.2759247e-01 -2.7728003e-01 5.0922525e-01 8.1952107e-01 -1.1492834e-01 2.2215767e-01 5.3752404e-02 5.8470018e-02 6.3050348e-01 -2.8083917e-01 7.0225298e-01 -1.6479143e+00 -2.7933556e-01 -8.7243271e-01 -3.1269479e-01 -1.2870295e+00 -1.0517711e-02 -1.3474908e+00 -1.4266616e-01 -1.3170328e+00 8.1735760e-02 -4.0352422e-01 -3.3592498e-01 5.5159903e-01 -2.4438608e-01 1.8528672e-01 6.5039299e-02 1.1376859e-01 -5.8610296e-01 3.6506420e-01 1.2116404e+00 2.6397920e-01 -4.3618459e-01 -1.9281514e-02 -7.5009739e-01 8.1021768e-01 6.4271206e-01 -5.9617507e-01 -4.0695262e-01 -2.9450920e-01 6.4641023e-01 -5.1062155e-01 4.7848549e-01 -1.0956659e+00 2.9275173e-01 4.8247033e-01 9.4222084e-02 -4.3207783e-01 -1.2984416e-01 -7.5199294e-01 -1.1281017e-01 6.6074371e-01 -4.4178703e-01 1.6291466e-01 2.1459390e-01 6.4653945e-01 1.8263894e-01 -1.3669778e-01 6.6909772e-01 7.6412492e-02 -4.7986254e-01 4.8149428e-01 -3.3083793e-02 3.5858250e-01 6.3541085e-01 -2.7686512e-01 -5.9684473e-01 -1.3349840e-01 -5.9178931e-01 3.4452716e-01 3.7717083e-01 3.5132197e-01 3.7379906e-01 -1.2185955e+00 -7.6386791e-01 6.4475137e-01 -4.7635064e-02 3.3078127e-02 -2.7682824e-02 1.2019958e+00 -4.2786080e-01 5.2700929e-02 1.3117547e-01 -6.7475444e-01 -1.0744914e+00 2.9998901e-01 5.1402235e-01 -5.2504182e-01 -1.0380832e+00 1.1065278e+00 2.9884112e-01 -2.5144070e-01 6.6695756e-01 -5.7175076e-01 4.0575165e-02 5.1187154e-02 3.4467965e-01 4.3989971e-01 3.7147295e-01 -1.2718098e-01 -4.2831874e-01 1.9482943e-01 -1.7600228e-01 4.2605150e-01 1.4755982e+00 2.2276418e-01 -1.7640670e-01 3.7565321e-01 1.3090452e+00 -9.0532295e-02 -1.4959773e+00 -4.8498917e-01 1.8880418e-01 -7.9458371e-02 -3.8059618e-02 -2.8652212e-01 -1.4665941e+00 7.7483392e-01 2.1744812e-01 8.6630309e-01 9.5442778e-01 1.4931223e-01 6.2145907e-01 3.7317136e-01 2.3849040e-01 -9.0562940e-01 -3.8498215e-02 5.8394700e-01 6.4606261e-01 -7.7814704e-01 -8.6532466e-02 -6.2030977e-01 -6.1838374e-02 9.1362685e-01 5.5002868e-01 -6.4417243e-01 1.0146104e+00 7.1425217e-01 -4.4075716e-01 -4.3358761e-01 -8.3055794e-01 -2.7029526e-01 2.0828313e-01 4.4980779e-01 3.9795938e-01 -2.7737906e-02 -1.9777234e-01 5.8820921e-01 -3.3333322e-01 -7.7474654e-02 3.8299474e-01 5.9282219e-01 -4.7428235e-01 -8.8874501e-01 2.5287887e-01 6.5377426e-01 -3.0696207e-01 -3.6295256e-01 -7.1768326e-01 1.1375610e+00 -5.0355629e-03 5.4640096e-01 2.8597161e-01 -5.9129542e-01 2.8443992e-01 9.4302915e-02 3.9211890e-01 -4.7005567e-01 -9.2417878e-01 -2.4501096e-01 1.1164808e-01 -7.1822107e-01 -2.3062229e-01 -3.7997940e-01 -1.1765822e+00 -5.3530860e-01 -5.9731072e-01 6.4595535e-02 3.9755848e-01 7.4319685e-01 5.4648626e-01 8.1975973e-01 4.8737237e-01 -7.5404775e-01 -7.8341532e-01 -8.9422977e-01 -7.4559623e-01 4.7229031e-01 5.1750284e-01 -4.3081748e-01 -8.9415294e-01 -5.1863807e-01]
[6.830257892608643, 6.037235260009766]
c1a57908-5c6c-4c64-859f-faf588463968
dual-path-convolutional-image-text-embedding
1711.05535
null
https://arxiv.org/abs/1711.05535v4
https://arxiv.org/pdf/1711.05535v4.pdf
Dual-Path Convolutional Image-Text Embeddings with Instance Loss
Matching images and sentences demands a fine understanding of both modalities. In this paper, we propose a new system to discriminatively embed the image and text to a shared visual-textual space. In this field, most existing works apply the ranking loss to pull the positive image / text pairs close and push the negative pairs apart from each other. However, directly deploying the ranking loss is hard for network learning, since it starts from the two heterogeneous features to build inter-modal relationship. To address this problem, we propose the instance loss which explicitly considers the intra-modal data distribution. It is based on an unsupervised assumption that each image / text group can be viewed as a class. So the network can learn the fine granularity from every image/text group. The experiment shows that the instance loss offers better weight initialization for the ranking loss, so that more discriminative embeddings can be learned. Besides, existing works usually apply the off-the-shelf features, i.e., word2vec and fixed visual feature. So in a minor contribution, this paper constructs an end-to-end dual-path convolutional network to learn the image and text representations. End-to-end learning allows the system to directly learn from the data and fully utilize the supervision. On two generic retrieval datasets (Flickr30k and MSCOCO), experiments demonstrate that our method yields competitive accuracy compared to state-of-the-art methods. Moreover, in language based person retrieval, we improve the state of the art by a large margin. The code has been made publicly available.
['Yi-Dong Shen', 'Mingliang Xu', 'Michael Garrett', 'Yi Yang', 'Zhedong Zheng', 'Liang Zheng']
2017-11-15
null
null
null
null
['person-retrieval', 'nlp-based-person-retrival']
['computer-vision', 'computer-vision']
[ 1.87723543e-02 -2.86362022e-01 -3.85544837e-01 -6.12333536e-01 -7.54972935e-01 -3.57458621e-01 6.70341790e-01 -3.80145595e-03 -6.56789541e-01 3.46822381e-01 1.59658432e-01 1.32317320e-01 -6.59204572e-02 -6.31595790e-01 -6.86175585e-01 -6.49394214e-01 3.74525398e-01 3.68046016e-01 -9.74329095e-03 -5.96718863e-02 -1.24756284e-01 1.16435885e-01 -1.42240763e+00 4.33928192e-01 6.24281526e-01 1.19669831e+00 3.26949239e-01 2.41483420e-01 -9.05642360e-02 5.73351681e-01 -1.94625765e-01 -6.07310891e-01 3.80394161e-01 -2.51800716e-01 -4.57104564e-01 9.42450669e-03 8.84310782e-01 -4.46360052e-01 -7.56128430e-01 1.02100265e+00 8.47718954e-01 8.94018561e-02 6.42846286e-01 -1.35575640e+00 -1.05997086e+00 3.59617263e-01 -7.71925747e-01 -7.71212876e-02 1.71183527e-01 4.68320549e-02 1.25834036e+00 -1.09741998e+00 3.72005641e-01 1.28454196e+00 4.35320586e-01 5.02626061e-01 -1.00918210e+00 -7.46829867e-01 3.64141047e-01 3.34281474e-01 -1.69358802e+00 -2.64422268e-01 1.03367591e+00 -4.79012400e-01 5.49074292e-01 2.52948076e-01 5.83069921e-01 1.04622793e+00 -1.20047607e-01 1.18950784e+00 1.04349577e+00 -4.24587727e-01 -2.62302130e-01 4.25119758e-01 1.43852875e-01 7.12230802e-01 5.77078946e-02 -8.66953433e-02 -4.17637885e-01 6.62283450e-02 5.74708104e-01 6.19834304e-01 -4.30240870e-01 -7.06758320e-01 -9.95917380e-01 7.50211239e-01 7.44062722e-01 3.28139156e-01 -1.83753520e-02 2.76068866e-01 4.63640898e-01 1.76487863e-01 4.03282106e-01 -1.88176498e-01 -1.65751338e-01 1.44429505e-01 -1.01917315e+00 1.61367744e-01 4.32410270e-01 8.85402799e-01 9.43274975e-01 -5.41351736e-01 -4.32067186e-01 1.07847095e+00 6.55446887e-01 7.27490187e-01 4.96464401e-01 -3.70056212e-01 8.13983202e-01 7.24901438e-01 -7.35882670e-02 -1.34803724e+00 -9.36590359e-02 -3.33281606e-01 -9.66746449e-01 -2.07846418e-01 1.84807077e-01 1.40446827e-01 -9.82486010e-01 1.72630680e+00 9.32230502e-02 1.42471477e-01 -1.54578730e-01 1.22931421e+00 8.24783921e-01 6.25747383e-01 -1.07941650e-01 8.27810466e-02 1.42565966e+00 -1.28882921e+00 -6.76571906e-01 -2.93916225e-01 4.19838518e-01 -6.97098553e-01 1.28661883e+00 1.75253421e-01 -8.82084131e-01 -6.98636472e-01 -9.31453824e-01 -4.15663898e-01 -5.95162690e-01 5.92880607e-01 4.15222466e-01 5.00550449e-01 -8.50062013e-01 2.43231952e-01 -7.11193740e-01 -3.63669068e-01 2.89732099e-01 2.35488653e-01 -4.94004458e-01 -5.34266174e-01 -1.28058517e+00 5.68396628e-01 3.79274577e-01 2.02635154e-01 -6.43541634e-01 -4.64005083e-01 -8.71707559e-01 7.93043673e-02 4.04840946e-01 -7.81013548e-01 7.68557847e-01 -1.15959108e+00 -1.27713001e+00 1.07081783e+00 -7.57397041e-02 -2.24328637e-01 7.09890902e-01 -3.00620705e-01 -3.11474741e-01 3.21871161e-01 1.18738256e-01 7.26667225e-01 9.68540430e-01 -1.30235767e+00 -5.30656278e-01 -5.67128062e-01 3.03226650e-01 2.78009593e-01 -9.93202090e-01 -1.07407965e-01 -1.13226998e+00 -7.57604659e-01 -1.87857121e-01 -9.33208048e-01 1.44866124e-01 3.39973539e-01 -3.59829068e-01 -4.41303194e-01 7.23966599e-01 -6.41917288e-01 1.26840150e+00 -2.28918862e+00 2.28388965e-01 2.05100462e-01 2.03979835e-01 1.63152918e-01 -2.33657613e-01 5.05628347e-01 -2.92384382e-02 2.52925605e-03 -3.23345438e-02 -7.66463995e-01 3.70177895e-01 9.41406116e-02 -2.17355773e-01 5.64579487e-01 -5.87268127e-03 1.00303674e+00 -7.40907669e-01 -6.69246912e-01 2.16487885e-01 7.65847206e-01 -5.23500979e-01 2.23515838e-01 1.74716264e-01 -1.61590222e-02 -4.77458090e-01 4.28679645e-01 8.18770468e-01 -3.96238506e-01 4.46383432e-02 -5.66199601e-01 2.41277814e-01 -1.81087613e-01 -1.14634168e+00 1.99862766e+00 -5.06923437e-01 5.15779078e-01 -1.73468217e-01 -1.14194942e+00 7.48297811e-01 9.84689221e-02 2.83517480e-01 -7.78914928e-01 1.86344981e-02 1.14950687e-01 -4.97928292e-01 -5.37994385e-01 4.94969964e-01 5.97676309e-03 -1.53084904e-01 1.64914608e-01 9.56713967e-03 4.66487437e-01 7.91621357e-02 2.61627644e-01 6.01532638e-01 9.33337063e-02 -1.41224161e-01 1.17289655e-01 6.65548384e-01 -4.25016850e-01 4.54853892e-01 6.74973011e-01 9.46019869e-03 8.39910984e-01 2.74643272e-01 -2.72688568e-01 -8.95937979e-01 -1.13109767e+00 -1.23387374e-01 1.19564223e+00 4.90120709e-01 -5.24286866e-01 -4.79140371e-01 -8.50462615e-01 1.27277628e-01 1.21794254e-01 -7.09148288e-01 -2.31352866e-01 -4.19347942e-01 -4.59661752e-01 4.13820952e-01 5.99526346e-01 7.15405405e-01 -6.84568942e-01 -5.97942919e-02 -1.71634838e-01 -3.21592212e-01 -1.06320000e+00 -9.01543915e-01 -2.81159997e-01 -3.65776390e-01 -8.83799136e-01 -1.09420061e+00 -9.82765377e-01 7.65678525e-01 4.94691491e-01 8.47617984e-01 1.69357270e-01 -3.85060698e-01 6.68323755e-01 -4.47368860e-01 -1.23921372e-02 3.18871170e-01 1.13009527e-01 -1.06425032e-01 4.99044448e-01 5.13788044e-01 -3.02707195e-01 -9.18241382e-01 2.70371050e-01 -1.10308468e+00 -3.08311563e-02 6.20865524e-01 1.07474613e+00 7.84776092e-01 1.08709335e-01 2.66594708e-01 -5.16385794e-01 3.98465097e-01 -2.97303170e-01 -3.38132799e-01 5.94786465e-01 -4.96035039e-01 -5.49056791e-02 6.49805307e-01 -4.78156269e-01 -7.39403248e-01 1.33097813e-01 1.58804029e-01 -8.44838202e-01 1.62711386e-02 6.86733961e-01 -4.23467845e-01 9.75835621e-02 1.35791272e-01 3.03071916e-01 -3.16406414e-02 -5.50728738e-01 5.78983605e-01 8.18723083e-01 3.25522006e-01 -5.38386941e-01 9.38259363e-01 5.50047457e-01 -3.19481164e-01 -7.52663553e-01 -8.54640543e-01 -6.99518979e-01 -4.85149205e-01 -6.99561164e-02 9.18480098e-01 -1.23954999e+00 -4.51163083e-01 3.86344135e-01 -9.29577410e-01 -1.17481582e-01 4.28214446e-02 5.48108637e-01 -2.55655110e-01 5.83455920e-01 -4.19216990e-01 -5.59497714e-01 -2.69490391e-01 -1.05257106e+00 1.34959686e+00 2.51335829e-01 3.71681154e-01 -9.80700731e-01 -1.00718010e-02 5.21539927e-01 2.09609360e-01 -1.91566497e-02 6.11121237e-01 -5.73132873e-01 -6.07945800e-01 -4.41571862e-01 -5.41773558e-01 4.98169005e-01 5.68218455e-02 -1.78030133e-01 -9.21967506e-01 -7.30822623e-01 -3.59650671e-01 -5.82698107e-01 1.36260068e+00 4.65169325e-02 1.35417271e+00 -3.07875127e-01 -4.18308020e-01 6.84615433e-01 1.59180272e+00 -4.44632918e-01 7.09966660e-01 2.77901560e-01 1.04035807e+00 6.34289682e-01 6.16219938e-01 3.41333896e-01 6.93979800e-01 9.66755927e-01 2.77851194e-01 -2.44403303e-01 -1.65433288e-02 -4.97202426e-01 4.03616905e-01 6.55078709e-01 2.41325989e-01 -3.80062312e-01 -7.25840807e-01 5.26599348e-01 -2.08400774e+00 -1.01553452e+00 3.44979823e-01 2.18694997e+00 8.53459716e-01 -1.18863866e-01 2.40752771e-02 -1.35055602e-01 7.48591483e-01 3.79970670e-01 -3.92764509e-01 1.74911857e-01 -9.81566608e-02 -1.33147702e-01 3.43898833e-01 3.44344914e-01 -1.27606380e+00 8.97020459e-01 4.33061981e+00 1.15235686e+00 -1.38247883e+00 4.47648913e-02 5.64478159e-01 -3.44391763e-01 -3.91164958e-01 -1.58962503e-01 -9.77032840e-01 6.01039231e-01 3.92511249e-01 -1.32294400e-02 3.27568203e-01 7.09747374e-01 -4.05091001e-03 3.24141979e-01 -1.22883415e+00 1.51971960e+00 5.53049326e-01 -1.00438583e+00 2.70424277e-01 1.00255191e-01 5.47238827e-01 -1.13162510e-01 3.29331160e-01 4.24862295e-01 -1.77946195e-01 -1.04463625e+00 6.93821669e-01 7.48782754e-01 9.13603306e-01 -7.84445345e-01 8.22965980e-01 2.43126646e-01 -1.32220936e+00 -8.55864659e-02 -6.14069819e-01 2.24425331e-01 6.45520538e-02 6.44960999e-01 -2.60313988e-01 7.73031652e-01 8.32013488e-01 1.07328665e+00 -8.33415926e-01 9.40830469e-01 -8.18861499e-02 2.05604315e-01 -1.46451905e-01 1.12409592e-01 2.37553462e-01 -1.77826926e-01 1.94011554e-01 1.26502645e+00 2.45023891e-01 -2.46406987e-01 6.39769614e-01 7.67650962e-01 -3.31970245e-01 3.05820912e-01 -6.33152544e-01 -1.48732245e-01 2.86425143e-01 1.42153871e+00 -3.13033909e-01 -2.94379443e-01 -6.00396991e-01 1.25439322e+00 4.48676616e-01 5.04637718e-01 -9.24335837e-01 -5.57380676e-01 4.41206634e-01 3.44151892e-02 5.75226247e-01 9.79030132e-03 1.67563483e-01 -1.46115160e+00 4.91083175e-01 -7.03086317e-01 3.84824246e-01 -6.69687212e-01 -1.79824555e+00 4.86054718e-01 -1.30057633e-01 -1.38499856e+00 -2.01090705e-02 -6.83851540e-01 -2.36530110e-01 8.94562721e-01 -1.86550999e+00 -1.56065595e+00 -4.98863399e-01 8.67985964e-01 1.70118570e-01 -1.93419635e-01 5.97187757e-01 7.39918768e-01 -5.82237184e-01 1.16622424e+00 2.15044335e-01 6.16238773e-01 1.13370681e+00 -1.06149113e+00 -2.34896183e-01 5.79394758e-01 1.77944496e-01 7.26078749e-01 2.48069480e-01 -2.48957291e-01 -1.39223552e+00 -1.18348670e+00 8.20511460e-01 -2.61178911e-01 6.22202754e-01 -5.84736228e-01 -7.84039676e-01 5.76622903e-01 3.80318314e-01 4.12004888e-01 7.99099505e-01 7.39641488e-02 -6.92569077e-01 -6.21758163e-01 -9.18188334e-01 5.13596296e-01 8.61131489e-01 -8.85671854e-01 -5.01956582e-01 3.42570007e-01 6.84575260e-01 -2.01391533e-01 -8.12285483e-01 2.94478804e-01 6.38495624e-01 -7.56082773e-01 1.04679680e+00 -5.11529624e-01 6.47141397e-01 -5.26338160e-01 -3.61719966e-01 -1.03150094e+00 -2.94601500e-01 -1.57769546e-01 -1.84713807e-02 1.50481284e+00 1.96075365e-01 -6.09774590e-01 6.04166806e-01 4.35107738e-01 7.72151053e-02 -8.74874830e-01 -8.66497099e-01 -6.97820067e-01 1.25112295e-01 -2.03787312e-01 4.84843671e-01 1.04465568e+00 -2.30625510e-01 4.32243526e-01 -6.61505282e-01 1.83916733e-01 5.92796922e-01 2.96496004e-01 8.20996642e-01 -1.10071647e+00 -3.89048606e-01 -3.43133926e-01 -6.12062275e-01 -1.37360656e+00 2.92740047e-01 -1.15161812e+00 -7.81023651e-02 -1.53964460e+00 6.65522933e-01 -4.83775914e-01 -6.05341434e-01 5.29523790e-01 -3.45365793e-01 4.50460017e-01 4.70057219e-01 2.34152034e-01 -8.92361403e-01 8.00792038e-01 1.10557222e+00 -5.65999091e-01 1.04045674e-01 -3.65324706e-01 -6.80706024e-01 6.04421854e-01 6.65931880e-01 -2.76786804e-01 -4.68146116e-01 -7.91648746e-01 3.46666604e-01 -2.46689662e-01 7.31490970e-01 -7.83899128e-01 3.63395989e-01 7.96951205e-02 5.51171899e-01 -7.58506477e-01 5.02651334e-01 -1.13519657e+00 -3.58207256e-01 9.51049030e-02 -5.38019121e-01 -4.34501320e-02 -8.43949467e-02 8.67264867e-01 -4.41785872e-01 -1.00346930e-01 5.94477117e-01 2.02951003e-02 -6.17745519e-01 6.57096088e-01 4.12025541e-01 2.04596356e-01 8.95926178e-01 -9.41097662e-02 -2.88675934e-01 -3.29006374e-01 -5.59341788e-01 5.12664676e-01 5.52857161e-01 6.78890765e-01 7.64533043e-01 -1.85392427e+00 -6.37950540e-01 1.87887415e-01 4.64335084e-01 -2.30971903e-01 4.80485499e-01 8.16875577e-01 -1.30169153e-01 3.99705917e-01 1.19834216e-02 -8.23479772e-01 -1.33160174e+00 6.76617920e-01 3.65128666e-01 -4.17085081e-01 -4.61794496e-01 7.37834275e-01 5.14508128e-01 -5.11995077e-01 5.32090902e-01 2.04397701e-02 -2.43710682e-01 2.06960753e-01 7.39199281e-01 -2.04418469e-02 -2.72439092e-01 -6.51243687e-01 -4.47083592e-01 9.07373548e-01 -3.62537295e-01 -1.94989089e-02 1.18951881e+00 -2.33568817e-01 -2.08178312e-01 4.45565671e-01 1.84635437e+00 -3.56001630e-02 -1.12193108e+00 -6.04565680e-01 -4.19451833e-01 -7.06692100e-01 6.71010837e-02 -6.33851767e-01 -1.36270976e+00 1.16341591e+00 9.21805441e-01 1.81087274e-02 1.14122975e+00 5.27847633e-02 9.17102695e-01 3.82032156e-01 2.34441638e-01 -1.28281164e+00 3.34563047e-01 3.50546449e-01 8.33510399e-01 -1.39131105e+00 5.30457776e-03 -2.10647866e-01 -6.02638304e-01 9.76390123e-01 6.36898458e-01 -1.68526039e-01 6.18875146e-01 -3.15458506e-01 -4.96255010e-02 -8.94667208e-03 -4.93059576e-01 -2.25616023e-01 6.89637482e-01 3.19700301e-01 3.87528032e-01 6.67655915e-02 -2.76296556e-01 6.36680305e-01 6.56618252e-02 3.32997963e-02 3.20264585e-02 6.37220383e-01 -2.31971592e-01 -1.26474988e+00 -1.87529519e-01 3.45338196e-01 -3.76488954e-01 -2.44821519e-01 -3.29665810e-01 7.74072647e-01 2.56244600e-01 7.20313668e-01 8.37101117e-02 -3.83347392e-01 3.69633049e-01 -9.13122445e-02 4.28755492e-01 -4.34338927e-01 -2.11489514e-01 1.04220450e-01 -2.38956496e-01 -4.14510995e-01 -5.39268196e-01 -3.78492713e-01 -1.08723450e+00 -1.75464600e-01 -2.98323810e-01 7.45361447e-02 3.97319078e-01 7.32880771e-01 3.85786176e-01 3.81664038e-01 7.74856448e-01 -7.02412963e-01 -5.88375628e-01 -7.73298442e-01 -6.25868618e-01 7.25970566e-01 3.69512528e-01 -6.43071115e-01 -3.44326019e-01 1.17274718e-02]
[10.892440795898438, 1.256798505783081]
e1184058-106f-4cc9-84b6-edff758e8d48
clarifying-system-1-2-through-the-common
2305.10654
null
https://arxiv.org/abs/2305.10654v1
https://arxiv.org/pdf/2305.10654v1.pdf
Clarifying System 1 & 2 through the Common Model of Cognition
There have been increasing challenges to dual-system descriptions of System-1 and System-2, critiquing them as imprecise and fostering misconceptions. We address these issues here by way of Dennett's appeal to use computational thinking as an analytical tool, specifically we employ the Common Model of Cognition. Results show that the characteristics thought to be distinctive of System-1 and System-2 instead form a spectrum of cognitive properties. By grounding System-1 and System-2 in the Common Model we aim to clarify their underlying mechanisms, persisting misconceptions, and implications for metacognition.
['Robert L. West', 'Brendan Conway-Smith']
2023-05-18
null
null
null
null
['misconceptions']
['miscellaneous']
[ 7.47176632e-03 2.51640171e-01 3.53990614e-01 -3.59853730e-03 4.16867554e-01 -8.02464306e-01 9.33939457e-01 6.03432119e-01 -1.13134488e-01 -5.32127023e-02 2.78753817e-01 -1.37583566e+00 -7.24200428e-01 -6.17892385e-01 -1.28482699e-01 -5.68560883e-02 1.60324171e-01 -2.05665343e-02 1.34728923e-01 -6.15430474e-01 1.08490169e+00 3.81465018e-01 -1.46667719e+00 4.03152376e-01 1.00684547e+00 3.48770648e-01 8.24403316e-02 5.04967749e-01 -1.68325782e-01 1.62053895e+00 -6.45966411e-01 -4.39733028e-01 -1.62156373e-01 -5.98050535e-01 -1.19708490e+00 -2.00917214e-01 1.63322568e-01 -2.80371923e-02 -1.16858527e-01 1.29298151e+00 -1.02111241e-02 -2.93836277e-02 5.58586597e-01 -1.15813673e+00 -1.03372455e+00 5.22587180e-01 2.60127038e-01 5.83946824e-01 6.12122416e-01 2.86629319e-01 3.02975178e-01 -5.72924137e-01 7.55169913e-02 1.58453500e+00 1.08133447e+00 2.24860221e-01 -1.47913086e+00 -4.92304415e-01 3.00438821e-01 2.75207400e-01 -1.72952271e+00 -5.69820881e-01 1.50567785e-01 -9.96935725e-01 1.23013854e+00 5.30545473e-01 1.14057004e+00 5.03963053e-01 7.94544935e-01 3.94394919e-02 1.60603690e+00 -7.24995017e-01 2.58674800e-01 5.86909771e-01 9.35553610e-01 2.41187871e-01 9.38353300e-01 3.57190788e-01 -3.11046153e-01 -2.67408639e-01 9.72404003e-01 -2.24732190e-01 -2.00937778e-01 2.22548932e-01 -6.97057128e-01 4.45485145e-01 8.00243020e-02 8.12971950e-01 -2.76710093e-01 1.03299737e-01 1.57336339e-01 6.04185820e-01 -7.67485872e-02 1.00795674e+00 -1.02679454e-01 -2.31748343e-01 -4.26978886e-01 4.23997521e-01 9.58250284e-01 5.98152578e-01 3.63096923e-01 2.70474106e-01 1.77056283e-01 2.75761068e-01 5.26500165e-01 -2.50533428e-02 8.60879302e-01 -9.08249438e-01 -4.14927602e-01 6.66773379e-01 2.38165095e-01 -1.43675339e+00 -6.39117956e-01 -6.74346447e-01 3.88373956e-02 3.65969092e-01 2.70534158e-01 9.93721113e-02 -5.51492751e-01 1.75730801e+00 -2.99838066e-01 -3.83350819e-01 2.82495439e-01 5.73220193e-01 4.62072968e-01 2.37487450e-01 6.23123527e-01 -1.81145430e-01 1.27924430e+00 -3.71997476e-01 -7.54792452e-01 -4.86272663e-01 8.24569166e-01 -7.98767805e-01 1.16207957e+00 6.79734230e-01 -1.59744537e+00 -6.50269151e-01 -1.04346156e+00 1.09685771e-01 -5.11075377e-01 -2.36085802e-01 9.39295352e-01 1.29438329e+00 -1.43433797e+00 5.71375370e-01 -8.80190074e-01 -3.41070890e-01 -2.49970615e-01 -2.74250954e-01 4.64788377e-01 3.43697667e-01 -1.07739484e+00 1.68902946e+00 8.37938249e-01 3.01641166e-01 -2.65613735e-01 -6.68313026e-01 -4.96604592e-01 1.97849125e-01 2.24805504e-01 -6.83122754e-01 1.37293315e+00 -1.21256042e+00 -1.12155914e+00 5.09613037e-01 1.91851005e-01 -3.19374986e-02 1.28405303e-01 -4.65055108e-02 -1.24688649e+00 -5.24810478e-02 1.05405837e-01 -2.81116098e-01 1.51159197e-01 -1.43312812e+00 -2.01993838e-01 -1.73661768e-01 2.64235020e-01 1.89903557e-01 2.11480409e-02 3.36403176e-02 3.76016378e-01 -4.25131261e-01 9.32916582e-01 -5.51190197e-01 3.99192609e-02 -2.89860874e-01 -1.43807650e-01 9.30163413e-02 -5.05117625e-02 -3.17144275e-01 1.93615723e+00 -1.90273798e+00 -7.09037542e-01 5.09096801e-01 7.83769131e-01 2.71083623e-01 3.13957691e-01 1.15946722e+00 -5.05315840e-01 4.89114940e-01 3.48544598e-01 6.16583467e-01 4.24433172e-01 -1.40492609e-02 -3.69510114e-01 2.37104088e-01 2.44643241e-02 6.73683703e-01 -1.05796695e+00 -1.73299193e-01 1.50454938e-01 1.91588730e-01 -3.27010840e-01 -1.17817819e-01 1.49346679e-01 -1.95205674e-01 -9.90566760e-02 4.11618024e-01 7.88455367e-01 -6.26126409e-01 5.91082215e-01 2.13995166e-02 -5.81068575e-01 8.40848744e-01 -1.13614321e+00 8.89992476e-01 -1.22736439e-01 6.86368227e-01 -4.67664510e-01 -5.99913299e-01 5.21428585e-01 3.11160833e-01 -4.55034256e-01 -8.33247304e-01 -2.80755181e-02 1.04793750e-01 1.01644230e+00 -7.23765373e-01 5.34135818e-01 -6.03414416e-01 1.65331841e-01 8.78838778e-01 6.66293800e-02 -4.15637195e-01 -8.57426897e-02 7.03326106e-01 8.30688119e-01 -2.52357960e-01 4.74800736e-01 -1.11552536e+00 3.01161706e-01 9.52124298e-02 2.31923714e-01 9.11070228e-01 -4.02545109e-02 -2.46639811e-02 4.37274486e-01 -3.12402099e-01 -6.79834545e-01 -1.07989585e+00 -5.05046189e-01 6.61731482e-01 2.28461117e-01 -9.00892496e-01 -7.01244771e-01 3.32337797e-01 -2.90342093e-01 1.83319676e+00 -5.84535003e-01 -5.19568920e-01 9.44115892e-02 -4.16699529e-01 7.71236300e-01 5.13001919e-01 4.95918006e-01 -3.03011775e-01 -1.42083001e+00 3.83533061e-01 4.43107486e-02 -4.31952119e-01 2.85742044e-01 8.72311741e-02 -8.10175180e-01 -1.05415046e+00 2.02185795e-01 -2.01242298e-01 7.19370961e-01 6.89683378e-01 1.29976594e+00 1.12516022e+00 5.76371439e-02 8.34727526e-01 -2.70366043e-01 -5.77018261e-01 -7.62495100e-01 -8.21925759e-01 -1.24492541e-01 -9.08394754e-01 6.31181479e-01 -5.17956078e-01 -3.68445247e-01 1.32996976e-01 -1.13460946e+00 2.75476038e-01 4.27596569e-01 3.81726265e-01 -5.06791592e-01 2.04468518e-01 4.85924602e-01 -9.66085076e-01 1.37577868e+00 -7.73155868e-01 -5.67081496e-02 3.52451861e-01 -1.41228056e+00 -1.47846833e-01 1.76553071e-01 -5.01545489e-01 -1.47997069e+00 -1.12980962e+00 2.83824325e-01 2.71403521e-01 -1.94915190e-01 1.02593422e+00 2.43179083e-01 -2.75065720e-01 1.05659938e+00 4.20080036e-01 1.82385147e-01 -1.18648849e-01 2.31425196e-01 2.33080178e-01 5.05596876e-01 -1.04735065e+00 4.46925342e-01 -6.70662373e-02 -3.83343637e-01 -7.57343650e-01 -6.67728543e-01 4.59564514e-02 -1.96338162e-01 -3.62278908e-01 2.67309010e-01 -9.89767134e-01 -8.27012420e-01 -3.08995806e-02 -1.14994359e+00 -8.89179483e-02 -6.28276020e-02 3.44882220e-01 -5.45299530e-01 5.14792204e-01 -7.08452046e-01 -1.33267796e+00 1.25147477e-01 -7.93908596e-01 -3.45106572e-02 2.23281100e-01 -1.03122354e+00 -1.31283391e+00 3.16622853e-01 -9.10744295e-02 8.07471812e-01 9.81991366e-02 1.36317873e+00 -1.12540269e+00 -7.75790364e-02 -3.59871909e-02 -1.31131560e-01 -6.27575666e-02 7.44581670e-02 -1.42033473e-02 -8.85195017e-01 -1.70130096e-02 6.76288486e-01 -2.04547212e-01 7.83176273e-02 -3.71788383e-01 7.15253353e-01 -6.29176855e-01 -7.85117298e-02 5.65327331e-02 1.72256804e+00 6.67895913e-01 8.64789963e-01 4.62239385e-01 -1.39516994e-01 9.07640100e-01 1.70352101e-01 2.38060415e-01 5.86038589e-01 -1.34704672e-02 -1.81563586e-01 6.35917068e-01 1.40569970e-01 -3.01009417e-01 1.41657382e-01 1.00545192e+00 -9.04667825e-02 2.99626946e-01 -1.43797112e+00 4.34849679e-01 -1.69359648e+00 -1.04189157e+00 -4.55186486e-01 2.15519929e+00 8.72500658e-01 3.99418473e-01 -2.96324193e-02 -1.97386779e-02 4.00635242e-01 -3.82798135e-01 -1.00169048e-01 -7.38766730e-01 1.96543753e-01 8.70637223e-03 -2.99594998e-01 6.76810801e-01 1.15441807e-01 7.53698647e-01 8.11393738e+00 2.18817234e-01 -8.46174419e-01 -7.54204318e-02 3.16146731e-01 2.13557467e-01 -8.07828069e-01 6.23087108e-01 -2.39813924e-01 4.33046162e-01 1.45017552e+00 -7.67791212e-01 1.66897416e-01 3.28786671e-01 9.51247104e-03 -6.15406513e-01 -1.36479759e+00 4.18790102e-01 1.14860684e-02 -1.11785364e+00 3.30640882e-01 -2.63989151e-01 2.66514838e-01 -9.03466105e-01 6.33302554e-02 4.73871112e-01 3.90164852e-01 -1.23777604e+00 1.31537652e+00 6.74866378e-01 1.51358053e-01 -5.08823395e-01 3.86128813e-01 3.07603896e-01 -4.94627655e-01 1.04664788e-01 -3.21163356e-01 -1.18514097e+00 -2.32989833e-01 1.36981532e-01 -2.79188752e-01 1.86097652e-01 3.09452742e-01 -6.88864067e-02 -8.37375641e-01 8.60514879e-01 6.04599416e-02 3.86357307e-01 7.49657582e-03 -3.48321125e-02 2.81395987e-02 -1.41327411e-01 1.64823622e-01 1.23120320e+00 3.28019202e-01 7.38483071e-01 -2.01548815e-01 1.70984864e+00 1.20899427e+00 -4.40154374e-01 -3.97141397e-01 -4.10647094e-01 1.00743186e+00 7.20735371e-01 -1.05707288e+00 -7.63916731e-01 -3.19507360e-01 4.51870620e-01 -1.76070616e-01 3.59032989e-01 -4.85556215e-01 -2.03632876e-01 4.25252229e-01 1.38795197e-01 -3.56624603e-01 -3.72407585e-01 -6.90356374e-01 -1.11476898e+00 -4.75183606e-01 -9.76267815e-01 -2.07679316e-01 -9.92820799e-01 -1.11463964e+00 4.86356825e-01 4.26207691e-01 -7.82681763e-01 -2.75414288e-01 -7.53395975e-01 -9.78186011e-01 1.23800325e+00 -8.42960835e-01 -9.07668829e-01 -3.03283900e-01 4.84094441e-01 3.44085880e-03 3.16380382e-01 1.11539960e+00 -3.01130116e-01 -3.49001706e-01 1.90140724e-01 -1.27874732e-01 -5.42811573e-01 3.30327421e-01 -1.28513706e+00 4.53592718e-01 6.66276753e-01 -3.44156802e-01 1.74089336e+00 1.12857926e+00 -7.36211777e-01 -1.35938025e+00 -2.05396593e-01 8.70956600e-01 -8.28184366e-01 9.19165909e-01 9.99623537e-02 -1.24788380e+00 9.55031574e-01 4.49263990e-01 -9.41814363e-01 1.28971446e+00 1.65601224e-01 -3.91529918e-01 7.18577981e-01 -1.03427219e+00 9.21456158e-01 7.73808420e-01 -6.83639526e-01 -1.42675161e+00 1.77981615e-01 4.71380174e-01 -1.99187726e-01 -7.15985477e-01 -3.67684305e-01 6.85871601e-01 -1.26662874e+00 9.83081937e-01 -5.71536064e-01 1.56476617e-01 -1.72658667e-01 -9.10272151e-02 -1.26105881e+00 -8.89499962e-01 -5.74588895e-01 1.99968204e-01 8.73336434e-01 2.48827577e-01 -1.26888967e+00 -4.74131592e-02 1.58288717e+00 -3.06650400e-01 -1.21183082e-01 -3.66259456e-01 -8.16795468e-01 6.02185428e-01 -7.39280939e-01 5.87974548e-01 1.48284078e+00 1.18705249e+00 3.18676740e-01 5.41368306e-01 1.67945996e-01 4.37143803e-01 -6.71419129e-02 2.15967998e-01 -1.60583818e+00 -4.03103590e-01 -8.40375364e-01 -1.22180834e-01 -6.92053556e-01 -4.06285375e-01 -6.42342925e-01 -3.35067242e-01 -1.10223651e+00 1.52852148e-01 -3.32490414e-01 -1.40857518e-01 1.75642401e-01 1.60323530e-01 -3.80366564e-01 5.46311378e-01 5.63730955e-01 -1.41190559e-01 -7.24475756e-02 7.55657434e-01 5.02047002e-01 -2.31117949e-01 -6.38941467e-01 -1.88444090e+00 1.12961423e+00 7.57390380e-01 -3.37087698e-02 -7.77223527e-01 -3.79401863e-01 9.30022478e-01 2.03681171e-01 7.88827360e-01 -1.19804573e+00 6.03738070e-01 -6.35549486e-01 5.21656394e-01 -1.79510564e-01 1.64716452e-01 -8.34671021e-01 6.15152240e-01 8.85729313e-01 -4.12006885e-01 4.86721873e-01 8.66367340e-01 1.86364338e-01 2.43882939e-01 -6.12740397e-01 2.53457159e-01 -4.71975714e-01 -7.17775702e-01 -1.06592679e+00 -1.25475180e+00 -2.77919114e-01 7.85359502e-01 -6.53381824e-01 -7.40012765e-01 -2.46186197e-01 -8.71122599e-01 -3.44175622e-02 5.83256304e-01 -1.79034211e-02 6.12132192e-01 -1.00287092e+00 -1.53516337e-01 2.85518169e-01 9.47775692e-02 -7.04252005e-01 3.69263411e-01 8.16645622e-01 -6.70114040e-01 7.94768691e-01 -5.89644194e-01 -1.37468696e-01 -6.42994523e-01 5.94290495e-01 5.79716444e-01 6.49142504e-01 -4.29057717e-01 5.53084731e-01 3.43186855e-01 -4.47985083e-02 -3.37373257e-01 -5.40440500e-01 -1.21301390e-01 -1.84067711e-01 6.94668591e-01 6.20450377e-01 -8.35226923e-02 -2.21359327e-01 -3.25802207e-01 -3.21442261e-03 -3.11090529e-01 -3.52880895e-01 5.98761022e-01 -3.98330390e-01 -5.19170403e-01 9.16037917e-01 1.43991485e-01 -2.77103573e-01 -3.78915250e-01 -6.64537251e-02 2.37566590e-01 -7.81964183e-01 1.01994380e-01 -1.29653537e+00 2.33745068e-01 7.98304558e-01 5.03045201e-01 6.38287842e-01 7.37712502e-01 -2.57391036e-01 -2.19715506e-01 5.75426459e-01 3.68350118e-01 -9.63385820e-01 -4.94520552e-03 5.29786527e-01 9.00724292e-01 -3.42489958e-01 9.44637507e-02 -3.86138439e-01 -3.31995338e-01 1.40139341e+00 1.01959705e+00 -1.82648599e-01 6.94599986e-01 1.81961954e-01 -7.68384263e-02 -5.55753469e-01 -1.23937702e+00 1.25314802e-01 -1.43720537e-01 5.13355672e-01 1.13121259e+00 1.11879557e-01 -1.01839399e+00 1.21206856e+00 -5.64728141e-01 2.45046645e-01 1.08762383e+00 1.39946699e+00 -5.98968029e-01 -6.84798181e-01 -7.57869601e-01 2.90167689e-01 -2.95079708e-01 -3.86418581e-01 -5.55991828e-01 1.00741827e+00 -9.72687304e-02 1.10945153e+00 2.84648597e-01 -4.95881468e-01 1.05972059e-01 2.90312171e-01 6.54536128e-01 -6.87215090e-01 -9.45005655e-01 1.68281019e-01 1.83869749e-01 -3.73972654e-01 -8.18891302e-02 -5.03675580e-01 -1.05579448e+00 -8.85151088e-01 -5.55021703e-01 3.09728980e-01 3.31833512e-01 1.07567191e+00 3.93775821e-01 6.87618434e-01 -3.69797826e-01 -2.73946822e-01 -1.06935346e+00 -9.76928711e-01 -5.16531587e-01 2.85904575e-02 8.10758844e-02 -7.16616988e-01 -2.88435340e-01 -2.97346748e-02]
[9.402811050415039, 7.082505226135254]
41c62aa4-44f0-40d2-aad4-c8585d2dea96
learning-from-synthetic-animals
1912.08265
null
https://arxiv.org/abs/1912.08265v2
https://arxiv.org/pdf/1912.08265v2.pdf
Learning from Synthetic Animals
Despite great success in human parsing, progress for parsing other deformable articulated objects, like animals, is still limited by the lack of labeled data. In this paper, we use synthetic images and ground truth generated from CAD animal models to address this challenge. To bridge the domain gap between real and synthetic images, we propose a novel consistency-constrained semi-supervised learning method (CC-SSL). Our method leverages both spatial and temporal consistencies, to bootstrap weak models trained on synthetic data with unlabeled real images. We demonstrate the effectiveness of our method on highly deformable animals, such as horses and tigers. Without using any real image label, our method allows for accurate keypoint prediction on real images. Moreover, we quantitatively show that models using synthetic data achieve better generalization performance than models trained on real images across different domains in the Visual Domain Adaptation Challenge dataset. Our synthetic dataset contains 10+ animals with diverse poses and rich ground truth, which enables us to use the multi-task learning strategy to further boost models' performance.
['Weichao Qiu', 'Gregory Hager', 'Jiteng Mu', 'Alan Yuille']
2019-12-17
learning-from-synthetic-animals-1
http://openaccess.thecvf.com/content_CVPR_2020/html/Mu_Learning_From_Synthetic_Animals_CVPR_2020_paper.html
http://openaccess.thecvf.com/content_CVPR_2020/papers/Mu_Learning_From_Synthetic_Animals_CVPR_2020_paper.pdf
cvpr-2020-6
['human-parsing']
['computer-vision']
[ 3.75153184e-01 2.80012578e-01 -2.15480879e-01 -5.60272157e-01 -9.86764133e-01 -9.46789265e-01 4.92666185e-01 -3.17031056e-01 -5.36976933e-01 8.05195332e-01 -4.57220487e-02 6.50922060e-02 4.50632721e-01 -3.98147583e-01 -1.34592175e+00 -1.91691056e-01 1.74951911e-01 7.02341616e-01 6.90708280e-01 -1.63562804e-01 -2.29111448e-01 2.03354880e-01 -1.29001689e+00 4.33685303e-01 8.67144763e-01 6.33058488e-01 4.64344084e-01 5.44898272e-01 2.40325660e-01 7.77926743e-01 -2.66972780e-01 -3.44696403e-01 3.34729582e-01 -3.33195895e-01 -9.11914587e-01 1.76007643e-01 9.14408863e-01 -6.57651484e-01 -7.69108683e-02 8.75535488e-01 3.10520262e-01 3.02578732e-02 6.37143373e-01 -1.15845907e+00 -7.56291151e-01 3.35472286e-01 -5.48073709e-01 -3.05876900e-02 2.86732405e-01 2.73131132e-01 9.22930658e-01 -8.07904243e-01 1.20448148e+00 1.27335417e+00 8.22011948e-01 1.09564519e+00 -1.50412524e+00 -7.31846154e-01 2.72663742e-01 -1.13941394e-01 -8.60413909e-01 -5.14984012e-01 8.23081076e-01 -6.05498552e-01 6.33547425e-01 -2.24592626e-01 6.88092530e-01 1.62287140e+00 -6.35282397e-02 9.86012578e-01 1.22312546e+00 -2.55595773e-01 2.61104316e-01 -1.78014636e-01 -2.07827225e-01 9.10003960e-01 3.03172141e-01 3.60618979e-01 -5.19821346e-01 1.15321115e-01 9.62224185e-01 -2.52213359e-01 -1.13073997e-01 -9.34364736e-01 -1.32681131e+00 6.15550458e-01 4.08492655e-01 -2.11923957e-01 -1.41768456e-01 3.20502430e-01 1.22800693e-01 -5.64045124e-02 4.86330181e-01 4.76991087e-01 -8.03853691e-01 1.57600105e-01 -7.25750983e-01 4.28917050e-01 6.81239784e-01 1.08586705e+00 4.74595934e-01 4.64873686e-02 2.73567855e-01 8.95924389e-01 4.45702732e-01 7.17017651e-01 2.31842443e-01 -1.23285413e+00 5.91372907e-01 4.23017770e-01 1.99589372e-01 -6.22883856e-01 -4.38854635e-01 -2.44190156e-01 -2.83782244e-01 3.97949904e-01 8.16446841e-01 -1.06944069e-01 -1.37254584e+00 2.05710459e+00 5.40123701e-01 1.27634928e-01 1.81156337e-01 9.55129504e-01 9.27425325e-01 3.13555181e-01 4.70572352e-01 3.23912621e-01 1.03920054e+00 -1.14083707e+00 -1.74936280e-01 -7.27990270e-01 5.83624244e-01 -4.60651904e-01 1.12869227e+00 5.19043393e-02 -1.05503917e+00 -5.44943213e-01 -1.06422186e+00 -9.58023593e-02 -6.46209493e-02 4.59713824e-02 6.03797197e-01 3.09353560e-01 -7.32494533e-01 6.59402490e-01 -1.29157865e+00 -3.98801744e-01 8.63795280e-01 1.92675576e-01 -7.76454747e-01 -2.59939194e-01 -7.72000194e-01 8.18620503e-01 3.83410901e-01 -1.87089771e-01 -1.07562006e+00 -8.32591355e-01 -1.16748786e+00 -5.51306605e-01 2.71290600e-01 -7.03329265e-01 1.30575109e+00 -1.03428674e+00 -1.25642121e+00 1.43208134e+00 2.13219449e-01 -5.35875142e-01 8.36143255e-01 -3.52326781e-01 -2.32845843e-02 3.58309358e-01 3.84868562e-01 1.40487516e+00 8.11364114e-01 -1.64610183e+00 -4.61073846e-01 -3.78969997e-01 2.77014021e-02 4.37841751e-02 2.20247582e-01 -2.73136646e-01 -4.30847317e-01 -7.63284206e-01 3.24666709e-01 -1.33104300e+00 -2.73748726e-01 4.11036521e-01 -3.96780103e-01 1.70435861e-01 6.88472331e-01 -6.24156833e-01 2.85358936e-01 -1.95161486e+00 3.90147805e-01 -2.97360301e-01 1.28202289e-01 2.46617228e-01 -3.83796364e-01 -8.44622627e-02 9.45858210e-02 -1.28309736e-02 -5.16709983e-01 -4.21578735e-01 -1.38549998e-01 6.44075096e-01 -2.78350532e-01 2.46133760e-01 6.52273059e-01 1.29094553e+00 -1.01625943e+00 -9.00601923e-01 1.29031315e-01 2.66019553e-01 -7.94751227e-01 2.64465183e-01 -6.52445257e-01 1.00154722e+00 -5.71302295e-01 7.92127848e-01 6.57522559e-01 -5.30082941e-01 2.38420740e-01 -2.40679964e-01 3.13669384e-01 -1.95238084e-01 -8.05915654e-01 2.15204358e+00 -2.22061589e-01 3.24456900e-01 8.35127011e-02 -8.29414964e-01 5.42525291e-01 -2.67918147e-02 3.08851302e-01 -7.63182700e-01 -1.93031326e-01 1.81769088e-01 -4.58138213e-02 -5.95247984e-01 1.04274586e-01 -2.43997291e-01 -1.81238517e-01 1.76200032e-01 4.18410808e-01 -6.23782337e-01 9.63751599e-02 1.48874462e-01 1.05607963e+00 1.16980076e+00 -3.43542770e-02 -5.33792600e-02 -1.17233343e-01 4.86417115e-01 8.00271273e-01 5.30950546e-01 -4.16975260e-01 9.53577816e-01 1.66077241e-01 -5.20013273e-01 -1.31493711e+00 -1.28275108e+00 -1.38124749e-01 1.16793430e+00 3.61054480e-01 1.34004399e-01 -7.98742652e-01 -1.00657833e+00 1.89268962e-01 3.81399632e-01 -7.13755190e-01 4.61436771e-02 -8.76342177e-01 -6.11160636e-01 5.73355377e-01 9.30431008e-01 5.27296245e-01 -1.10147834e+00 -8.12721193e-01 2.42958874e-01 -2.80107796e-01 -1.59995759e+00 -3.20874214e-01 1.08957991e-01 -8.78532290e-01 -1.10727382e+00 -7.34478533e-01 -8.72887731e-01 6.04853749e-01 8.31209682e-03 1.37338889e+00 -1.07594498e-01 -4.44841713e-01 5.22389352e-01 -3.80140960e-01 -3.06446612e-01 -7.46922851e-01 -4.47437428e-02 -8.59139711e-02 -4.90182251e-01 -1.76957861e-01 -4.99597579e-01 -5.77144623e-01 4.97278631e-01 -7.20620453e-01 3.40722442e-01 5.77694714e-01 1.00028777e+00 9.27622259e-01 -8.68590891e-01 6.55755162e-01 -1.08300221e+00 -1.46904975e-01 -3.22758228e-01 -5.89723825e-01 2.42089927e-01 -1.48587152e-01 3.52921933e-01 3.91487271e-01 -6.69157028e-01 -1.09458804e+00 5.53591073e-01 -7.73753226e-02 -4.52717453e-01 -2.22236544e-01 6.60058856e-02 -8.43298733e-02 -3.62643957e-01 8.23290765e-01 -1.55669361e-01 -1.02845561e-02 -5.26253104e-01 5.24132669e-01 8.94763134e-03 8.49464059e-01 -9.41222847e-01 7.96939075e-01 6.25854075e-01 2.30545606e-02 -3.97301108e-01 -1.10495198e+00 8.74505490e-02 -7.94619083e-01 -5.04483767e-02 1.21976495e+00 -1.22432745e+00 -3.07075948e-01 4.47392225e-01 -1.01796579e+00 -8.41492295e-01 -2.75165409e-01 5.67306757e-01 -8.33277822e-01 1.87470764e-01 -7.28730500e-01 -3.77657920e-01 -2.51858197e-02 -1.00364959e+00 1.50709951e+00 7.22111389e-02 -3.43025088e-01 -1.01819313e+00 5.87199256e-02 8.41486275e-01 7.50296935e-02 8.16642344e-01 7.38793194e-01 -6.10333323e-01 -7.82977283e-01 1.96795948e-02 -1.68172076e-01 3.49819362e-01 -5.57841966e-03 -1.13999434e-01 -8.73995304e-01 -3.35674673e-01 -3.92518252e-01 -1.21781433e+00 9.00590539e-01 2.81802297e-01 1.04272342e+00 6.47321418e-02 -5.07950485e-01 7.11826444e-01 1.07513666e+00 3.34851854e-02 2.11035594e-01 2.09431902e-01 1.01177859e+00 7.93629169e-01 8.29562604e-01 -7.05099553e-02 6.41534328e-01 5.75561523e-01 5.27383327e-01 -2.81945080e-01 -4.47847068e-01 -6.53157413e-01 1.19687263e-02 5.71479023e-01 -7.58963376e-02 -2.97436088e-01 -1.18271494e+00 6.59191370e-01 -1.71626520e+00 -7.67102897e-01 1.68303534e-01 1.88747132e+00 9.05966163e-01 2.39215687e-01 -2.41189301e-02 -4.14313883e-01 5.85892081e-01 5.27416170e-02 -9.48864222e-01 2.60447234e-01 -2.37502068e-01 2.93485641e-01 5.28127551e-01 3.23156774e-01 -1.45115161e+00 1.15267158e+00 6.71668482e+00 4.91505742e-01 -9.16928768e-01 1.02825478e-01 5.51317275e-01 6.97072968e-02 -1.87836096e-01 -1.76114187e-01 -6.52125597e-01 3.52878630e-01 5.49322188e-01 4.83448297e-01 7.59097859e-02 1.01589894e+00 -1.04987569e-01 6.85270727e-02 -1.36679423e+00 7.19725847e-01 1.93000922e-03 -1.32517254e+00 2.51939073e-02 -7.07891658e-02 8.51385117e-01 2.24254593e-01 4.30773050e-02 2.60563761e-01 8.64601076e-01 -1.04180765e+00 8.66152585e-01 2.28981152e-01 8.61679316e-01 -2.10226804e-01 3.06611925e-01 3.43523860e-01 -1.02403057e+00 1.25543237e-01 -5.78330457e-02 3.05353761e-01 1.44116342e-01 -1.13202065e-01 -8.13134313e-01 3.92431378e-01 7.29643166e-01 8.72174740e-01 -5.68793654e-01 4.83688563e-01 -2.46551514e-01 6.30040526e-01 -4.17849660e-01 3.29231441e-01 2.99815256e-02 -3.76099050e-02 3.55699092e-01 9.67203379e-01 -7.15781450e-02 1.03503212e-01 3.59480858e-01 8.14312816e-01 -1.42727584e-01 -2.32723624e-01 -6.32794857e-01 -1.12961695e-01 4.22304988e-01 1.00926244e+00 -8.18043709e-01 -1.97031528e-01 -3.30737889e-01 9.62943912e-01 5.57181835e-01 3.68867129e-01 -9.22432721e-01 1.82084516e-01 4.42606002e-01 1.21201448e-01 3.86105746e-01 -3.73146504e-01 -2.59284526e-01 -1.34130621e+00 1.57651722e-01 -9.13230658e-01 2.74405479e-01 -9.75713909e-01 -1.26787806e+00 4.90270227e-01 2.52722412e-01 -1.31598032e+00 -4.55675274e-01 -6.98108315e-01 -5.80131859e-02 3.69231582e-01 -1.30166638e+00 -1.79421794e+00 -3.00161123e-01 2.85760045e-01 6.26501501e-01 1.09326560e-03 6.54914737e-01 1.70257881e-01 -2.29979381e-01 4.47065413e-01 -2.04386711e-01 3.00953895e-01 8.77073884e-01 -1.28485954e+00 7.21192956e-01 6.95164740e-01 1.56057775e-01 2.98746586e-01 6.49803340e-01 -7.23434508e-01 -1.22956467e+00 -1.23747504e+00 6.26336783e-02 -9.08400178e-01 4.01141107e-01 -4.28419828e-01 -1.03263772e+00 9.72211361e-01 -2.96699136e-01 6.37796223e-01 4.90007490e-01 -2.30635867e-01 -6.75266266e-01 2.74678499e-01 -1.33298433e+00 4.98069763e-01 1.62479675e+00 -3.05613518e-01 -7.63471544e-01 4.33309376e-01 1.00421202e+00 -8.28557491e-01 -8.65199387e-01 9.06877756e-01 7.72662520e-01 -6.18333936e-01 1.25324249e+00 -1.06957197e+00 8.39723289e-01 -2.81213343e-01 -4.42380905e-01 -1.29388034e+00 5.75887375e-02 -6.98927641e-02 4.79841083e-02 1.00283074e+00 4.97825742e-01 -3.95332158e-01 1.03472686e+00 5.73297679e-01 -1.27511963e-01 -5.56195796e-01 -8.63660574e-01 -8.58479381e-01 4.04731274e-01 -2.52492011e-01 2.30558410e-01 1.02574503e+00 -6.01311564e-01 3.07905436e-01 -3.53333950e-01 1.79993585e-01 9.88436162e-01 3.80063742e-01 1.00900006e+00 -1.26783657e+00 -5.59069991e-01 2.46100873e-01 -5.15224397e-01 -1.19006503e+00 5.26868463e-01 -8.30261111e-01 3.15252781e-01 -1.43237174e+00 2.39511266e-01 -6.12131417e-01 1.41564041e-01 8.17906380e-01 -1.86976790e-01 7.05748260e-01 1.74939036e-01 2.53873080e-01 -7.84039795e-01 4.68150735e-01 1.65798771e+00 -8.80747512e-02 1.92161296e-02 -3.59620541e-01 -3.40330750e-01 1.04250801e+00 4.50903893e-01 -5.77723026e-01 -4.76538897e-01 -8.43073308e-01 -6.96767941e-02 1.93503603e-01 7.85019040e-01 -9.49506760e-01 -1.98142976e-01 -3.54839861e-01 6.81401372e-01 -4.05103654e-01 5.73198378e-01 -6.54833317e-01 1.13393389e-01 4.87073839e-01 -2.82471329e-01 -1.69777706e-01 3.41891795e-01 9.25840914e-01 4.38843854e-02 1.08521238e-01 1.07818174e+00 -2.49241441e-01 -9.52455580e-01 3.34298670e-01 2.47473180e-01 5.75004578e-01 1.05416512e+00 -1.26916990e-01 -3.87144327e-01 2.39545051e-02 -9.19601679e-01 4.19914871e-01 8.07142913e-01 4.92960304e-01 5.18304110e-01 -1.14476216e+00 -6.48887694e-01 8.95507410e-02 4.41797286e-01 4.97969717e-01 2.26997226e-01 3.91108304e-01 -7.08231807e-01 -6.11622930e-02 -5.16416788e-01 -9.79243934e-01 -1.22661793e+00 4.18705404e-01 3.67511004e-01 -3.42192948e-02 -5.85282087e-01 9.95966733e-01 4.59490389e-01 -8.04755986e-01 -1.32331997e-01 -6.00042224e-01 1.46069288e-01 -2.98793823e-01 5.45271188e-02 -9.66485031e-03 -1.80960000e-01 -7.36222386e-01 -5.16019046e-01 8.64561915e-01 -1.73081979e-01 -3.46264988e-01 1.51532972e+00 1.49630129e-01 4.41490620e-01 3.17406684e-01 1.09348214e+00 -5.17087057e-02 -1.99456072e+00 -3.07810783e-01 -1.52646182e-02 -3.74718428e-01 -5.15867114e-01 -8.75928581e-01 -8.41882646e-01 7.55060732e-01 6.70904577e-01 -4.10729647e-01 7.09687948e-01 3.96563619e-01 7.49803185e-01 3.88427943e-01 7.31530309e-01 -9.20635462e-01 5.92660189e-01 2.80957043e-01 8.04618299e-01 -1.90540576e+00 -4.94091846e-02 -5.96245587e-01 -8.89870405e-01 7.14729488e-01 8.40837121e-01 -2.99360693e-01 1.77341118e-01 3.45402986e-01 2.38414377e-01 6.95860805e-03 -6.93034828e-01 -2.36442566e-01 2.02243179e-01 1.02048683e+00 1.73547372e-01 -1.09262217e-03 1.20898835e-01 4.50751156e-01 -1.80256397e-01 1.24262720e-01 2.21395954e-01 1.27818143e+00 -3.15188825e-01 -1.19706678e+00 -9.65264812e-02 1.29107118e-01 -5.21535456e-01 1.67068258e-01 -3.95941198e-01 1.05274069e+00 2.01025054e-01 6.42225325e-01 -5.08763380e-02 -2.38124490e-01 4.17981297e-01 -8.07668176e-03 9.74820971e-01 -8.44942689e-01 -9.79935303e-02 -2.31444836e-01 1.63022622e-01 -5.52183509e-01 -7.19767451e-01 -6.75456643e-01 -1.40902627e+00 2.00642347e-01 -1.97369456e-02 -2.49773413e-01 6.68490589e-01 1.00601566e+00 3.17707777e-01 4.73234415e-01 9.47409272e-02 -1.11836457e+00 -5.94696999e-01 -7.78136432e-01 -1.08711228e-01 8.68670166e-01 4.59873468e-01 -8.84894431e-01 -1.54706672e-01 6.12106919e-01]
[7.4850335121154785, -1.062756896018982]
51d4d1fe-a02e-4e7b-b2fb-67f432ebba26
unsupervised-skin-tissue-segmentation-for
null
null
https://www.sciencedirect.com/science/article/abs/pii/S0167865517303860
https://www.sciencedirect.com/science/article/abs/pii/S0167865517303860
Unsupervised skin tissue segmentation for remote photoplethysmography
Segmentation is a critical step for many algorithms, especially for remote photoplethysmography (rPPG) applications as only the skin surface provides information. Moreover, it has been shown that the rPPG signal is not distributed homogeneously across the skin. Most of the time, algorithms get input information from face detection provided by a supervised learning of physical appearance and skin pixel selection. However, both methods show several limitations. In this paper, we propose a simple approach to implicitly select skin tissues based on their distinct pulsatility feature. The input video frames are decomposed into several temporal superpixels from which the pulse signals are extracted. A pulsatility measure from each temporal superpixel is then used to merge the pulse traces and estimate the photoplethysmogram signal. Since the most pulsatile signals provide high quality information, areas where the information is predominant are favored. We evaluated our contribution using a new publicly available dataset dedicated to rPPG algorithms comparison. The results of our experiments show that our method outperforms state of the art algorithms, without any critical face or skin detection.
['Julien Dubois', 'Alamin Mansouri', 'Yannick Benezeth', 'Richard Macwan', 'Serge Bobbia']
2019-06-01
null
null
null
pattern-recognition-letters-2019-6
['face-detection']
['computer-vision']
[ 3.97222221e-01 2.28869710e-02 -2.02008516e-01 -2.08971605e-01 -5.41242540e-01 -4.18458968e-01 2.51633793e-01 -9.37996283e-02 -2.94408441e-01 7.15899944e-01 -1.88217372e-01 4.88117278e-01 2.64775246e-01 -5.25625587e-01 -8.11034590e-02 -1.25334132e+00 -2.98581142e-02 2.37096593e-01 4.04114753e-01 1.92716554e-01 2.62545764e-01 7.21543133e-01 -1.66271389e+00 2.02027962e-01 1.04122984e+00 1.24381757e+00 -1.97966143e-01 6.83372915e-01 -1.89835280e-01 4.38788295e-01 -4.78075683e-01 -9.95696858e-02 2.43853763e-01 -9.23873842e-01 -4.01826441e-01 1.93225056e-01 6.63784862e-01 -2.80287236e-01 1.00964969e-02 8.70179832e-01 8.20389092e-01 -5.36362156e-02 6.97664738e-01 -1.06042278e+00 2.97607094e-01 2.19725966e-01 -9.15790796e-01 3.08497727e-01 1.96108729e-01 2.99650282e-01 5.13074398e-01 -9.06810820e-01 8.50639284e-01 6.83606029e-01 5.65848887e-01 6.34469271e-01 -1.39846110e+00 -3.71370018e-01 -3.68610680e-01 4.03445780e-01 -1.33427846e+00 -5.03536582e-01 1.15738213e+00 -3.23200315e-01 4.04252946e-01 6.27138972e-01 7.41040766e-01 5.66928625e-01 1.92953944e-02 7.40399897e-01 1.80698729e+00 -3.78253073e-01 3.03143144e-01 4.51205224e-01 6.47280738e-02 6.59446955e-01 1.89747270e-02 -1.71215147e-01 -6.07106030e-01 -2.48974502e-01 4.83011365e-01 -5.06420553e-01 -6.65489137e-01 -2.57822067e-01 -5.46027184e-01 2.88127691e-01 -1.42301202e-01 5.75693607e-01 -6.08122051e-01 -7.68472180e-02 2.83266991e-01 2.84696799e-02 5.59970856e-01 -2.86743436e-02 -3.63911353e-02 -2.68179417e-01 -1.50018144e+00 -2.96797156e-01 1.04646671e+00 1.30624130e-01 8.19524348e-01 -2.07950860e-01 -2.40569159e-01 8.82346213e-01 1.10600106e-01 5.60768425e-01 2.23788559e-01 -1.10409153e+00 -1.66149661e-01 3.22362512e-01 -3.28750536e-02 -9.21890199e-01 -3.14848006e-01 5.24387360e-02 -6.27184629e-01 4.93109316e-01 8.29980612e-01 -1.57072008e-01 -8.10477734e-01 1.10605526e+00 6.14152610e-01 5.43778539e-01 -2.93571442e-01 1.29398251e+00 7.77456939e-01 4.53941703e-01 1.48472965e-01 -9.06135440e-01 1.35503352e+00 -4.58811432e-01 -7.50413001e-01 2.71026611e-01 9.52378288e-03 -6.98884189e-01 6.42742634e-01 6.66183233e-01 -1.05404139e+00 -4.20797735e-01 -8.66078019e-01 3.03632736e-01 -9.87110101e-03 3.15817177e-01 1.63214609e-01 1.04605973e+00 -1.05056012e+00 9.47162807e-01 -7.15998173e-01 -3.28208119e-01 2.80413836e-01 1.79500520e-01 -1.92899376e-01 6.22399263e-02 -9.77103531e-01 8.19069207e-01 -1.06616504e-02 3.01925272e-01 -2.99436003e-01 -8.28049481e-01 -5.52759290e-01 -2.23397642e-01 1.79083318e-01 -1.40211031e-01 8.11002076e-01 -1.11430085e+00 -1.93457448e+00 1.24797320e+00 -2.59478480e-01 -3.14814925e-01 8.11250031e-01 3.15630734e-01 -2.42877096e-01 1.07564771e+00 -4.07796592e-01 2.29735479e-01 1.27210295e+00 -1.36705887e+00 -3.50111991e-01 -3.17680120e-01 -6.60011411e-01 1.74745813e-01 -1.43594354e-01 2.39647910e-01 -6.14959717e-01 -1.44338965e-01 -9.91507806e-03 -7.33443439e-01 -2.87746917e-02 2.50474721e-01 -2.20256373e-01 -4.94593231e-04 8.34008574e-01 -1.05645978e+00 1.15435767e+00 -2.05150390e+00 -1.54324621e-01 4.85482335e-01 3.19137633e-01 3.21163237e-01 6.23391308e-02 9.41476226e-02 6.63141310e-02 -7.93853998e-02 -5.33847809e-01 -2.57224441e-01 -2.44351655e-01 -6.55381903e-02 -8.78794026e-03 9.22965705e-01 -1.68569945e-02 6.12806261e-01 -7.60475039e-01 -1.08297229e+00 6.63291037e-01 7.63901770e-01 2.64498055e-01 3.27315815e-02 6.89261705e-02 5.93543231e-01 -2.25059137e-01 7.27067113e-01 1.14014280e+00 1.76557049e-01 3.07800442e-01 -4.00565416e-01 -2.77361453e-01 -2.46745691e-01 -9.14874554e-01 1.22544801e+00 -1.83938757e-01 8.00374806e-01 4.44366664e-01 -7.19218254e-01 9.76799428e-01 5.75229228e-01 1.20075369e+00 -6.46908522e-01 3.24941337e-01 4.21621799e-01 -1.61166415e-01 -7.88663983e-01 5.71099818e-02 -3.39923173e-01 5.96197009e-01 4.32216704e-01 -9.97643103e-04 -2.86522031e-01 5.22790968e-01 -7.73697719e-02 6.91893876e-01 2.32613042e-01 2.05397621e-01 -3.33153576e-01 8.32355559e-01 -4.46194932e-02 4.65463161e-01 4.04718667e-01 -5.87262511e-01 1.05884206e+00 8.83909583e-01 -1.56576931e-01 -7.74604142e-01 -8.47874701e-01 -4.45076019e-01 4.09845293e-01 3.17644447e-01 -8.80247429e-02 -1.11856055e+00 -6.02415621e-01 -4.96523380e-02 3.36244166e-01 -7.64505208e-01 3.93886089e-01 -5.53566813e-01 -9.51802433e-01 3.21490437e-01 4.96661328e-02 3.64131451e-01 -1.17480779e+00 -9.58554268e-01 2.26666227e-01 -2.69391298e-01 -1.12751794e+00 -3.29651922e-01 -3.87564272e-01 -9.36665237e-01 -1.27141893e+00 -1.23489010e+00 -2.88524479e-01 6.69029772e-01 5.58165573e-02 9.67534006e-01 -3.08985785e-02 -9.04088378e-01 6.95525110e-01 -2.12153763e-01 9.48537793e-03 -3.37674469e-01 -3.74913901e-01 -2.95437723e-01 6.81288302e-01 3.52229774e-01 -6.29009187e-01 -8.66864860e-01 2.33608261e-01 -4.55840647e-01 -1.94968700e-01 4.28953588e-01 3.97888184e-01 7.95793533e-01 -4.88832705e-02 2.16113955e-01 -9.65148926e-01 4.14496750e-01 -3.57077643e-02 -5.62661469e-01 3.69662702e-01 -3.20702314e-01 -2.96023726e-01 4.39497977e-01 -3.57348561e-01 -1.38374162e+00 2.73286343e-01 1.90766156e-01 -5.60648382e-01 -3.89342219e-01 1.78701609e-01 2.34588549e-01 -4.44549739e-01 7.46816039e-01 5.25010824e-01 3.46697778e-01 -3.39185774e-01 1.06988631e-01 5.34203887e-01 4.82958049e-01 -2.98172504e-01 4.69939202e-01 8.20487916e-01 2.51096904e-01 -1.46931064e+00 -2.28310481e-01 -6.64949417e-01 -8.70464504e-01 -8.69093418e-01 7.27185071e-01 -2.29637355e-01 -7.60081589e-01 7.80032337e-01 -9.31233406e-01 -2.79030532e-01 -3.04914266e-01 4.79302615e-01 -5.41923702e-01 8.99839759e-01 -6.95906401e-01 -1.35178816e+00 -5.86784899e-01 -7.13719010e-01 7.84203649e-01 6.94610298e-01 -1.74555816e-02 -1.02493584e+00 3.11094761e-01 2.49911517e-01 4.85602975e-01 6.03103042e-01 4.23435807e-01 5.82284965e-02 -3.19427371e-01 -1.41304284e-01 -1.10074595e-01 3.59606117e-01 2.46859774e-01 5.90413094e-01 -1.49788749e+00 1.10341579e-01 2.17340261e-01 -3.05419620e-02 1.04813790e+00 8.82066190e-01 1.08381414e+00 1.80732727e-01 -1.87895492e-01 5.34961879e-01 1.48779440e+00 3.01924776e-02 9.54755902e-01 -2.95082688e-01 3.58869970e-01 1.08086312e+00 7.34044790e-01 4.60069031e-01 -2.40050673e-01 4.26142752e-01 1.86985984e-01 -4.10274982e-01 -4.63893175e-01 2.15449244e-01 4.38570142e-01 2.33892247e-01 -5.88883698e-01 1.10919252e-01 -6.30532563e-01 4.33128387e-01 -1.40820789e+00 -8.99895012e-01 -4.51924175e-01 2.53427005e+00 9.65785563e-01 -3.70911509e-01 3.88563186e-01 1.06454559e-01 9.82233644e-01 8.20694715e-02 -3.65330398e-01 -2.35974789e-01 -1.34392425e-01 5.81758022e-01 5.54937959e-01 5.13197482e-01 -1.08309650e+00 4.51470882e-01 6.23067474e+00 6.72007442e-01 -1.61259770e+00 -4.39504720e-02 7.70242929e-01 -1.08020887e-01 -2.55890172e-02 -2.32494339e-01 -4.88747686e-01 7.08951175e-01 7.15594172e-01 -1.29993349e-01 1.83633223e-01 3.73346597e-01 4.88404959e-01 -8.51498365e-01 -6.33921087e-01 1.20730495e+00 3.34819645e-01 -9.19351399e-01 -5.92654407e-01 -1.20308839e-01 3.96730423e-01 -3.55438381e-01 -1.21958600e-02 -4.79608178e-01 -5.05940795e-01 -8.95388305e-01 2.46243253e-02 7.50311971e-01 9.73176956e-01 -5.02039850e-01 5.07540345e-01 -1.58888742e-01 -1.07050288e+00 4.14723605e-01 -2.74738997e-01 5.33325851e-01 2.90064692e-01 1.04123604e+00 -9.02772129e-01 4.32078242e-01 3.42636138e-01 5.59337854e-01 -4.26882088e-01 1.42815137e+00 -3.54327530e-01 7.19924629e-01 -5.53987801e-01 8.87525678e-02 -3.00211221e-01 -5.14958918e-01 6.70977056e-01 1.23645735e+00 2.93485165e-01 2.95026332e-01 -1.80358335e-01 9.22014594e-01 2.97569871e-01 4.61310387e-01 -3.25452417e-01 -3.86707187e-02 1.44364715e-01 1.82507849e+00 -1.18007004e+00 -3.33070487e-01 -4.49969620e-01 9.61448014e-01 -4.44099367e-01 4.58745062e-01 -5.81092954e-01 -3.77050459e-01 2.90756434e-01 4.42525208e-01 -6.51226342e-02 1.41246602e-01 -2.44155526e-01 -1.06595838e+00 4.11313213e-02 -3.08888197e-01 6.64977610e-01 -6.32537723e-01 -1.04903376e+00 4.70430672e-01 -1.51305705e-01 -1.17642069e+00 -1.54373750e-01 -5.49895883e-01 -8.34233880e-01 1.12499392e+00 -1.80664885e+00 -8.18642497e-01 -7.18346715e-01 4.35622185e-01 1.39841437e-01 4.79198426e-01 5.59324980e-01 2.83013493e-01 -4.93138522e-01 2.82693565e-01 -2.24483117e-01 -5.08983620e-02 9.12431180e-01 -1.37078488e+00 -2.93363988e-01 8.93476844e-01 -1.69529244e-01 1.16235264e-01 7.59144783e-01 -6.94530666e-01 -1.25167727e+00 -4.36909229e-01 6.34425879e-01 -8.59685168e-02 3.40738833e-01 1.71612035e-02 -9.73870575e-01 -2.01212570e-01 3.42503786e-01 3.29914600e-01 6.10920727e-01 -4.84795868e-01 1.57080472e-01 -4.07494336e-01 -1.50628352e+00 1.94020554e-01 2.11926833e-01 -1.83826223e-01 -1.08669154e-01 8.58070180e-02 -3.29227507e-01 -3.66624773e-01 -9.98018682e-01 2.25860387e-01 6.79460108e-01 -1.25393987e+00 4.92664218e-01 2.74328142e-01 2.27574542e-01 -3.14993352e-01 6.08738244e-01 -1.03112888e+00 2.38462508e-01 -8.99872661e-01 -1.60972878e-01 1.18562925e+00 1.91918418e-01 -7.43272007e-01 1.04817879e+00 8.32834244e-01 2.88154185e-01 -3.27842653e-01 -8.22883427e-01 -3.76424819e-01 -3.24337840e-01 -6.77658769e-04 -2.01835990e-01 8.72727871e-01 2.06884712e-01 -2.33174846e-01 -4.53528106e-01 -1.08543307e-01 1.21085775e+00 5.04760385e-01 3.56857538e-01 -1.27895761e+00 -2.31149361e-01 -4.11409885e-01 -3.25823843e-01 -4.14348453e-01 -1.77149236e-01 -4.71283168e-01 1.98731139e-01 -1.28754473e+00 2.87690401e-01 -2.60761768e-01 -3.76290828e-01 3.98835391e-01 -1.25223950e-01 7.37993658e-01 3.79384946e-05 5.69226481e-02 2.65378729e-02 -4.64159288e-02 1.42532957e+00 8.75846222e-02 -6.24182880e-01 3.48718241e-02 -1.11108094e-01 4.91407067e-01 8.26865256e-01 -1.48778319e-01 -2.74244487e-01 4.55519140e-01 -3.64155680e-01 3.30814630e-01 2.72296429e-01 -1.01446450e+00 2.01221406e-01 -1.39366627e-01 3.86186332e-01 -5.95646799e-01 5.09766459e-01 -5.44774950e-01 1.28547981e-01 6.87232316e-01 1.48705259e-01 -9.32898641e-01 1.70531049e-02 6.14877343e-02 -2.03611597e-01 -3.18275124e-01 1.38471198e+00 -1.75682917e-01 -6.90944016e-01 2.69380122e-01 -3.10124516e-01 -1.70266494e-01 1.07658112e+00 -5.20496726e-01 -1.61932230e-01 -3.82411808e-01 -9.48703945e-01 -3.08590159e-02 5.59207141e-01 -1.85385361e-01 6.16932571e-01 -7.01181889e-01 -8.96037817e-01 1.94905564e-01 -2.40113571e-01 -5.17147064e-01 6.97635949e-01 1.78668642e+00 -7.69061804e-01 -7.68263787e-02 -4.50879961e-01 -8.04555893e-01 -1.71016276e+00 3.29697222e-01 7.03093886e-01 1.79917037e-01 -8.52881849e-01 6.32585824e-01 -6.27435669e-02 4.05076861e-01 -1.07053474e-01 -1.21206343e-01 -4.73609596e-01 4.27896410e-01 7.89565027e-01 6.22027934e-01 1.22014796e-02 -7.58058190e-01 -3.10260564e-01 9.48001385e-01 2.97534794e-01 -1.69080198e-01 1.10238552e+00 -1.36380613e-01 -4.47677046e-01 2.81191409e-01 9.97447431e-01 2.12144792e-01 -1.43863714e+00 8.78307670e-02 -2.37738639e-01 -8.31758559e-01 1.04960047e-01 -6.96301818e-01 -1.37254012e+00 1.09269166e+00 8.11223626e-01 4.14478749e-01 1.65138960e+00 -2.53399581e-01 5.39008319e-01 -5.16872764e-01 1.99623957e-01 -1.38759136e+00 -1.88558996e-01 -2.89945036e-01 6.69785202e-01 -8.43861282e-01 1.32532865e-01 -1.04127228e+00 -8.37676346e-01 1.38964701e+00 3.56718540e-01 -1.63119137e-01 5.02729654e-01 5.22746325e-01 3.94129753e-01 -5.54828569e-02 -3.58022511e-01 -4.07530248e-01 2.33869225e-01 6.89030766e-01 3.44307482e-01 -1.06648095e-01 -9.36026752e-01 8.86432827e-02 3.46140623e-01 1.23352125e-01 5.43654621e-01 4.13668960e-01 -5.81450939e-01 -1.07426631e+00 -4.32898104e-01 3.62985969e-01 -6.92430198e-01 1.53823331e-01 -5.79485893e-01 6.63968980e-01 -8.32158253e-02 6.65702283e-01 -1.62439004e-01 1.21302299e-01 -2.76817773e-02 2.46091127e-01 8.94827366e-01 -2.59094477e-01 -4.19863284e-01 4.38485086e-01 -9.84729547e-03 -6.38350427e-01 -6.89882278e-01 -8.04830909e-01 -1.25178432e+00 -1.05670832e-01 -2.71956604e-02 1.41170964e-01 6.98572516e-01 5.93980014e-01 -2.11549271e-02 -1.28720691e-02 9.25507009e-01 -9.13047194e-01 -8.49188194e-02 -7.35710025e-01 -1.17098248e+00 4.33184355e-01 1.93234146e-01 -3.41412425e-01 -6.24332607e-01 3.45711172e-01]
[13.8807954788208, 2.735976219177246]
562e678a-4cad-4923-b571-81a055772a46
decoupled-and-memory-reinforced-networks
2102.10795
null
https://arxiv.org/abs/2102.10795v1
https://arxiv.org/pdf/2102.10795v1.pdf
Decoupled and Memory-Reinforced Networks: Towards Effective Feature Learning for One-Step Person Search
The goal of person search is to localize and match query persons from scene images. For high efficiency, one-step methods have been developed to jointly handle the pedestrian detection and identification sub-tasks using a single network. There are two major challenges in the current one-step approaches. One is the mutual interference between the optimization objectives of multiple sub-tasks. The other is the sub-optimal identification feature learning caused by small batch size when end-to-end training. To overcome these problems, we propose a decoupled and memory-reinforced network (DMRNet). Specifically, to reconcile the conflicts of multiple objectives, we simplify the standard tightly coupled pipelines and establish a deeply decoupled multi-task learning framework. Further, we build a memory-reinforced mechanism to boost the identification feature learning. By queuing the identification features of recently accessed instances into a memory bank, the mechanism augments the similarity pair construction for pairwise metric learning. For better encoding consistency of the stored features, a slow-moving average of the network is applied for extracting these features. In this way, the dual networks reinforce each other and converge to robust solution states. Experimentally, the proposed method obtains 93.2% and 46.9% mAP on CUHK-SYSU and PRW datasets, which exceeds all the existing one-step methods.
['Yi Yang', 'Nong Sang', 'Changxin Gao', 'Zhedong Zheng', 'Chuchu Han']
2021-02-22
null
null
null
null
['person-search']
['computer-vision']
[-1.69870928e-01 -5.66457570e-01 1.25568941e-01 -5.58721006e-01 -6.67613149e-01 -2.89120287e-01 3.70581120e-01 5.71546936e-03 -8.73523355e-01 6.12356901e-01 9.28643644e-02 2.76657969e-01 -2.13065609e-01 -7.21839666e-01 -5.77605069e-01 -7.69571722e-01 1.22366033e-01 3.83622646e-01 4.07016128e-01 2.95556299e-02 1.36082619e-01 2.12506607e-01 -1.80143273e+00 2.79073089e-01 8.20019305e-01 1.08537292e+00 3.36042434e-01 4.92258132e-01 -1.15620047e-01 4.55627084e-01 -3.95256937e-01 -5.81173539e-01 3.23455632e-01 -7.46702822e-03 -7.08031952e-01 -1.17457561e-01 4.37936097e-01 -4.55368221e-01 -5.79351962e-01 1.05437922e+00 9.72401619e-01 3.94789249e-01 2.01812252e-01 -1.33121789e+00 -6.16022766e-01 4.48766887e-01 -7.44802654e-01 3.52361888e-01 2.87640661e-01 2.79965281e-01 7.60831654e-01 -9.94375646e-01 1.19529469e-02 1.41835296e+00 6.64658368e-01 4.62708890e-01 -9.87220705e-01 -8.93749893e-01 2.94421226e-01 5.48538864e-01 -1.69520962e+00 -6.42256737e-01 4.40657973e-01 -3.33289772e-01 8.55204642e-01 2.98135042e-01 4.65928048e-01 8.79102468e-01 -4.01219964e-01 8.29110682e-01 7.61035025e-01 -1.94720909e-01 -1.18688948e-01 2.62725323e-01 4.40382302e-01 7.72936404e-01 3.24537188e-01 1.61579117e-01 -7.16137767e-01 -5.26067987e-02 4.97329324e-01 3.35114747e-01 -5.18364385e-02 -1.35929972e-01 -1.12966907e+00 5.67455947e-01 4.45966959e-01 1.32199094e-01 -2.25880280e-01 -1.77665986e-02 4.71540064e-01 1.00144826e-01 7.33441859e-02 6.30836785e-02 -2.81361431e-01 -5.84937632e-02 -8.48580241e-01 2.84953713e-01 4.12193716e-01 9.54899013e-01 8.97815585e-01 -4.26343381e-01 -4.82663035e-01 9.37154949e-01 5.19674122e-01 4.31271315e-01 4.12482798e-01 -4.64015603e-01 6.61035419e-01 6.87402725e-01 1.25024170e-01 -1.06997180e+00 -4.97589976e-01 -6.00745201e-01 -8.90583515e-01 -2.92563677e-01 4.41811562e-01 -1.05833635e-01 -6.54111028e-01 1.82813990e+00 5.05321562e-01 4.96854752e-01 -6.04538918e-02 1.13936388e+00 9.66166317e-01 4.05452490e-01 2.80979961e-01 2.77865324e-02 1.53302920e+00 -1.34234643e+00 -4.01223183e-01 -2.92493910e-01 5.37146032e-01 -6.92582726e-01 7.05558956e-01 -9.36220884e-02 -9.54454839e-01 -1.11886179e+00 -1.06829810e+00 -2.70289443e-02 -3.96325111e-01 4.17118907e-01 4.17878777e-01 6.05793178e-01 -8.36946368e-01 3.59902918e-01 -4.80131865e-01 -2.87905753e-01 3.91881943e-01 6.27345145e-01 -3.17480981e-01 -1.62263662e-01 -1.22829676e+00 7.60003686e-01 5.64462125e-01 5.05596995e-01 -5.86154461e-01 -5.35804868e-01 -7.30437517e-01 1.25119582e-01 3.84506345e-01 -7.37700164e-01 9.34931695e-01 -6.22484207e-01 -1.27352762e+00 8.18560421e-01 -3.16601068e-01 -2.87047148e-01 5.28432965e-01 -3.58742356e-01 -4.79043782e-01 -1.79852009e-01 1.95820183e-01 6.87342048e-01 5.78972101e-01 -9.01170731e-01 -1.00915194e+00 -5.35375237e-01 -6.29053637e-02 4.01294947e-01 -6.84476435e-01 2.87531257e-01 -9.74743664e-01 -3.98518831e-01 -7.05238581e-02 -8.20199788e-01 -1.91287473e-01 -1.00520588e-01 -2.64505595e-01 -3.82375926e-01 4.98299181e-01 -6.22857809e-01 1.36672723e+00 -2.32342052e+00 1.78843066e-01 2.68572927e-01 3.32842767e-01 5.01472473e-01 -3.01457226e-01 3.96533497e-03 5.74608929e-02 -2.49733225e-01 1.40992895e-01 -8.14105570e-01 2.24341024e-02 3.15288268e-02 1.40906543e-01 4.36495095e-01 1.11335352e-01 1.02288425e+00 -6.95439160e-01 -6.60306931e-01 1.54777989e-01 2.95382321e-01 -3.46669734e-01 4.00881886e-01 3.84858429e-01 3.71569008e-01 -4.61028337e-01 6.43555224e-01 1.01948154e+00 -2.80369103e-01 -3.18660811e-02 -4.26321834e-01 -2.68659353e-01 -3.16325054e-02 -1.71637321e+00 1.82832062e+00 -7.19447583e-02 1.62109196e-01 -5.30037843e-02 -1.14584887e+00 8.61369073e-01 -8.85990355e-03 3.65477949e-01 -8.91232133e-01 1.02274574e-01 1.13091759e-01 1.02356290e-02 -5.63010395e-01 4.34424371e-01 3.61455172e-01 -6.70701126e-03 3.78791898e-01 1.05477892e-01 9.27466452e-01 1.66033432e-01 -7.86461830e-02 8.06420267e-01 -1.55066466e-02 7.38559067e-02 -1.32893443e-01 1.10948980e+00 -3.41477454e-01 7.83680558e-01 9.06238079e-01 -4.44028854e-01 4.97017741e-01 -3.61485519e-02 -7.92671263e-01 -8.60429943e-01 -1.06688440e+00 -3.10760569e-02 1.44865072e+00 4.22050059e-01 -5.13329208e-01 -6.86832726e-01 -5.97980857e-01 1.05690919e-01 9.52536911e-02 -4.40186024e-01 -2.84217358e-01 -7.46527612e-01 -1.01556814e+00 7.43912935e-01 6.58907175e-01 9.53732312e-01 -7.29896367e-01 -4.43229705e-01 2.38901690e-01 -3.18017572e-01 -1.15849042e+00 -6.89648926e-01 -1.55414432e-01 -3.17020893e-01 -1.03510296e+00 -7.79545963e-01 -9.36274230e-01 6.89799786e-01 6.15559757e-01 8.49481583e-01 3.49470556e-01 -3.47606331e-01 1.89288601e-01 -1.42200738e-01 -5.63492030e-02 1.78747222e-01 1.77516252e-01 4.66318816e-01 3.54873896e-01 7.28373885e-01 -3.84741426e-01 -7.54879892e-01 5.50431132e-01 -4.62589979e-01 1.49469703e-01 6.91447914e-01 9.39699233e-01 5.70503652e-01 2.10296698e-02 4.05705214e-01 -4.51716892e-02 4.25409734e-01 -3.50610793e-01 -7.05017567e-01 6.23817444e-01 -4.76158559e-01 1.72922537e-01 5.06182611e-01 -6.48903370e-01 -9.94991481e-01 3.39778125e-01 -1.29242718e-01 -3.89956862e-01 -1.40730947e-01 1.30051613e-01 -3.03337634e-01 -1.67896330e-01 2.58480579e-01 4.72950011e-01 -1.00541212e-01 -4.80505377e-01 2.03278318e-01 8.13735485e-01 7.01081038e-01 -7.40571320e-01 9.13806498e-01 2.41838917e-01 -2.85467267e-01 -4.54627216e-01 -7.72074401e-01 -6.80972755e-01 -7.02750206e-01 -2.47025102e-01 9.95549262e-01 -1.16009295e+00 -1.13319421e+00 8.86870444e-01 -1.18228662e+00 7.87923783e-02 1.26013845e-01 4.94607478e-01 -4.30234745e-02 5.28874099e-01 -4.21106249e-01 -8.92053366e-01 -5.13103724e-01 -1.27744341e+00 8.41282964e-01 7.15611160e-01 1.99423790e-01 -5.08276999e-01 -9.94257703e-02 4.67821151e-01 4.26049501e-01 -2.98458248e-01 3.51540416e-01 -6.93849385e-01 -6.18783712e-01 -3.53555940e-02 -7.43408084e-01 9.35875699e-02 3.34110414e-03 -4.06986594e-01 -1.05008173e+00 -4.84583020e-01 -2.00087637e-01 -3.54688853e-01 8.38920593e-01 2.20696226e-01 1.14574254e+00 -1.19214989e-01 -4.90775585e-01 8.30312788e-01 1.16538680e+00 1.19898387e-03 2.91318983e-01 5.67955375e-01 6.92099333e-01 5.50147593e-01 5.17277777e-01 5.44437528e-01 8.06779981e-01 8.10210526e-01 5.45648076e-02 -6.07745349e-02 7.42091164e-02 -2.53035516e-01 1.79635957e-01 6.03903532e-01 -5.12451641e-02 2.50776764e-02 -8.15582812e-01 2.78028578e-01 -2.26358652e+00 -9.82673347e-01 1.39785334e-02 2.30234432e+00 5.53927839e-01 1.10891253e-01 3.52858603e-01 -2.64162660e-01 1.01407743e+00 1.14245757e-01 -6.64307773e-01 3.26706976e-01 -1.99693501e-01 -1.28824607e-01 4.50881153e-01 2.71572858e-01 -1.33893132e+00 7.50248849e-01 5.74672747e+00 7.18836427e-01 -8.27755809e-01 2.86820322e-01 6.26929641e-01 -4.46030736e-01 3.34594667e-01 -1.63169980e-01 -1.27521503e+00 8.16986084e-01 6.58385336e-01 1.34393498e-02 4.98400748e-01 7.08594739e-01 6.15053773e-02 -1.45499438e-01 -1.14704740e+00 1.64664328e+00 1.31357893e-01 -1.16884625e+00 -1.16559520e-01 3.75751918e-03 2.55680412e-01 3.35038640e-02 8.10762271e-02 3.74608308e-01 3.47835869e-02 -8.43201280e-01 6.49086595e-01 6.62162781e-01 5.66793859e-01 -8.95756304e-01 8.74129593e-01 4.59156066e-01 -1.75517559e+00 -4.65703577e-01 -5.49814403e-01 1.68306734e-02 2.89266407e-01 4.26256001e-01 -1.32799074e-01 6.75126374e-01 1.07816112e+00 4.71622318e-01 -8.71842504e-01 1.23221314e+00 1.96989030e-01 4.04132307e-02 -4.56275672e-01 4.38114069e-02 1.28054276e-01 -2.50177443e-01 2.84792662e-01 1.34913874e+00 2.34028518e-01 1.25174262e-02 4.96661812e-01 7.49701202e-01 7.46890530e-02 -6.66707009e-02 -1.28304064e-01 4.50918347e-01 6.39853597e-01 1.37540555e+00 -3.85159492e-01 -4.15790021e-01 -6.18009627e-01 1.16105616e+00 7.47275472e-01 2.20368475e-01 -1.19245780e+00 -3.22494090e-01 8.55434597e-01 -3.42561364e-01 2.32425854e-01 -2.50994086e-01 -8.90416801e-02 -1.21185708e+00 4.23576266e-01 -8.08236182e-01 6.05735004e-01 -1.94936380e-01 -1.32331789e+00 4.71695334e-01 -1.16892241e-01 -9.81110215e-01 3.48870158e-02 -4.22078460e-01 -5.32956898e-01 1.11314332e+00 -1.71100819e+00 -1.25601506e+00 -6.26468062e-01 7.71810591e-01 4.28571433e-01 -4.23956245e-01 6.29929483e-01 9.05009508e-01 -1.13857698e+00 1.12309217e+00 -1.74111843e-01 4.15833443e-01 9.73339319e-01 -8.68194103e-01 4.17470694e-01 1.12435496e+00 -1.27283871e-01 8.10038567e-01 2.80787438e-01 -4.88834441e-01 -1.35218799e+00 -1.06177974e+00 8.06318700e-01 -2.95478284e-01 2.52469242e-01 -3.99368972e-01 -9.23653305e-01 2.83482760e-01 -3.76306660e-02 1.82587564e-01 8.00176144e-01 1.60110772e-01 -4.69384730e-01 -3.73381436e-01 -9.36914384e-01 3.73017848e-01 1.36006165e+00 -4.94658381e-01 -5.25476158e-01 2.03823239e-01 5.50907552e-01 -3.55686396e-01 -6.42116487e-01 2.69618005e-01 6.46137595e-01 -9.93456423e-01 1.25254977e+00 -6.13006234e-01 -1.25317141e-01 -6.30208313e-01 -1.78202346e-01 -8.70773017e-01 -8.06301177e-01 -4.36334014e-01 -2.89527237e-01 1.44292819e+00 1.77197829e-01 -6.90163195e-01 6.61979198e-01 8.28257024e-01 1.21277541e-01 -7.06231713e-01 -9.62622702e-01 -6.80540621e-01 -4.67862248e-01 -1.10312872e-01 8.44105780e-01 6.04089975e-01 -3.23019952e-01 3.38799059e-01 -5.90437949e-01 4.54208493e-01 9.39788997e-01 -1.09127816e-02 8.37729573e-01 -1.16073513e+00 -4.38283622e-01 -4.74359393e-01 -3.40107292e-01 -1.38251972e+00 7.10854083e-02 -7.32456386e-01 1.21493870e-02 -1.14212930e+00 6.60214782e-01 -7.10819423e-01 -6.36715293e-01 4.06662226e-01 -5.99527657e-01 -2.92766597e-02 2.13241026e-01 3.31141025e-01 -1.04855561e+00 5.03863513e-01 6.95533633e-01 -9.22131687e-02 -2.77907372e-01 5.67409676e-03 -7.31435359e-01 4.44882870e-01 7.29613543e-01 -3.49122465e-01 -2.50062138e-01 -8.13281476e-01 -1.16947658e-01 -4.01520163e-01 5.84344685e-01 -1.33445954e+00 9.17500377e-01 -3.77231129e-02 7.33620882e-01 -8.21874917e-01 3.83997262e-01 -6.17902100e-01 4.36541177e-02 4.32436556e-01 -1.71400890e-01 4.79232907e-01 6.88628554e-02 4.60942447e-01 -1.65153325e-01 -4.71142791e-02 8.25544000e-01 -1.09762646e-01 -9.27727401e-01 6.30045712e-01 1.58174753e-01 -9.55222026e-02 1.06352377e+00 -3.00553411e-01 -2.21634135e-01 1.12212375e-01 -3.92440438e-01 7.90032923e-01 6.72709495e-02 6.56544864e-01 7.00343311e-01 -1.45888889e+00 -8.00362647e-01 4.15412307e-01 1.05020978e-01 1.26591399e-01 6.06744528e-01 9.01788473e-01 -6.66907849e-03 4.03712451e-01 -3.05453181e-01 -5.88098645e-01 -1.39852214e+00 5.94487309e-01 4.92040902e-01 -2.82241762e-01 -3.02411705e-01 9.61581588e-01 7.67387748e-02 -4.80089635e-01 5.04289746e-01 3.04577470e-01 -2.68846154e-01 1.31033301e-01 9.65690494e-01 5.06645203e-01 -1.74867123e-01 -6.94104552e-01 -6.12976730e-01 5.78582525e-01 -4.11842018e-01 1.07646637e-01 1.33809543e+00 -3.32786560e-01 -2.06695184e-01 -5.19559607e-02 1.20331645e+00 -3.58002007e-01 -1.11515105e+00 -5.27292430e-01 -3.12060416e-02 -4.96149987e-01 -2.30081245e-01 -4.76980388e-01 -9.37476337e-01 6.44625366e-01 9.69091952e-01 -1.60209686e-01 1.06228304e+00 -1.27373338e-01 8.99151623e-01 5.81780851e-01 2.01198280e-01 -1.36597252e+00 4.32660617e-02 4.38040435e-01 3.81424755e-01 -1.37320709e+00 -6.13244213e-02 -2.12550670e-01 -3.28316212e-01 9.61109638e-01 1.14495289e+00 2.24041253e-01 4.57574457e-01 9.29914191e-02 -1.22078054e-01 5.46849556e-02 -4.31600213e-01 -4.38273191e-01 3.56286615e-01 5.69382370e-01 1.06193043e-01 -1.27692163e-01 -1.40756816e-01 8.51752996e-01 1.07352242e-01 -5.91146648e-02 -2.92688996e-01 6.56718194e-01 -3.94785345e-01 -1.08715999e+00 -5.01772821e-01 3.01015943e-01 -1.21494263e-01 -8.67208689e-02 -4.39970940e-02 3.74047637e-01 4.72542733e-01 9.43294883e-01 1.06673792e-01 -6.69650733e-01 3.94867331e-01 1.79159399e-02 2.37377554e-01 -2.21423373e-01 -6.76607370e-01 -1.77467972e-01 -2.51844171e-02 -7.06443608e-01 -3.58825743e-01 -6.15527749e-01 -1.02236021e+00 -3.62762630e-01 -4.18307126e-01 6.12497777e-02 2.97678739e-01 1.07283390e+00 6.48656607e-01 3.78453583e-01 6.43393517e-01 -6.37123585e-01 -7.93675721e-01 -7.60779321e-01 -1.47392213e-01 4.96387839e-01 1.79403275e-01 -8.25075626e-01 3.71354888e-03 -3.64303142e-01]
[14.782841682434082, 0.826081395149231]
36057931-3bb2-4fb7-a048-d9daf9eb942d
abess-a-fast-best-subset-selection-library-in
2110.09697
null
https://arxiv.org/abs/2110.09697v2
https://arxiv.org/pdf/2110.09697v2.pdf
abess: A Fast Best Subset Selection Library in Python and R
We introduce a new library named abess that implements a unified framework of best-subset selection for solving diverse machine learning problems, e.g., linear regression, classification, and principal component analysis. Particularly, the abess certifiably gets the optimal solution within polynomial times with high probability under the linear model. Our efficient implementation allows abess to attain the solution of best-subset selection problems as fast as or even 20x faster than existing competing variable (model) selection toolboxes. Furthermore, it supports common variants like best group subset selection and $\ell_2$ regularized best-subset selection. The core of the library is programmed in C++. For ease of use, a Python library is designed for conveniently integrating with scikit-learn, and it can be installed from the Python library Index. In addition, a user-friendly R library is available at the Comprehensive R Archive Network. The source code is available at: https://github.com/abess-team/abess.
['Xueqin Wang', 'Junxian Zhu', 'Shiyun Lin', 'Yanhang Zhang', 'Kangkang Jiang', 'Junhao Huang', 'Liyuan Hu', 'Jin Zhu']
2021-10-19
null
null
null
null
['sparse-learning']
['methodology']
[-1.51973844e-01 -4.69400227e-01 -4.62839395e-01 -4.32906449e-01 -1.28325522e+00 -5.77790797e-01 -1.81190774e-01 2.31021550e-03 -9.57674384e-02 6.60502374e-01 -3.43279511e-01 -4.69372153e-01 -1.63234085e-01 -6.25737607e-01 -6.67259097e-01 -8.64181161e-01 -7.62207434e-02 5.39857030e-01 -1.55375183e-01 1.08255416e-01 2.79475987e-01 5.39597988e-01 -1.50345886e+00 -1.26077160e-01 9.57433283e-01 1.02420640e+00 4.58431154e-01 4.16424513e-01 1.96690634e-01 6.51695728e-02 -1.11746371e-01 -1.64447337e-01 4.95822489e-01 -3.34819764e-01 -6.46484792e-01 -4.81586754e-01 1.73787922e-01 -6.12198785e-02 -1.15093499e-01 8.53266001e-01 8.64811301e-01 1.24484196e-01 2.60672212e-01 -1.43333423e+00 -3.58786017e-01 3.48319352e-01 -5.74213624e-01 -1.40415188e-02 3.78177941e-01 3.84275287e-01 1.23331034e+00 -1.13573563e+00 3.78883481e-01 1.13942981e+00 4.66911495e-01 4.17790085e-01 -1.45227945e+00 -9.64495540e-01 3.18078585e-02 1.50404587e-01 -1.85392535e+00 -5.67288876e-01 5.95232904e-01 -3.63086760e-01 7.74364591e-01 9.29835320e-01 6.30774498e-01 8.61510217e-01 -7.96807259e-02 1.05588830e+00 1.05367124e+00 -6.84594586e-02 2.54301220e-01 -3.88232395e-02 5.10275662e-01 6.81675971e-01 1.97543770e-01 -4.74540219e-02 -6.59017742e-01 -8.51560175e-01 6.62939131e-01 8.85782465e-02 -3.01241726e-01 -3.15019071e-01 -1.31712830e+00 7.81655550e-01 4.06289995e-02 -2.51326323e-01 -3.57125521e-01 5.17955795e-02 1.90195724e-01 2.69967437e-01 4.25974995e-01 6.20281577e-01 -1.03264344e+00 -6.27777679e-03 -7.89300621e-01 3.78474206e-01 7.98469484e-01 1.07208860e+00 6.74027264e-01 -1.41027838e-01 3.25876400e-02 1.09316671e+00 2.49250531e-01 5.92322171e-01 1.84031099e-01 -1.15381992e+00 -7.65108243e-02 5.38646102e-01 4.84435260e-02 -7.28775680e-01 -7.45814323e-01 -5.90064526e-01 -8.02495956e-01 -1.01732410e-01 2.78271765e-01 -2.10998684e-01 -3.39303434e-01 1.65161479e+00 8.38181019e-01 4.27788526e-01 -4.58031356e-01 8.58529031e-01 9.58745897e-01 4.24828053e-01 -7.06130713e-02 -6.28262043e-01 1.24345779e+00 -6.30288243e-01 -3.58547330e-01 -7.25581869e-02 6.02121294e-01 -7.05405295e-01 1.10225916e+00 6.00848556e-01 -9.68069077e-01 -7.76066780e-02 -7.70980775e-01 -3.64651866e-02 -2.29848891e-01 1.43947020e-01 1.03887188e+00 2.75015473e-01 -8.65996420e-01 4.96218294e-01 -1.15411651e+00 -8.56763497e-02 5.07916927e-01 6.04032457e-01 -4.09189671e-01 -6.89782901e-03 -9.34715509e-01 6.97741568e-01 1.33035153e-01 9.06984434e-02 -7.32589424e-01 -1.07629335e+00 -9.36821401e-01 -1.21066429e-01 5.75441837e-01 -9.50823843e-01 1.06492031e+00 -6.27435565e-01 -1.42742240e+00 8.65749061e-01 -5.34202158e-01 1.10497788e-01 3.83045107e-01 -8.51645842e-02 1.11549772e-01 -1.40829116e-01 4.65073325e-02 2.99245656e-01 5.93939006e-01 -8.68343115e-01 -2.38965198e-01 -5.93675733e-01 -4.20126677e-01 1.42368615e-01 -1.70754164e-01 5.48726857e-01 -6.65507972e-01 -5.92144966e-01 1.63332701e-01 -1.15155315e+00 -6.29603446e-01 1.53028771e-01 -4.83840883e-01 -2.68823713e-01 2.15204686e-01 -7.26308286e-01 1.46459150e+00 -2.12745070e+00 3.19144070e-01 2.64426500e-01 2.38719329e-01 2.45528519e-02 -1.26207232e-01 3.27787787e-01 -4.99190211e-01 2.20056519e-01 -3.68339747e-01 -3.53917062e-01 -2.32491158e-02 -4.00624186e-01 8.91459510e-02 8.91055703e-01 -1.26723289e-01 7.15977848e-01 -7.29839981e-01 -2.78606594e-01 3.89094502e-01 3.74178767e-01 -3.30253929e-01 2.35202000e-01 -2.65712798e-01 5.88978350e-01 -6.20952904e-01 9.43893135e-01 9.84747112e-01 -4.79791939e-01 2.75631487e-01 1.61153838e-01 -2.75899202e-01 1.78039968e-01 -1.53871858e+00 1.48868871e+00 -1.98783010e-01 3.09823722e-01 2.56220281e-01 -6.70367181e-01 9.06345606e-01 1.17925547e-01 6.92437947e-01 -2.30958313e-02 7.81467184e-02 2.93317080e-01 -2.82794744e-01 -2.54044026e-01 -1.13493036e-02 3.38029325e-01 3.74230705e-02 5.57831645e-01 -1.68606579e-01 -1.65254101e-01 1.37362584e-01 1.83464605e-02 9.48685169e-01 -3.76845419e-05 6.93158031e-01 -3.65248919e-01 4.61134285e-01 1.31950304e-01 8.42541873e-01 6.88359976e-01 -9.01571382e-03 6.46475792e-01 5.43802500e-01 -3.54401648e-01 -6.28378510e-01 -8.87173772e-01 -6.94632769e-01 1.29234886e+00 -1.07935667e-01 -5.84757447e-01 -1.88406169e-01 -1.99493930e-01 4.20157999e-01 7.85203040e-01 -4.14737225e-01 7.79874027e-02 -2.93058962e-01 -1.10773635e+00 2.66379923e-01 2.03393579e-01 1.17396377e-01 -8.80388439e-01 1.70403346e-02 -1.28076956e-01 -5.45369089e-02 -5.39914310e-01 -4.57787812e-01 2.85819799e-01 -7.52653122e-01 -1.16948295e+00 -5.90904117e-01 -6.61213100e-01 6.09786749e-01 3.79370600e-01 8.34441245e-01 3.49737555e-01 -5.25414765e-01 2.70562321e-01 -2.86363035e-01 -4.18017089e-01 2.50576437e-01 1.80163518e-01 1.11253008e-01 -2.36086324e-01 4.38162804e-01 -4.03579891e-01 -6.91163719e-01 3.74438882e-01 -4.29954022e-01 9.94667336e-02 1.70046762e-01 9.51671302e-01 1.11607575e+00 -1.06968001e-01 4.60133910e-01 -6.16496027e-01 2.78846383e-01 -8.44877541e-01 -1.10182953e+00 1.59841672e-01 -7.54198253e-01 -1.69851035e-01 5.69367468e-01 -1.45642683e-01 -5.33770025e-01 4.59788620e-01 -3.59829545e-01 -3.31382126e-01 -1.83714703e-01 7.86137164e-01 -2.73071676e-01 -1.82747915e-01 3.27377111e-01 3.60492855e-01 1.77230805e-01 -7.49650061e-01 -5.51507995e-02 7.33720362e-01 2.46430263e-02 -5.64282775e-01 4.46057677e-01 2.36040071e-01 1.55068664e-02 -7.24968374e-01 -6.56861484e-01 -6.44773901e-01 -4.54894662e-01 -1.09526753e-01 2.31150910e-01 -1.00286233e+00 -8.92675996e-01 7.51999974e-01 -6.57037258e-01 -5.54248929e-01 4.26651537e-01 4.80908871e-01 -4.74441200e-01 1.67044654e-01 -3.68713230e-01 -5.74899912e-01 -6.25338018e-01 -1.36950397e+00 1.07534564e+00 2.99616277e-01 -3.34625065e-01 -7.49642909e-01 -9.51493159e-03 2.80338049e-01 1.61526039e-01 2.81977475e-01 6.90676153e-01 -9.04078782e-01 -4.76837993e-01 -4.46883291e-01 1.93131745e-01 1.19495867e-02 -1.20692983e-01 4.96516377e-01 -7.30878770e-01 -4.33500677e-01 -3.64838153e-01 -4.47004959e-02 7.52990842e-01 8.18466902e-01 1.71850967e+00 -4.64667343e-02 -6.16739154e-01 1.24832201e+00 1.20277989e+00 1.08741418e-01 3.61629546e-01 3.43953609e-01 4.10797477e-01 3.81523520e-01 8.00214171e-01 9.02108490e-01 3.12674522e-01 8.03162038e-01 1.99769258e-01 -1.53809831e-01 3.54982018e-01 1.74644396e-01 7.29348436e-02 5.86562514e-01 2.13184878e-01 1.12713948e-01 -1.08261538e+00 1.34365484e-01 -1.97144628e+00 -8.40065479e-01 -6.03480637e-01 2.70808649e+00 1.08135676e+00 -4.98075575e-01 3.36264163e-01 -1.94961041e-01 5.74536026e-01 -1.21775083e-01 -8.75939369e-01 -3.90353091e-02 -2.84972996e-01 1.17543936e-01 5.81636846e-01 6.31750286e-01 -1.18993270e+00 9.22896743e-01 6.26984262e+00 9.68847096e-01 -1.09188437e+00 -3.38837095e-02 8.91568720e-01 -5.79231322e-01 -1.09680310e-01 -1.25293983e-02 -9.55327213e-01 6.72023773e-01 7.73924112e-01 -6.34110212e-01 8.75183225e-01 1.08886290e+00 4.53501493e-01 -2.09219311e-03 -8.67442131e-01 1.02403915e+00 -3.60476166e-01 -1.27823675e+00 -7.34522164e-01 1.15693761e-02 4.06631678e-01 3.06969196e-01 9.50447768e-02 1.78087607e-01 3.28779966e-01 -9.67997253e-01 4.37289566e-01 3.18127900e-01 1.02982807e+00 -6.99174464e-01 5.12909830e-01 3.66997749e-01 -1.08416772e+00 1.48174301e-01 -3.60028982e-01 6.38380200e-02 -8.72513372e-03 8.89182985e-01 -5.35940588e-01 5.84848344e-01 8.79451215e-01 8.11833560e-01 -6.93643153e-01 1.50920260e+00 -6.14189394e-02 6.76035941e-01 -6.12597048e-01 7.35974833e-02 -5.20495296e-01 -3.24594766e-01 6.95513129e-01 1.08047020e+00 3.73312414e-01 2.96121955e-01 2.73143560e-01 5.30149043e-01 3.30105603e-01 3.66705507e-01 -2.38975525e-01 1.57815292e-01 9.69228089e-01 1.45028436e+00 -5.72527826e-01 -7.53846020e-02 -4.31541622e-01 5.81462502e-01 3.31906617e-01 3.91792774e-01 -7.59328663e-01 -4.70584422e-01 8.71892631e-01 -2.94822693e-01 1.84463218e-01 -2.06373915e-01 -6.48173094e-01 -1.14913142e+00 -1.47081420e-01 -1.30775201e+00 7.56838679e-01 -8.00916672e-01 -1.28584707e+00 3.60127300e-01 2.25289062e-01 -1.04249358e+00 1.52233571e-01 -8.16894770e-01 -4.03611273e-01 1.06493247e+00 -8.56479168e-01 -8.29989731e-01 -3.59527767e-01 6.46252394e-01 1.29667267e-01 -1.02528706e-01 1.00562608e+00 2.71959245e-01 -1.42610431e+00 6.62138462e-01 4.82712030e-01 -2.30462596e-01 8.87670279e-01 -1.15535676e+00 8.51514339e-02 6.81640625e-01 -5.59918821e-01 1.13078249e+00 7.52086937e-01 -6.92040205e-01 -1.86776805e+00 -1.07355237e+00 6.71360552e-01 -2.39713147e-01 8.28242898e-01 -3.90906483e-01 -8.59718204e-01 1.03213274e+00 -1.48805484e-01 1.48089141e-01 1.29249966e+00 4.85507101e-01 -1.99550107e-01 -2.61586726e-01 -9.45969701e-01 6.45549417e-01 7.56648362e-01 -1.15017369e-01 1.24683112e-01 7.94632196e-01 3.89621824e-01 -6.76494300e-01 -1.16741288e+00 4.18543726e-01 6.85394406e-01 -4.90405262e-01 1.18133998e+00 -5.89681566e-01 4.94908355e-02 -5.70129156e-01 -1.78888768e-01 -1.09410787e+00 -5.08586586e-01 -9.46769655e-01 -2.38369063e-01 9.74980235e-01 6.43762171e-01 -1.07646930e+00 3.83383363e-01 8.69454622e-01 -1.57153428e-01 -1.08957589e+00 -1.12081182e+00 -6.83987975e-01 1.94255859e-02 -4.77032125e-01 8.59661400e-01 9.43429530e-01 8.80392566e-02 -3.96452798e-03 -3.23974818e-01 2.93712646e-01 6.93225801e-01 3.94982100e-01 9.73038197e-01 -1.12435102e+00 -5.38975954e-01 -7.26917744e-01 -3.30726393e-02 -8.12943220e-01 3.21795702e-01 -1.21028662e+00 -2.51472086e-01 -1.21208096e+00 5.17593682e-01 -8.09148371e-01 -3.99969041e-01 8.92732024e-01 -6.03654742e-01 4.47767042e-02 7.00978339e-02 3.10494035e-01 -6.12997830e-01 3.90723288e-01 1.06742799e+00 1.58261701e-01 -5.12210965e-01 4.09000009e-01 -1.08723974e+00 4.82404888e-01 1.08382392e+00 -6.59378171e-01 1.11879773e-01 -1.63953677e-01 1.20919913e-01 2.74812609e-01 2.49932349e-01 -3.51318389e-01 -2.56215408e-02 -6.44039810e-01 3.88812959e-01 -5.64605236e-01 2.76923478e-01 -5.13989449e-01 7.20701694e-01 3.10514092e-01 -3.34239095e-01 6.03013597e-02 3.59198213e-01 -1.90231409e-02 3.54832172e-01 -2.57991076e-01 7.67261028e-01 -1.14264578e-01 -3.81019920e-01 6.29251897e-01 -1.54323146e-01 -3.26164037e-01 1.17445481e+00 1.93432987e-01 -4.56249952e-01 -2.69850165e-01 -7.17293262e-01 5.77907741e-01 6.57291412e-01 -7.25312275e-05 3.94618958e-01 -1.18802810e+00 -9.06783819e-01 2.28572518e-01 9.76335257e-02 -1.42377866e-02 2.09162712e-01 1.40631711e+00 -4.91721958e-01 4.51329857e-01 2.15300605e-01 -4.95609522e-01 -1.53129864e+00 6.00950420e-01 2.85449207e-01 1.15161605e-01 -3.94911051e-01 1.14000368e+00 2.54361983e-02 -7.25236893e-01 1.39869347e-01 2.73348570e-01 1.53953627e-01 -1.96462184e-01 9.44769859e-01 7.30937064e-01 -1.51133075e-01 -3.81387889e-01 -7.48387635e-01 3.14153820e-01 1.82670099e-03 3.27216715e-01 1.57488179e+00 -3.27897556e-02 -6.93333387e-01 4.24818575e-01 1.23271978e+00 1.75288588e-01 -9.22371089e-01 1.42086726e-02 -2.33439282e-01 -5.15142202e-01 1.56539500e-01 -7.76246548e-01 -1.03376889e+00 3.41982067e-01 2.76062995e-01 -1.27773136e-01 1.34870994e+00 1.00070842e-01 3.03960502e-01 2.48942599e-01 4.85148847e-01 -6.85590565e-01 -7.97118127e-01 4.14428085e-01 1.19451857e+00 -1.45616734e+00 5.05754471e-01 -4.67146814e-01 -6.67009413e-01 1.01075828e+00 5.12695253e-01 -1.27349600e-01 9.00401354e-01 3.91078293e-01 -1.13117807e-02 -9.05809272e-03 -1.06099033e+00 -9.82573479e-02 2.78898329e-01 3.43916684e-01 6.79832399e-01 3.01989198e-01 -4.87148196e-01 1.14075506e+00 -3.60569090e-01 -1.02364151e-02 4.33942750e-02 5.48379481e-01 -7.90862143e-02 -1.10901916e+00 -6.22449994e-01 7.89960623e-01 -4.34549689e-01 -3.77654016e-01 -2.58704305e-01 4.69848692e-01 -3.37698668e-01 8.38203907e-01 -1.54393166e-01 -4.71949577e-01 -8.67281482e-02 -8.33349898e-02 9.22886804e-02 -5.13191640e-01 -6.06273830e-01 4.73243713e-01 -1.65412773e-03 -8.90353262e-01 1.94415182e-01 -1.25816703e+00 -1.10321319e+00 -4.20254141e-01 -3.93008202e-01 1.91272467e-01 5.73515773e-01 5.72196603e-01 6.86513484e-01 6.95520043e-02 7.14627802e-01 -8.16675246e-01 -4.32998896e-01 -7.19860077e-01 -6.04293168e-01 -1.94477275e-01 1.77007124e-01 -7.04221606e-01 -5.42240679e-01 -2.82546461e-01]
[7.305428981781006, 4.387258052825928]
6eb21137-e74e-4c5f-971f-5641b3811b1e
simple-unsupervised-summarization-by-1
1907.13337
null
https://arxiv.org/abs/1907.13337v1
https://arxiv.org/pdf/1907.13337v1.pdf
Simple Unsupervised Summarization by Contextual Matching
We propose an unsupervised method for sentence summarization using only language modeling. The approach employs two language models, one that is generic (i.e. pretrained), and the other that is specific to the target domain. We show that by using a product-of-experts criteria these are enough for maintaining continuous contextual matching while maintaining output fluency. Experiments on both abstractive and extractive sentence summarization data sets show promising results of our method without being exposed to any paired data.
['Alexander M. Rush', 'Jiawei Zhou']
2019-07-31
simple-unsupervised-summarization-by
https://aclanthology.org/P19-1503
https://aclanthology.org/P19-1503.pdf
acl-2019-7
['abstractive-sentence-summarization', 'unsupervised-sentence-summarization']
['natural-language-processing', 'natural-language-processing']
[ 4.98675883e-01 7.01626658e-01 -1.18122727e-01 -4.72381055e-01 -1.08811128e+00 -4.17107821e-01 5.68113565e-01 6.42553210e-01 -5.87069690e-01 7.20572054e-01 7.89549828e-01 -1.20123342e-01 1.97999701e-01 -6.12249494e-01 -5.84387720e-01 -3.36896479e-01 3.33529770e-01 5.41226864e-01 2.68394172e-01 -5.40029526e-01 6.39983058e-01 1.71818107e-01 -1.36147964e+00 6.67794585e-01 1.35806525e+00 4.66761440e-01 3.18123937e-01 1.08480036e+00 -5.10153353e-01 1.02562582e+00 -1.02826834e+00 -4.99227345e-01 1.22963607e-01 -7.30944753e-01 -1.06184113e+00 3.55632365e-01 5.56960821e-01 1.10328503e-01 -1.77646086e-01 9.79885817e-01 5.89785039e-01 4.26002502e-01 7.03067482e-01 -5.40299296e-01 -4.83604431e-01 1.00514770e+00 -7.44918659e-02 2.80902207e-01 5.64553559e-01 3.11287865e-02 1.09154010e+00 -6.96323514e-01 8.05448294e-01 1.05369496e+00 5.12837231e-01 4.97173876e-01 -1.37041306e+00 -8.98997262e-02 2.21780732e-01 -2.06346259e-01 -9.59295392e-01 -8.32593679e-01 8.55475843e-01 -3.68582010e-01 1.50682068e+00 4.18570131e-01 2.90573359e-01 9.25877452e-01 3.71625930e-01 8.31597209e-01 8.17284644e-01 -8.49970400e-01 3.55545968e-01 3.29823285e-01 5.87698400e-01 5.12550414e-01 2.14014187e-01 -2.95382917e-01 -6.57242656e-01 -1.79716200e-01 1.19857810e-01 -5.23836195e-01 -2.14192733e-01 -2.41225317e-01 -8.01212728e-01 8.40119243e-01 -1.84141710e-01 6.09015942e-01 -4.15030152e-01 -2.26089790e-01 7.42594063e-01 5.56008279e-01 8.08440804e-01 8.06561232e-01 -3.69202197e-01 -9.53840278e-03 -1.52263021e+00 2.53311247e-01 1.02056587e+00 1.09953320e+00 5.19482732e-01 1.09242171e-01 -3.87793452e-01 9.52263653e-01 -1.04171365e-01 4.74008590e-01 8.48228395e-01 -8.05387855e-01 8.53454590e-01 6.50346875e-01 7.20946118e-02 -8.73028100e-01 -3.35071921e-01 -3.98501605e-01 -6.91547275e-01 -3.27271342e-01 -3.40909660e-02 -1.10560849e-01 -8.40000033e-01 1.79518855e+00 -3.00796419e-01 -1.35251820e-01 3.32774222e-01 1.56036496e-01 9.89922285e-01 7.38611400e-01 1.28234491e-01 -7.39651024e-01 1.01904941e+00 -1.12169671e+00 -8.82406235e-01 -4.77962881e-01 8.80256116e-01 -6.81779861e-01 1.18716812e+00 3.43850106e-01 -1.56600511e+00 -4.94258761e-01 -1.26172888e+00 -2.30046168e-01 -2.10304439e-01 2.86077708e-01 2.28007078e-01 7.36562431e-01 -1.29642200e+00 6.91955268e-01 -6.09395027e-01 -7.31834054e-01 5.10584600e-02 3.98698747e-01 -4.18599278e-01 3.68452102e-01 -1.07350111e+00 1.12663734e+00 7.65561104e-01 -3.52474719e-01 -5.46107948e-01 -2.84719437e-01 -8.83530676e-01 1.74871430e-01 2.44859606e-01 -1.06762755e+00 1.57585979e+00 -1.29069638e+00 -1.94008076e+00 8.85317326e-01 -3.57393682e-01 -7.18896627e-01 5.05209267e-01 -3.58307779e-01 -2.08867237e-01 4.87626135e-01 1.09539777e-01 3.85608822e-01 7.82336712e-01 -1.30327380e+00 -4.66587842e-01 -1.32388070e-01 -1.16983719e-01 4.12179440e-01 -6.09313905e-01 2.54675984e-01 -1.94965824e-01 -8.53556573e-01 -1.75521269e-01 -5.86280525e-01 -4.51396793e-01 -7.55189300e-01 -7.34213471e-01 -2.92398304e-01 4.62630689e-01 -9.58561361e-01 1.53353655e+00 -1.62697136e+00 4.40409124e-01 7.29513168e-02 -6.70985729e-02 3.70241106e-01 -3.33302587e-01 8.50497305e-01 -1.77840441e-01 3.16277444e-01 -6.42584145e-01 -7.76992977e-01 -3.01444769e-01 4.84943613e-02 -4.85545009e-01 9.78625491e-02 1.73298255e-01 7.85385311e-01 -1.10396636e+00 -7.12043524e-01 -4.24841009e-02 -1.94002360e-01 -6.54220760e-01 4.49013293e-01 -3.37076217e-01 1.45611659e-01 -3.65044415e-01 -8.92421156e-02 2.94796735e-01 1.87565580e-01 3.10845047e-01 -9.43638608e-02 -7.95511156e-02 6.13721132e-01 -9.92606640e-01 2.09511065e+00 -5.21293283e-01 4.69155699e-01 -2.51709163e-01 -1.13980460e+00 9.47151899e-01 3.66957337e-01 1.13576800e-01 -4.35612947e-01 2.70445019e-01 1.13050029e-01 -2.24372670e-01 -6.25032723e-01 1.07958686e+00 -3.79265219e-01 -3.02577287e-01 7.18537390e-01 6.04804575e-01 -4.85529959e-01 6.10651433e-01 7.01468587e-01 1.10637605e+00 -1.09151550e-01 5.61496794e-01 -6.90760970e-01 4.79975641e-01 1.09052122e-01 5.46426475e-01 9.74094689e-01 1.72942877e-01 6.35050237e-01 5.90687513e-01 2.02547401e-01 -1.16945827e+00 -9.42573547e-01 4.31490004e-01 1.01612067e+00 -1.46915102e-02 -8.39039147e-01 -1.08028984e+00 -7.30990410e-01 -4.31578487e-01 1.45417511e+00 -4.05334622e-01 -4.18646753e-01 -6.52737617e-01 -2.71395087e-01 6.19990468e-01 5.55339992e-01 2.09561005e-01 -1.11551237e+00 -7.32519269e-01 2.48788357e-01 -3.00332040e-01 -8.44988406e-01 -5.93712807e-01 1.80824205e-01 -1.20141912e+00 -6.18688762e-01 -4.30947512e-01 -8.89711797e-01 7.66642928e-01 1.73469827e-01 1.35094905e+00 -4.66021486e-02 3.55689615e-01 5.51590323e-01 -2.96973020e-01 -5.04681766e-01 -9.20503795e-01 4.67006177e-01 9.01754573e-02 -1.72702864e-01 1.42842799e-01 -6.90478861e-01 -1.03777468e-01 -4.06421840e-01 -9.92931664e-01 1.46410376e-01 7.57201850e-01 6.93113863e-01 2.74075091e-01 -8.05234537e-02 7.37975001e-01 -1.32986307e+00 1.20144272e+00 -1.68495327e-01 7.19705448e-02 5.08960783e-01 -4.08274144e-01 2.22502694e-01 8.61038387e-01 -2.86366701e-01 -1.26235950e+00 -2.94288970e-03 1.60839744e-02 2.43734680e-02 -2.32600898e-01 7.30623484e-01 -3.57037008e-01 4.64211583e-01 9.24157858e-01 3.74992222e-01 -9.93196815e-02 -5.98363340e-01 6.03007913e-01 7.73031652e-01 5.91147900e-01 -5.01410604e-01 5.92042029e-01 1.52238175e-01 -5.51548004e-01 -1.33652258e+00 -7.05597997e-01 -4.77790743e-01 -8.13647628e-01 -6.02057986e-02 8.35016191e-01 -6.96075797e-01 2.48064641e-02 2.09769428e-01 -1.40409386e+00 -1.61852092e-01 -7.74003565e-01 1.96950004e-01 -7.41681337e-01 7.61055171e-01 -7.60737717e-01 -7.70491600e-01 -7.59880424e-01 -8.27629030e-01 8.94190490e-01 2.05512196e-01 -6.25172377e-01 -1.07568502e+00 3.93790185e-01 2.50575602e-01 2.84112990e-01 -8.02615657e-02 8.41895819e-01 -1.34718955e+00 1.98115166e-02 -3.05223674e-01 3.18009287e-01 7.69804895e-01 1.80490255e-01 1.19811870e-01 -7.67824650e-01 -1.77452669e-01 3.88447225e-01 -4.17222381e-01 1.19905317e+00 2.64077187e-01 7.37302244e-01 -7.39278793e-01 -1.44759268e-01 -8.53816643e-02 1.05460322e+00 -6.82248622e-02 7.13698983e-01 3.41557823e-02 5.08475840e-01 6.66080058e-01 3.26649815e-01 1.41468436e-01 3.32790792e-01 4.27028120e-01 -1.89993963e-01 -9.46601033e-02 7.65083730e-02 -5.29024839e-01 5.97889066e-01 1.38776076e+00 1.60976008e-01 -4.74713564e-01 -7.97973871e-01 7.91943252e-01 -2.03224039e+00 -1.17865670e+00 -2.72858087e-02 1.98394871e+00 1.06396759e+00 3.64920914e-01 1.93316892e-01 -4.59918678e-02 6.44002378e-01 2.82269984e-01 -2.70196617e-01 -1.12903416e+00 -3.43593627e-01 3.62314165e-01 2.33224586e-01 5.87329984e-01 -1.05694246e+00 1.10370541e+00 7.22662735e+00 8.44782054e-01 -7.72089779e-01 -9.48871672e-02 4.65079337e-01 -9.92900208e-02 -6.67955935e-01 1.63986161e-01 -3.73895973e-01 2.12377399e-01 1.18857801e+00 -5.97955108e-01 1.39643354e-02 5.77052593e-01 3.94040495e-01 -2.86876768e-01 -1.05384052e+00 6.14000082e-01 4.42611337e-01 -1.29659104e+00 5.19228339e-01 -4.99017656e-01 8.32393587e-01 -3.57296556e-01 -3.57602894e-01 3.70007306e-01 1.53840482e-01 -8.34158242e-01 8.63700330e-01 6.57511532e-01 3.11172843e-01 -8.13727856e-01 7.51531720e-01 7.23741293e-01 -6.92038953e-01 2.25144312e-01 -3.27661842e-01 -9.79050919e-02 2.69581616e-01 4.70769197e-01 -5.02961099e-01 9.45785522e-01 -4.22859713e-02 6.85238600e-01 -8.37105036e-01 8.34913790e-01 -4.27073658e-01 7.25735307e-01 -1.60796583e-01 6.30458491e-03 -1.84791852e-02 -1.54729098e-01 1.04932630e+00 1.69962931e+00 1.05018921e-01 4.81285341e-02 4.61432636e-01 4.60926890e-01 -6.55726194e-02 6.22125328e-01 -6.76763058e-01 -1.03720672e-01 2.94449776e-01 1.10698032e+00 -5.78950584e-01 -7.70648539e-01 -6.29795566e-02 1.14727104e+00 3.48669112e-01 3.29310298e-01 -2.55735189e-01 -7.05874383e-01 -9.56885293e-02 -8.08995664e-02 1.89176321e-01 -1.71551347e-01 -5.41041672e-01 -1.45709956e+00 -2.85093486e-02 -9.35378671e-01 3.92200977e-01 -5.86433411e-01 -1.32194138e+00 7.28639245e-01 3.43281299e-01 -9.96320486e-01 -4.05650914e-01 -2.74297655e-01 -9.54356670e-01 8.30388010e-01 -1.05862045e+00 -1.12983501e+00 2.74970710e-01 2.04885572e-01 8.71274054e-01 -3.23830664e-01 8.51362050e-01 -1.47272617e-01 -5.86320758e-01 5.11537492e-01 -6.95836693e-02 -2.63356209e-01 7.57992148e-01 -1.39374769e+00 2.92195678e-01 1.20463192e+00 2.24506974e-01 8.43634725e-01 1.31041312e+00 -7.27681339e-01 -1.02688754e+00 -1.02495515e+00 1.58059943e+00 -5.16197681e-01 4.40515220e-01 -3.17663044e-01 -1.05882275e+00 6.34105980e-01 7.90245116e-01 -9.50492680e-01 9.59211707e-01 5.44683002e-02 -2.29283094e-01 9.91485119e-02 -1.10472834e+00 6.33359194e-01 8.91717672e-01 -4.93781120e-01 -1.46926045e+00 6.00204229e-01 9.20313776e-01 -2.96829641e-01 -5.57561159e-01 3.01882505e-01 6.88300282e-02 -8.89478922e-01 4.84872609e-01 -1.03286314e+00 6.85793638e-01 1.60112511e-02 -7.10848868e-02 -1.48028231e+00 -1.01029441e-01 -6.13324404e-01 8.92896764e-03 1.51900184e+00 6.67862296e-01 -3.41188937e-01 5.35098195e-01 5.15447199e-01 -5.10761917e-01 -5.00617385e-01 -5.98346412e-01 -9.68746245e-01 2.85036564e-01 -2.77979136e-01 8.59396383e-02 6.85964108e-01 4.70465869e-01 1.15091252e+00 -1.95806369e-01 -1.96788087e-01 2.00152427e-01 1.45308271e-01 7.98072577e-01 -9.64672923e-01 -3.72547626e-01 -6.25854731e-01 -1.63035572e-01 -1.08149576e+00 5.46207726e-01 -1.08764386e+00 1.99042603e-01 -1.95592439e+00 4.34564233e-01 3.19544733e-01 -1.78123072e-01 3.28889430e-01 -2.81755954e-01 -2.09027395e-01 1.53216392e-01 8.16132501e-02 -7.98826754e-01 6.74705744e-01 5.76535404e-01 -2.36866236e-01 -6.72590494e-01 1.55328913e-02 -9.22684312e-01 6.63703561e-01 1.05852807e+00 -4.57558841e-01 -5.90619981e-01 -3.75498503e-01 -8.32401589e-02 -3.75272594e-02 -1.30390450e-02 -7.66416848e-01 3.94732416e-01 -8.09043124e-02 -1.13490723e-01 -6.23939216e-01 -7.11141825e-02 -2.48432398e-01 -1.76871255e-01 3.44719619e-01 -9.02874887e-01 1.09254308e-01 2.64466494e-01 4.60951358e-01 -2.49954417e-01 -6.74310982e-01 7.49233365e-01 -1.61414817e-01 -3.86963457e-01 -2.81510293e-01 -6.77314818e-01 3.35797846e-01 6.12540305e-01 1.22942878e-02 -2.20170259e-01 -6.74090087e-01 -4.67993140e-01 1.94958001e-01 5.74294567e-01 2.75429130e-01 4.96218145e-01 -9.47933137e-01 -1.02361178e+00 -2.39000112e-01 1.47828385e-01 -2.92038113e-01 9.98142138e-02 5.27466595e-01 -3.60193849e-01 4.58056360e-01 1.58696577e-01 -3.66168857e-01 -1.33927035e+00 5.97543895e-01 7.51385912e-02 -7.27765739e-01 -6.64905131e-01 5.57324469e-01 -1.31267369e-01 -4.64266777e-01 9.10653993e-02 -2.76049614e-01 -3.64003032e-01 2.96863914e-01 4.18999344e-01 2.52917171e-01 3.04908723e-01 -5.84416866e-01 -2.62775630e-01 2.71585822e-01 -2.01138049e-01 -5.15543640e-01 1.21591091e+00 -1.36283949e-01 -2.95561284e-01 8.23549151e-01 8.76477659e-01 3.90163034e-01 -5.15800297e-01 -1.85816407e-01 5.89369237e-01 9.17745754e-02 -9.32177156e-02 -7.20989347e-01 -5.01089096e-01 8.70448709e-01 -2.67140940e-02 3.61047447e-01 1.17218339e+00 -2.60613918e-01 6.36730731e-01 7.53993034e-01 9.35261548e-02 -1.44710505e+00 1.51074314e-02 7.16898739e-01 1.13612270e+00 -7.28471696e-01 1.95246667e-01 -3.10363084e-01 -8.19226503e-01 1.07135439e+00 3.14427495e-01 -3.91136259e-01 1.76948369e-01 7.08514750e-02 -4.54082787e-02 6.84732124e-02 -1.02874863e+00 -1.15828671e-01 3.99723858e-01 3.84613514e-01 4.91090953e-01 -1.09495133e-01 -8.24887574e-01 9.23967779e-01 -4.20932651e-01 -1.69050604e-01 6.46433353e-01 1.13600421e+00 -7.34333396e-01 -1.29261196e+00 1.00637460e-02 7.21076488e-01 -4.02259141e-01 -2.23296925e-01 -8.34639728e-01 5.64510345e-01 -3.39446545e-01 1.29165876e+00 -6.95353821e-02 -1.86778232e-01 7.54391730e-01 3.38196307e-01 5.96566856e-01 -1.14301860e+00 -1.04045177e+00 5.55250198e-02 5.96890450e-01 -1.64191425e-01 -4.14837629e-01 -5.26682556e-01 -1.05609810e+00 1.41183985e-02 -3.78318310e-01 5.36522567e-01 4.36522961e-01 9.95407820e-01 4.32258219e-01 4.19843495e-01 6.99168801e-01 -7.53406107e-01 -8.65506589e-01 -1.24379456e+00 -3.51993412e-01 4.30162013e-01 6.59664199e-02 2.70369574e-02 -2.16007099e-01 3.61277103e-01]
[12.486316680908203, 9.48287296295166]
ce625b41-6de1-48af-8cf9-13b676ebf8bb
tell-me-how-to-ask-again-question-data
2010.01475
null
https://arxiv.org/abs/2010.01475v1
https://arxiv.org/pdf/2010.01475v1.pdf
Tell Me How to Ask Again: Question Data Augmentation with Controllable Rewriting in Continuous Space
In this paper, we propose a novel data augmentation method, referred to as Controllable Rewriting based Question Data Augmentation (CRQDA), for machine reading comprehension (MRC), question generation, and question-answering natural language inference tasks. We treat the question data augmentation task as a constrained question rewriting problem to generate context-relevant, high-quality, and diverse question data samples. CRQDA utilizes a Transformer autoencoder to map the original discrete question into a continuous embedding space. It then uses a pre-trained MRC model to revise the question representation iteratively with gradient-based optimization. Finally, the revised question representations are mapped back into the discrete space, which serve as additional question data. Comprehensive experiments on SQuAD 2.0, SQuAD 1.1 question generation, and QNLI tasks demonstrate the effectiveness of CRQDA
['Ming Zhou', 'Nan Duan', 'Jiancheng Lv', 'Jiusheng Chen', 'Yu Yan', 'Jie Fu', 'Yeyun Gong', 'Dayiheng Liu']
2020-10-04
null
https://aclanthology.org/2020.emnlp-main.467
https://aclanthology.org/2020.emnlp-main.467.pdf
emnlp-2020-11
['question-rewriting']
['natural-language-processing']
[ 5.69194734e-01 4.51230049e-01 2.80785978e-01 -6.32399797e-01 -9.82369840e-01 -5.59064448e-01 4.81064200e-01 2.54960895e-01 -3.48228157e-01 7.17485249e-01 5.05653679e-01 -7.55416155e-01 -8.33128113e-03 -1.04714119e+00 -7.21030235e-01 -2.84113400e-02 7.65799105e-01 4.28661495e-01 6.86565787e-02 -6.75811231e-01 -8.19566473e-02 -2.59299781e-02 -1.46510792e+00 1.67847559e-01 1.28100097e+00 7.57671952e-01 2.77952790e-01 9.79741096e-01 -4.69254881e-01 1.19761348e+00 -8.90515506e-01 -5.65287530e-01 -1.59284726e-01 -9.77084517e-01 -1.40966070e+00 -2.60542314e-02 4.47243869e-01 -5.71362495e-01 -4.20721978e-01 7.30356455e-01 2.32520387e-01 6.47479713e-01 3.16884220e-01 -1.12771511e+00 -1.62490666e+00 3.32327843e-01 3.11993718e-01 2.49124929e-01 8.47965360e-01 2.24108890e-01 1.22159123e+00 -9.57910001e-01 5.10971367e-01 1.33735967e+00 1.29784539e-01 8.22492898e-01 -1.23730981e+00 -2.49654457e-01 1.28888384e-01 3.40936989e-01 -7.27181613e-01 -8.25872868e-02 9.68770087e-01 3.60193774e-02 8.97837579e-01 3.77404243e-01 5.26524663e-01 1.04438162e+00 7.51746222e-02 8.64266157e-01 9.54747379e-01 -6.20867610e-01 2.43117541e-01 -9.00662839e-02 5.60414732e-01 7.80416131e-01 -2.56098628e-01 -1.58942118e-01 -1.44424558e-01 -1.96149021e-01 5.54562151e-01 -1.49619892e-01 -4.34039891e-01 -1.01235330e-01 -1.03255844e+00 1.22531509e+00 5.34004092e-01 6.98930547e-02 -4.92476195e-01 -5.12813181e-02 2.37338185e-01 8.50683689e-01 2.42641568e-01 1.11198390e+00 -6.80013657e-01 -9.35731754e-02 -1.21454597e-01 7.74615109e-01 8.76552343e-01 9.58367586e-01 7.26016402e-01 1.39738292e-01 -7.81680465e-01 9.19661760e-01 3.59055281e-01 5.41424453e-01 6.90992713e-01 -1.08496332e+00 6.20128930e-01 9.54802215e-01 -3.48908082e-02 -8.83634746e-01 -1.50303245e-01 -2.04051077e-01 -7.34453857e-01 -1.75109610e-01 1.48767740e-01 -2.43422762e-01 -9.82421994e-01 1.81576252e+00 5.44677079e-01 -2.55072325e-01 6.24876201e-01 7.75928974e-01 1.44930053e+00 1.12493896e+00 3.02750971e-02 2.86078900e-01 1.57701051e+00 -1.26208699e+00 -9.86622512e-01 -2.47722939e-01 3.23738009e-01 -3.00114900e-01 1.74454582e+00 4.31579873e-02 -1.12495768e+00 -8.66468072e-01 -9.39431429e-01 -9.49966311e-01 -3.91477704e-01 2.34250650e-02 1.67763427e-01 1.30302340e-01 -8.98728311e-01 4.33687046e-02 -2.72205949e-01 -5.52954488e-02 2.46542722e-01 1.00414559e-01 -9.32015777e-02 -2.65353829e-01 -1.61554158e+00 7.64277399e-01 2.68315583e-01 -9.19139162e-02 -1.05715847e+00 -7.62424707e-01 -1.40071893e+00 1.48295045e-01 4.13161159e-01 -8.08368742e-01 1.73366642e+00 -5.65684497e-01 -1.85216177e+00 7.33518362e-01 -2.16624409e-01 -4.22593504e-01 -3.09142739e-01 -2.31465504e-01 -4.43000704e-01 1.99581906e-01 1.53528377e-01 7.81702757e-01 8.57428312e-01 -1.03380239e+00 -2.09918261e-01 -4.68688868e-02 6.84172392e-01 3.07824582e-01 3.88339125e-02 -1.82062358e-01 -4.60953936e-02 -7.50603914e-01 -8.36039335e-02 -4.76430207e-01 -3.42995435e-01 -2.17276156e-01 -3.00855666e-01 -8.69529843e-01 6.15670979e-01 -1.08806181e+00 1.06680155e+00 -1.86438954e+00 4.39079791e-01 -8.42708051e-02 3.62950802e-01 3.55970263e-01 -8.66911232e-01 4.07726288e-01 -3.58616486e-02 -6.61196234e-03 -4.66277570e-01 -1.29554227e-01 1.12922162e-01 4.50300992e-01 -6.14154398e-01 -3.60207081e-01 8.65680814e-01 1.42137110e+00 -1.14951420e+00 -1.74310625e-01 -8.88779238e-02 -5.46103045e-02 -6.63801193e-01 1.06823575e+00 -1.03941023e+00 2.58882016e-01 -6.00715160e-01 3.49640578e-01 3.93970937e-01 -3.35878402e-01 -3.08608204e-01 -5.21744415e-03 4.03303802e-01 6.86649740e-01 -4.41551566e-01 1.81131256e+00 -5.98765910e-01 5.94991148e-01 -2.74934560e-01 -7.45982528e-01 1.48152828e+00 9.63757485e-02 -3.18528563e-01 -1.06111217e+00 1.17117621e-01 -3.05616796e-01 -2.23344684e-01 -1.01495671e+00 9.84761655e-01 -3.79298143e-02 -2.04277977e-01 6.67568088e-01 3.47942084e-01 -5.33617795e-01 4.74703163e-02 2.27489561e-01 9.73877847e-01 1.89772099e-01 1.67580172e-01 -3.17694619e-02 7.24664032e-01 8.98589268e-02 1.36682466e-01 5.90295434e-01 9.95095354e-03 6.10268295e-01 5.58494031e-01 -1.72947630e-01 -1.04472423e+00 -1.28057575e+00 2.97371358e-01 1.24671721e+00 -5.05029336e-02 -2.81207770e-01 -9.25405979e-01 -9.23266172e-01 -5.58634773e-02 1.06609464e+00 -7.95948327e-01 -6.28836691e-01 -6.94952369e-01 -1.92263220e-02 5.19524455e-01 4.87751245e-01 6.56760335e-01 -1.44443345e+00 -3.34338307e-01 3.42680693e-01 -4.24034476e-01 -9.70156670e-01 -5.67880332e-01 -1.28141388e-01 -6.16019130e-01 -9.97279823e-01 -5.71602225e-01 -1.27173162e+00 6.70317531e-01 -3.19144055e-02 1.48176801e+00 2.31235951e-01 1.61503419e-01 4.64858294e-01 -7.20812738e-01 -1.68488652e-01 -6.52432680e-01 3.24620634e-01 -4.13337290e-01 -6.55591935e-02 3.02563906e-01 -3.52706224e-01 -4.92862850e-01 -9.29311439e-02 -1.16652668e+00 7.55523816e-02 2.25224793e-01 1.21965444e+00 6.98118031e-01 -7.06492901e-01 1.25052881e+00 -7.71151543e-01 1.61817145e+00 -7.07795203e-01 -5.08838356e-01 5.45829415e-01 -3.77968013e-01 3.71825367e-01 9.97945070e-01 -4.73521233e-01 -1.18699241e+00 -5.19543886e-01 -5.40179729e-01 -2.21008688e-01 -1.96660742e-01 7.08315194e-01 -4.20551807e-01 4.24951881e-01 8.46888125e-01 5.15028715e-01 2.24047467e-01 -4.56185073e-01 8.84229422e-01 7.90662944e-01 9.08413529e-01 -5.51951528e-01 9.66973782e-01 -2.56756693e-01 -5.11844754e-01 -4.94408101e-01 -1.31493485e+00 -1.82252135e-02 -3.03901345e-01 9.70969275e-02 1.28628886e+00 -6.48575127e-01 -5.65718174e-01 3.21101338e-01 -1.31812239e+00 -5.31651497e-01 -8.72121453e-01 -9.88179818e-02 -4.69421923e-01 2.30995044e-01 -4.93696690e-01 -4.68479395e-01 -7.89999843e-01 -9.57353592e-01 9.04851377e-01 7.54812241e-01 -2.80252129e-01 -8.46000969e-01 3.60497206e-01 8.11338246e-01 4.24467683e-01 2.21982017e-01 1.42649019e+00 -9.45656538e-01 -6.32754683e-01 -8.98221657e-02 -7.33555928e-02 7.85899997e-01 2.91432023e-01 -4.86445516e-01 -5.19271374e-01 -3.78247611e-02 2.53609389e-01 -9.74310100e-01 3.43771249e-01 -3.52327973e-01 1.35505879e+00 -4.20171142e-01 5.81382096e-01 4.54780310e-01 9.37282681e-01 1.05681546e-01 5.66693068e-01 1.94011509e-01 5.75718224e-01 4.55371529e-01 6.72177672e-01 -1.99033581e-02 1.03531313e+00 2.72126734e-01 3.95563722e-01 -5.42708486e-03 -1.46623582e-01 -6.63920522e-01 1.50515949e-02 1.23634899e+00 7.82991648e-01 -1.91755086e-01 -7.40618467e-01 7.33494341e-01 -1.34761560e+00 -5.44656754e-01 -4.37950231e-02 1.58168733e+00 1.30316234e+00 -1.73459321e-01 -2.50660151e-01 -2.52950974e-02 2.99917072e-01 1.55353069e-01 -8.82441103e-01 -7.29462862e-01 1.16410619e-02 7.68756270e-01 -5.53067625e-01 6.09369636e-01 -5.40752351e-01 8.96902621e-01 6.11004877e+00 2.16279283e-01 -5.40329814e-01 4.62728702e-02 3.77983809e-01 3.85060132e-01 -9.61744428e-01 3.11567411e-02 -3.08553815e-01 1.71936944e-01 1.00206888e+00 -5.00236452e-01 4.87245530e-01 6.65359497e-01 -2.30709165e-01 1.06490716e-01 -1.03115630e+00 6.86907709e-01 4.05972570e-01 -1.24799573e+00 5.89405537e-01 -5.69781959e-01 6.13871098e-01 -4.04508710e-01 -2.97405627e-02 9.19759929e-01 4.98608232e-01 -1.12714779e+00 1.33911967e-01 8.38270843e-01 5.54801106e-01 -8.10577512e-01 5.66782296e-01 3.45188498e-01 -7.73160875e-01 -1.60628080e-01 -4.64662224e-01 -9.08405334e-02 2.24581212e-01 5.17286621e-02 -6.21407986e-01 6.65413916e-01 3.74029040e-01 2.80012369e-01 -1.00750279e+00 4.65445310e-01 -6.88067436e-01 8.01248074e-01 -1.19519820e-02 -4.64794397e-01 1.76802516e-01 -1.83182746e-01 2.55308986e-01 4.20948446e-01 -2.64201939e-01 6.22990668e-01 -3.83675918e-02 1.18804264e+00 -5.70078313e-01 1.22742593e-01 -1.31737649e-01 -1.28121719e-01 4.19477850e-01 1.04465199e+00 4.11680996e-01 -4.27714884e-01 -3.41245025e-01 1.02499461e+00 7.06830263e-01 4.97745007e-01 -6.06146872e-01 -9.07968342e-01 5.30656338e-01 -3.07857156e-01 9.76990610e-02 -6.55923635e-02 1.70730799e-01 -1.52337360e+00 2.77219027e-01 -1.46146560e+00 3.17283988e-01 -1.33900952e+00 -1.28782856e+00 8.26560676e-01 -6.92726299e-02 -7.88684785e-01 -4.82813269e-01 -2.91053712e-01 -8.44687939e-01 1.18734396e+00 -1.62935221e+00 -9.32580590e-01 -5.57074547e-01 5.68390131e-01 7.76120245e-01 -3.03216100e-01 1.05348051e+00 -4.82597351e-02 -5.20142972e-01 9.09593642e-01 -7.86393583e-02 3.15835267e-01 2.63730407e-01 -1.52136552e+00 1.00216615e+00 5.56678593e-01 -8.09433907e-02 4.79614258e-01 3.78363132e-01 -4.50470388e-01 -1.44545686e+00 -1.40047896e+00 9.50551391e-01 -5.09803712e-01 6.39466703e-01 -5.05860806e-01 -1.57996964e+00 6.91066444e-01 6.71544433e-01 -2.94668227e-01 8.80960643e-01 -1.30296379e-01 -3.12254936e-01 -4.91480194e-02 -1.09036112e+00 8.29288006e-01 6.57064259e-01 -8.76852036e-01 -1.63432312e+00 2.25601152e-01 1.62578106e+00 -4.63938534e-01 -1.23982906e+00 2.60403425e-01 -1.04991272e-01 -1.93525195e-01 7.20589578e-01 -1.33294582e+00 8.09616387e-01 -8.22160244e-02 -6.63181096e-02 -1.38885462e+00 -2.18078613e-01 -4.75580990e-01 -3.94000739e-01 1.40378690e+00 5.07644355e-01 -5.71391046e-01 6.20461881e-01 5.44590235e-01 -1.37869403e-01 -1.16577530e+00 -8.36100757e-01 -3.36082608e-01 5.82652986e-01 4.08658683e-02 1.18464029e+00 8.56721401e-01 -1.48565650e-01 9.45980012e-01 -2.86517311e-02 -7.74358734e-02 3.10451746e-01 3.37769687e-01 9.00315940e-01 -8.20255935e-01 -3.40322882e-01 6.83829412e-02 1.21500149e-01 -1.73049963e+00 3.57914299e-01 -9.97345865e-01 1.88302770e-01 -1.80497134e+00 -3.53985012e-01 1.57856233e-02 -1.14816474e-02 1.86754763e-01 -9.46808219e-01 -4.71754938e-01 1.78217426e-01 -2.73736537e-01 -5.54175675e-01 1.32144439e+00 1.70636988e+00 -1.66546494e-01 -1.50316041e-02 -1.20247573e-01 -9.36670482e-01 1.69271722e-01 1.07348049e+00 -1.91137016e-01 -7.69467235e-01 -9.16753232e-01 3.78960371e-01 4.37141091e-01 4.37207937e-01 -6.32489264e-01 4.91303392e-02 -1.35724097e-01 2.22386733e-01 -6.61356747e-01 3.49555105e-01 -4.35097456e-01 -7.95952857e-01 1.44457355e-01 -1.08817387e+00 4.73077923e-01 1.91692084e-01 4.98364449e-01 -4.18417722e-01 -4.91562605e-01 4.81254667e-01 -5.29829320e-03 -3.32272470e-01 2.20798001e-01 -4.41205293e-01 7.66346097e-01 6.00349545e-01 1.63421288e-01 -4.43365395e-01 -7.28617311e-01 -6.81712449e-01 1.00311232e+00 2.32818164e-02 9.46514189e-01 9.79121506e-01 -1.52636075e+00 -1.01959014e+00 2.21660301e-01 3.31242889e-01 4.88911271e-01 3.19895476e-01 -2.97593772e-01 -3.38301182e-01 2.54340261e-01 -1.54369567e-02 -1.85317248e-01 -8.61745656e-01 5.38553715e-01 5.28896809e-01 -4.96279716e-01 -3.44726890e-01 8.50122094e-01 -2.12069049e-01 -1.33840942e+00 2.74506677e-02 -6.56381190e-01 -6.40316129e-01 -3.14685881e-01 6.18813992e-01 1.38695329e-01 -1.91328526e-01 -1.32225096e-01 1.41979009e-01 1.51464805e-01 -3.85736078e-01 -8.79845992e-02 1.04109919e+00 -1.39964789e-01 -1.60581663e-01 2.38566339e-01 1.24532712e+00 -4.19320047e-01 -8.73795092e-01 -5.99971175e-01 -4.36908565e-02 5.93109913e-02 -4.92187589e-01 -9.72887635e-01 -5.44692636e-01 8.12447309e-01 3.05093408e-01 3.62232596e-01 1.17779100e+00 2.18006760e-01 1.07923508e+00 9.97429192e-01 -3.40557337e-01 -6.75797880e-01 5.58959663e-01 9.32170570e-01 1.52588880e+00 -9.69450116e-01 -4.22754586e-01 -7.68835619e-02 -4.94800299e-01 8.75722826e-01 1.22481477e+00 -2.77735621e-01 1.98577479e-01 -4.13722366e-01 2.14094132e-01 -1.24673404e-01 -1.12622690e+00 -2.29845971e-01 1.71044022e-01 6.85524464e-01 5.98926134e-02 -2.41305336e-01 -2.97913160e-02 9.88505006e-01 -6.99037731e-01 1.66137457e-01 7.07329988e-01 9.64211881e-01 -3.84042501e-01 -1.17101574e+00 -1.59018606e-01 7.46194363e-01 -1.35923866e-02 -3.92744809e-01 -4.37282890e-01 4.76516902e-01 -2.60338545e-01 1.25830615e+00 1.00811899e-01 -4.47385103e-01 9.55530524e-01 3.50080729e-01 3.19122165e-01 -9.44506049e-01 -8.93799722e-01 -1.05170882e+00 1.98892683e-01 -2.25069538e-01 2.24481925e-01 -4.12650704e-01 -1.33971584e+00 1.94090143e-01 -3.23832512e-01 6.04820371e-01 3.01247805e-01 1.14666915e+00 6.01656795e-01 8.52021456e-01 7.20123231e-01 3.20034206e-01 -1.05535924e+00 -1.16077518e+00 2.65304238e-01 5.85686147e-01 6.65343523e-01 -6.58935010e-02 -1.87434316e-01 8.36349800e-02]
[11.441704750061035, 8.111491203308105]
0cd83a58-fd37-4ac2-898a-231f14e2cda7
shadow-background-noise-3d-spatial
2207.03064
null
https://arxiv.org/abs/2207.03064v2
https://arxiv.org/pdf/2207.03064v2.pdf
Shadow-Background-Noise 3D Spatial Decomposition Using Sparse Low-Rank Gaussian Properties for Video-SAR Moving Target Shadow Enhancement
Moving target shadows among video synthetic aperture radar (Video-SAR) images are always interfered by low scattering backgrounds and cluttered noises, causing poor detec-tion-tracking accuracy. Thus, a shadow-background-noise 3D spatial decomposition (SBN-3D-SD) model is proposed to enhance shadows for higher detection-tracking accuracy. It leverages the sparse property of shadows, the low-rank property of back-grounds, and the Gaussian property of noises to perform 3D spatial three-decomposition. It separates shadows from back-grounds and noises by the alternating direction method of multi-pliers (ADMM). Results on the Sandia National Laboratories (SNL) data verify its effectiveness. It boosts the shadow saliency from the qualitative and quantitative evaluation. It boosts the shadow detection accuracy of Faster R-CNN, RetinaNet and YOLOv3. It also boosts the shadow tracking accuracy of TransTrack, FairMOT and ByteTrack.
['Xu Zhan', 'Jun Shi', 'Zhenyu Yang', 'Tianwen Zhang', 'Xiaowo Xu', 'Xiaoling Zhang']
2022-07-07
null
null
null
null
['shadow-detection']
['computer-vision']
[ 9.27597508e-02 -6.12292945e-01 1.59163490e-01 -2.75942180e-02 -5.81831276e-01 -4.94297802e-01 6.15766525e-01 -7.13257730e-01 -2.26823360e-01 5.43292046e-01 4.29784954e-01 -4.82048005e-01 -2.30303891e-02 -3.81823421e-01 -3.90910029e-01 -1.10800743e+00 -1.98979452e-01 1.04433978e-02 6.57671690e-01 -1.68440565e-01 -1.51848987e-01 7.65380621e-01 -1.18275726e+00 4.85211104e-01 7.05441356e-01 1.10801339e+00 2.97978550e-01 7.42237270e-01 4.58370656e-01 9.04123068e-01 -4.66890693e-01 1.18950367e-01 6.90571904e-01 1.38996507e-03 4.35857981e-01 -8.58036652e-02 8.92588794e-01 -5.16488314e-01 -8.77170801e-01 9.57447231e-01 6.34424210e-01 -1.33825066e-02 4.99270260e-01 -1.19898117e+00 -5.41439652e-01 -1.60500079e-01 -9.40969706e-01 6.18027627e-01 -1.27178863e-01 6.08954132e-02 6.21621847e-01 -1.22504532e+00 4.92397398e-01 1.43694735e+00 1.15044546e+00 4.07475345e-02 -9.61486638e-01 -8.19488049e-01 2.25268409e-01 2.53406733e-01 -1.23356807e+00 -2.92475671e-01 4.30051744e-01 -3.13395441e-01 6.12018824e-01 4.51279998e-01 6.02834463e-01 1.35656345e+00 5.48406959e-01 9.82624173e-01 1.35147679e+00 3.81327011e-02 1.60062741e-02 -3.66615921e-01 3.11097443e-01 7.82532632e-01 7.93066740e-01 6.71079755e-01 -8.13949764e-01 -2.40650952e-01 6.41040921e-01 1.02306440e-01 -7.80752242e-01 -5.05116522e-01 -1.07361686e+00 6.50243282e-01 4.75500464e-01 6.87959343e-02 -2.77537614e-01 2.50316560e-01 -2.03166064e-02 -1.01625226e-01 4.12279069e-01 1.67746335e-01 -3.56482446e-01 4.78146404e-01 -1.13872492e+00 2.82334149e-01 5.75302064e-01 7.56207705e-01 2.94163972e-01 8.67149472e-01 -3.91214460e-01 4.26744133e-01 3.59642208e-01 1.64125216e+00 -9.61352363e-02 -6.83063149e-01 1.82375297e-01 1.59972146e-01 3.67052197e-01 -1.47103322e+00 -6.85304523e-01 -1.11130416e+00 -1.04687726e+00 5.91766953e-01 1.67482331e-01 -2.62988150e-01 -1.18685567e+00 1.35871673e+00 2.70803481e-01 5.81792831e-01 1.78611562e-01 1.27959692e+00 9.68021333e-01 4.42627639e-01 -3.28334630e-01 -1.43460348e-01 1.24703562e+00 -7.37735868e-01 -8.20106387e-01 -6.08723819e-01 2.92758137e-01 -7.71549106e-01 4.00702268e-01 3.92091572e-01 -3.71372461e-01 -5.09479523e-01 -1.28923512e+00 3.59246492e-01 -1.10314898e-01 3.86635214e-01 7.35587537e-01 8.46447945e-01 -6.47447109e-01 -8.90419260e-03 -8.28713000e-01 6.24159537e-02 7.95608580e-01 -1.86604425e-01 -1.75906289e-02 -5.87306678e-01 -9.90196943e-01 9.41261113e-01 -3.23089987e-01 4.61101621e-01 -1.32752109e+00 -1.19978774e+00 -5.38553357e-01 -3.61415088e-01 5.07827580e-01 -4.36302781e-01 5.92504084e-01 -7.64070094e-01 -8.00644457e-01 2.97025114e-01 -3.44830491e-02 -7.93593526e-01 4.63235289e-01 -5.83666861e-01 -8.27196717e-01 1.33190900e-01 1.16036326e-01 1.74666211e-01 1.11628520e+00 -1.56792760e+00 -6.11744940e-01 -6.72810972e-01 -3.26878965e-01 2.89801836e-01 2.61320025e-01 -2.72705615e-01 -3.36406887e-01 -9.98633027e-01 3.70937109e-01 -1.09472787e+00 -2.49030188e-01 1.37683734e-01 -3.13044518e-01 8.29439640e-01 1.46239769e+00 -9.15168583e-01 8.19192827e-01 -2.45561242e+00 -2.86747396e-01 1.95304051e-01 5.13668239e-01 4.01062310e-01 -2.96613693e-01 -1.43718600e-01 8.62358138e-02 -8.29619110e-01 -7.40809143e-02 5.09822220e-02 -1.31233722e-01 1.75723955e-01 -7.19141781e-01 1.05387259e+00 -1.69772625e-01 8.72800231e-01 -7.32653141e-01 -2.19172109e-02 -1.31313633e-02 7.35143483e-01 -2.31985509e-01 -1.84811890e-01 2.54513808e-02 2.48527870e-01 -6.23972118e-01 1.13618827e+00 1.18463516e+00 -3.60970572e-02 -1.47460714e-01 -7.11400628e-01 -3.16934019e-01 -3.38658065e-01 -1.26627386e+00 9.11419928e-01 -1.67835101e-01 1.07017422e+00 6.01785481e-01 -4.05130595e-01 7.59565294e-01 -2.18073919e-01 3.93878609e-01 -1.09703672e+00 7.25893974e-02 1.28375337e-01 8.41715485e-02 -2.64485866e-01 5.16640186e-01 -1.18780218e-01 1.55691326e-01 -6.73376098e-02 -4.77086246e-01 1.53059259e-01 -4.66363162e-01 3.78799200e-01 1.45889854e+00 5.22341989e-02 -1.28852293e-01 -4.97983068e-01 -1.29805990e-02 2.66305923e-01 7.76885927e-01 9.42772985e-01 -2.15371683e-01 5.52019775e-01 -1.10212751e-01 -4.17642295e-01 -4.60784227e-01 -1.36589992e+00 -1.41995922e-01 9.76689219e-01 3.36714536e-01 2.74843238e-02 -9.87745821e-02 -5.17438114e-01 3.41068000e-01 6.55358672e-01 -5.65806448e-01 -6.60489500e-02 -4.69389379e-01 -1.06162918e+00 7.76213348e-01 3.88128430e-01 7.37420022e-01 -3.47012162e-01 -1.04093993e+00 -1.92961725e-03 -3.45054179e-01 -1.51991546e+00 -2.95424879e-01 3.50414306e-01 -5.60469925e-01 -1.07372296e+00 -7.28017271e-01 -1.63278997e-01 4.28990364e-01 1.25964177e+00 7.99254298e-01 -2.00862899e-01 -7.28616774e-01 4.61722314e-01 -3.23233575e-01 -6.89342797e-01 1.95808992e-01 -1.01906300e+00 4.73257124e-01 3.04064870e-01 1.34590089e-01 -3.50545108e-01 -5.94852626e-01 5.45336962e-01 -5.18244565e-01 -8.15095846e-03 9.18965280e-01 8.50406528e-01 3.44791144e-01 2.64616013e-01 -1.39230534e-01 -4.94591743e-01 -4.93121101e-03 -8.25719684e-02 -1.01205790e+00 1.28965499e-02 -3.42795163e-01 -4.80473757e-01 2.14035332e-01 -4.68400329e-01 -1.23645747e+00 5.80413416e-02 6.12322927e-01 -7.81029820e-01 2.46243671e-01 2.60346401e-02 -9.96634588e-02 -7.23651230e-01 7.09183693e-01 2.95271218e-01 -2.70574868e-01 -2.00673729e-01 2.46500283e-01 2.55043626e-01 6.38050377e-01 2.23446995e-01 1.51309013e+00 1.28915155e+00 3.06222588e-01 -1.44224751e+00 -1.24316549e+00 -4.04146463e-01 -2.47468889e-01 -5.81317067e-01 7.72799194e-01 -1.34465325e+00 -5.06409526e-01 6.49016201e-01 -8.76760244e-01 -4.38835442e-01 1.65902719e-01 6.94230855e-01 1.05838962e-01 5.27342200e-01 -1.50017977e-01 -1.18908286e+00 -2.14867309e-01 -6.94962800e-01 1.04806566e+00 1.19466677e-01 2.53518879e-01 -5.78649879e-01 -3.28335315e-01 4.18049365e-01 5.87547243e-01 3.53559285e-01 3.90107542e-01 -1.95687801e-01 -1.19160211e+00 -1.48713246e-01 -6.64427221e-01 3.92184615e-01 -2.49395058e-01 -5.63109577e-01 -9.82153654e-01 -4.69861895e-01 1.03713185e-01 2.26350382e-01 1.29234910e+00 1.00049043e+00 3.52971613e-01 -1.68222845e-01 -6.98026538e-01 1.00376761e+00 1.36514628e+00 6.39341250e-02 6.51375294e-01 3.18208814e-01 9.27005112e-01 2.27894396e-01 1.02663064e+00 4.84071672e-01 -4.16134979e-04 6.94649696e-01 4.72381443e-01 -5.48511505e-01 -6.45894527e-01 2.98735827e-01 6.42301977e-01 2.43614569e-01 -8.57776031e-03 -3.77804902e-03 -9.75939512e-01 3.51808667e-01 -1.67323196e+00 -1.00720429e+00 -8.08709264e-01 1.76936162e+00 1.22003660e-01 1.77803040e-01 -4.83219177e-01 -3.14893782e-01 3.55617583e-01 5.55926681e-01 -6.02531552e-01 6.71740890e-01 -9.05969203e-01 -4.98002321e-02 1.30809724e+00 7.18611896e-01 -1.07197416e+00 1.01524019e+00 5.80817652e+00 1.00774515e+00 -9.07526255e-01 1.33820459e-01 2.55542994e-01 -7.60666654e-02 -5.43239191e-02 -1.30712464e-01 -1.21448386e+00 1.26639500e-01 4.02110249e-01 3.31811845e-01 1.87333405e-01 6.31177902e-01 1.94746614e-01 -5.40918291e-01 -3.40964407e-01 9.51690078e-01 4.42653716e-01 -1.64656484e+00 -2.08420902e-01 5.57080247e-02 6.51340663e-01 5.04854381e-01 4.24378693e-01 2.99919903e-01 5.12329996e-01 -6.91187978e-01 6.40707672e-01 5.64144790e-01 5.16782165e-01 -4.68797773e-01 6.79261923e-01 1.91970468e-01 -1.15262651e+00 -3.93982798e-01 -3.92497987e-01 1.81335866e-01 1.62535205e-01 1.23664832e+00 -8.22974265e-01 6.24312222e-01 9.73703146e-01 4.90655303e-01 -4.13177639e-01 1.00272512e+00 -1.38751179e-01 5.96908391e-01 -3.89579952e-01 2.41623700e-01 4.73457336e-01 -2.40671650e-01 1.14912975e+00 1.53622878e+00 1.03194617e-01 4.98113543e-01 5.16885161e-01 5.34926414e-01 3.67823869e-01 -6.69492424e-01 -6.44867837e-01 2.55634695e-01 1.01960734e-01 1.44289637e+00 -7.19220042e-01 -1.07730180e-01 -3.01499814e-01 6.67863190e-01 -7.32657015e-01 7.29205072e-01 -1.19320059e+00 -1.51490942e-01 8.52411270e-01 2.04268783e-01 7.57576168e-01 -7.48821855e-01 -3.73616874e-01 -8.37910175e-01 -2.46774703e-01 -1.13682997e+00 5.36513841e-03 -1.03731048e+00 -1.16113532e+00 6.73998773e-01 -2.25122467e-01 -1.47299194e+00 8.29359233e-01 -7.59132445e-01 -3.32189113e-01 6.65903091e-01 -1.84419835e+00 -1.39616430e+00 -7.52496600e-01 5.83605587e-01 2.49200076e-01 -4.74148601e-01 4.47240591e-01 4.67862874e-01 -3.39283764e-01 1.05418861e-01 1.98257893e-01 2.27015495e-01 5.92249990e-01 -6.43673778e-01 3.60672772e-01 1.35250282e+00 8.53060856e-02 1.13290943e-01 9.24749792e-01 -1.17510307e+00 -1.90727150e+00 -1.39068210e+00 2.69916862e-01 -4.63249117e-01 9.82719898e-01 -4.65634793e-01 -6.53268158e-01 3.33402961e-01 -1.33449480e-01 1.65692076e-01 4.98768359e-01 -1.80745199e-01 -7.07655787e-01 -2.70101756e-01 -6.43710256e-01 6.87229156e-01 1.03258991e+00 -2.40660980e-01 -4.42016900e-01 6.57000065e-01 6.17713749e-01 -6.06803358e-01 -4.22705039e-02 6.53283417e-01 4.90195602e-01 -9.67622876e-01 1.44596136e+00 -2.91712910e-01 -1.56606779e-01 -7.02803195e-01 -8.59983146e-01 -1.06953573e+00 -6.79725409e-01 -5.02714396e-01 -3.16871375e-01 5.37634194e-01 1.93281829e-01 -5.79790473e-01 8.65705490e-01 -2.86494553e-01 -3.95901710e-01 -4.03698862e-01 -9.45709527e-01 -1.21070671e+00 -9.12349880e-01 -5.18647432e-01 -1.10093147e-01 8.68183553e-01 -1.04595780e+00 3.58968288e-01 -5.73450804e-01 1.18237996e+00 1.45280027e+00 5.35672188e-01 7.47077286e-01 -1.26420093e+00 -2.98102736e-01 1.29026294e-01 -1.14119284e-01 -1.23664939e+00 -1.12426080e-01 -6.17737830e-01 2.48343185e-01 -1.38381672e+00 -1.95034314e-02 -4.30625021e-01 -1.61957234e-01 2.64159709e-01 8.22719708e-02 5.89856386e-01 2.45716453e-01 1.18974350e-01 -7.66691089e-01 8.35456610e-01 1.19604039e+00 -1.13582984e-01 1.36686996e-01 2.53695965e-01 -4.66780335e-01 9.68464971e-01 4.21774268e-01 -5.57091832e-01 -4.07819450e-02 -5.23411393e-01 -8.35298076e-02 -2.00130716e-02 9.18652177e-01 -1.40378118e+00 3.92819554e-01 -1.23853371e-01 9.60874557e-01 -1.26922309e+00 7.21651137e-01 -8.65412116e-01 4.46112175e-03 7.06532300e-01 4.59603369e-01 -2.74062365e-01 4.80767250e-01 1.14435434e+00 6.89782798e-02 6.04315221e-01 1.19388545e+00 1.24060683e-01 -8.86693895e-01 3.31946462e-01 -5.75814247e-01 -1.84592232e-03 8.34705055e-01 -3.62477571e-01 -4.87519741e-01 -3.70923221e-01 -2.55587429e-01 1.41402975e-01 -1.29601613e-01 3.31590325e-01 8.00126672e-01 -1.10837102e+00 -9.00199175e-01 3.15678179e-01 -1.57927454e-01 -6.46197975e-01 5.57210326e-01 1.41890323e+00 -3.05347264e-01 4.58833426e-01 -2.32812628e-01 -6.87362313e-01 -1.61847341e+00 1.58550128e-01 3.72358799e-01 5.55400252e-02 -1.04250443e+00 1.16652489e+00 6.71493471e-01 -3.07442825e-02 4.35038447e-01 -1.94292381e-01 9.37943384e-02 -8.16349126e-03 7.21083045e-01 5.76333702e-01 -9.90392175e-03 -6.58400655e-01 -6.45688832e-01 5.08804798e-01 9.14185271e-02 -6.21349998e-02 1.41790199e+00 5.75929806e-02 1.51110977e-01 1.77452806e-02 7.91563749e-01 4.90408748e-01 -1.50662339e+00 -3.36881936e-01 -1.15750253e-01 -7.85810947e-01 8.05382013e-01 -9.87665355e-01 -1.17790246e+00 6.05169356e-01 8.76397967e-01 -1.11574687e-01 1.11197007e+00 -3.60983044e-01 7.85511315e-01 5.80645084e-01 3.71949017e-01 -9.09577310e-01 3.98853332e-01 9.34129596e-01 8.83490145e-01 -8.31676006e-01 5.20072818e-01 -4.50018287e-01 -8.85636926e-01 8.24949145e-01 3.65706414e-01 -2.66731173e-01 7.50163734e-01 8.93021226e-01 2.42339775e-01 -4.46309179e-01 -3.73178273e-01 -5.08493304e-01 3.19486529e-01 1.05714333e+00 -4.03876483e-01 1.09109811e-01 4.90727544e-01 7.07623124e-01 2.57705450e-01 -4.94286269e-01 4.09942597e-01 6.76599681e-01 -7.10305572e-01 -3.33413869e-01 -8.35094154e-01 3.51052940e-01 -5.59638180e-02 -4.62758094e-01 -4.73211944e-01 9.08765912e-01 1.61460578e-01 9.56780136e-01 -2.86761314e-01 -4.71262753e-01 3.93759251e-01 -5.39889753e-01 3.64672512e-01 -1.71834961e-01 -2.52029419e-01 4.59604710e-01 3.05559635e-01 -9.39691246e-01 -1.06516175e-01 -7.18873620e-01 -1.05824924e+00 -1.71421111e-01 -5.16054749e-01 -1.12598777e-01 6.50347769e-01 6.99702680e-01 4.69098926e-01 8.79328668e-01 3.82360965e-01 -8.41320932e-01 -4.73641157e-01 -4.92772907e-01 -9.40249860e-01 -3.21891546e-01 6.81219101e-01 -1.05379200e+00 -5.86714029e-01 -1.86559841e-01]
[8.21466064453125, -1.095902919769287]
3421c301-759d-4cfe-85d8-c673381a2869
extractive-summarization-of-legal-decisions
2210.12437
null
https://arxiv.org/abs/2210.12437v1
https://arxiv.org/pdf/2210.12437v1.pdf
Extractive Summarization of Legal Decisions using Multi-task Learning and Maximal Marginal Relevance
Summarizing legal decisions requires the expertise of law practitioners, which is both time- and cost-intensive. This paper presents techniques for extractive summarization of legal decisions in a low-resource setting using limited expert annotated data. We test a set of models that locate relevant content using a sequential model and tackle redundancy by leveraging maximal marginal relevance to compose summaries. We also demonstrate an implicit approach to help train our proposed models generate more informative summaries. Our multi-task learning model variant leverages rhetorical role identification as an auxiliary task to further improve the summarizer. We perform extensive experiments on datasets containing legal decisions from the US Board of Veterans' Appeals and conduct quantitative and expert-ranked evaluations of our models. Our results show that the proposed approaches can achieve ROUGE scores vis-\`a-vis expert extracted summaries that match those achieved by inter-annotator comparison.
['Matthias Grabmair', 'Shanshan Xu', 'Abhishek Agarwal']
2022-10-22
null
null
null
null
['extractive-summarization']
['natural-language-processing']
[ 4.40813303e-01 5.29216945e-01 -7.69320130e-01 -4.17784870e-01 -1.97830844e+00 -1.00136435e+00 7.25700438e-01 7.63355851e-01 -5.91093481e-01 1.25790823e+00 1.16934919e+00 -5.59913099e-01 -2.79824376e-01 -2.47570232e-01 -2.73068875e-01 -1.99104443e-01 3.84301007e-01 6.61617756e-01 9.84243602e-02 -2.13954359e-01 9.60326910e-01 2.60740548e-01 -8.89631569e-01 8.83030176e-01 1.62334895e+00 3.34816068e-01 -2.59883910e-01 8.88752460e-01 -1.70194462e-01 1.62113786e+00 -1.10181546e+00 -9.53194201e-01 3.18102896e-01 -4.72248614e-01 -8.85686696e-01 -2.70322472e-01 8.41803014e-01 -5.00556648e-01 -4.11612988e-01 8.08982015e-01 8.42807293e-01 8.55658352e-02 1.14609194e+00 -6.94324315e-01 -5.91296673e-01 1.06397593e+00 -6.02254570e-01 7.58066356e-01 6.38968885e-01 -1.51411528e-02 1.45795405e+00 -4.01220590e-01 8.74770761e-01 1.22759366e+00 4.88727421e-01 4.01731074e-01 -1.04906988e+00 -3.76224250e-01 3.41006905e-01 3.52796704e-01 -7.18082130e-01 -8.78239334e-01 8.42604876e-01 -5.73164582e-01 1.05634665e+00 2.83263147e-01 1.66646149e-02 1.12586117e+00 3.18201214e-01 1.01574194e+00 9.55683291e-01 -4.30180609e-01 3.25175822e-01 4.85967919e-02 6.11906171e-01 5.49213231e-01 8.11752677e-01 -6.02327228e-01 -4.60529059e-01 -9.03440654e-01 1.48381351e-03 -3.25049281e-01 -1.09282240e-01 3.84156466e-01 -7.16279387e-01 9.84339416e-01 -2.31837556e-01 2.74026513e-01 -6.66629851e-01 1.00895196e-01 7.71422982e-01 1.97973758e-01 5.53400517e-01 9.65840995e-01 -2.38226265e-01 -3.27304423e-01 -1.25452304e+00 5.73364794e-01 1.01560092e+00 8.87566507e-01 2.24447176e-01 -2.50436932e-01 -1.19757044e+00 8.30028653e-01 -1.56961739e-01 4.49302375e-01 2.62070864e-01 -1.27875888e+00 1.04384232e+00 5.53494573e-01 3.80668551e-01 -1.04191720e+00 -3.28557417e-02 -4.30211514e-01 -4.45061386e-01 -1.53883383e-01 7.01197758e-02 -2.94251740e-01 -7.19948590e-01 1.21718287e+00 2.32743137e-02 -2.99340189e-01 1.53275087e-01 3.50315720e-01 1.05110025e+00 3.57408345e-01 3.91687155e-01 -5.26614606e-01 1.34156752e+00 -9.76819396e-01 -1.00382245e+00 -2.35389963e-01 7.58980632e-01 -7.66487598e-01 8.59319627e-01 2.44046122e-01 -1.27660680e+00 -2.13584796e-01 -9.03908253e-01 -4.63458419e-01 2.46646792e-01 7.49175966e-01 4.43298250e-01 4.82956022e-01 -5.85092723e-01 5.48046291e-01 -4.35840815e-01 -8.77358019e-02 8.61138701e-01 -2.67326653e-01 -3.24320234e-02 1.65859371e-01 -1.00764573e+00 1.13265324e+00 3.25173467e-01 -2.63045341e-01 -7.06121624e-01 -7.35872746e-01 -8.65421355e-01 2.57798702e-01 8.06387186e-01 -7.29801655e-01 1.59095287e+00 -3.96465734e-02 -9.61498201e-01 7.81676710e-01 -4.94712234e-01 -6.92092836e-01 7.35154033e-01 -5.20658791e-01 -2.72318006e-01 5.90075076e-01 9.60208952e-01 1.19722396e-01 4.38497066e-01 -1.12266779e+00 -9.68224823e-01 -8.14014897e-02 3.17905039e-01 1.58030599e-01 -2.18514740e-01 3.56553167e-01 -5.82802817e-02 -6.99885249e-01 -4.62027520e-01 -6.03065193e-01 -4.31393594e-01 -6.60143316e-01 -8.29870880e-01 -4.98155326e-01 2.91733384e-01 -1.06071699e+00 1.85305583e+00 -1.35744417e+00 -1.51484624e-01 2.51281336e-02 4.14845884e-01 3.31127673e-01 -4.67967466e-02 7.95012057e-01 1.94453508e-01 5.01538098e-01 -3.51730078e-01 -3.45448136e-01 -5.95073178e-02 3.96369258e-03 -8.41158688e-01 2.60184050e-01 5.00913933e-02 8.04697216e-01 -1.09062517e+00 -1.07208967e+00 -1.84985831e-01 -3.60404819e-01 -5.10072112e-01 -1.55283526e-01 -2.77588189e-01 2.92521745e-01 -8.94691944e-01 6.65209353e-01 1.53258592e-01 -1.91804379e-01 1.18465237e-01 -1.60082549e-01 -6.44096658e-02 8.34246457e-01 -4.88974094e-01 1.74955463e+00 -2.51772851e-01 6.48367047e-01 -2.01591477e-01 -1.00906849e+00 4.80728567e-01 3.24017018e-01 3.04892033e-01 -6.30118012e-01 1.80103287e-01 2.54015595e-01 -1.97916441e-02 -8.12277138e-01 7.75806844e-01 1.03695415e-01 -5.10118902e-01 7.14329243e-01 -1.13757476e-01 9.80042294e-02 8.50259185e-01 9.32834923e-01 1.55669308e+00 -1.38640985e-01 8.61820042e-01 -1.67268664e-01 4.90915775e-01 5.85357070e-01 7.28230298e-01 1.38659251e+00 -3.15904133e-02 1.71205834e-01 9.56905365e-01 -1.66535899e-01 -9.22796905e-01 -3.97487700e-01 -1.52044240e-02 9.35509145e-01 -3.52107376e-01 -7.21668482e-01 -4.65902627e-01 -1.14012873e+00 4.65034135e-02 1.29607427e+00 -6.83751404e-01 1.82699576e-01 -8.25559556e-01 -4.18092012e-01 7.51406908e-01 4.30639505e-01 2.80671328e-01 -8.22311342e-01 -9.27501976e-01 3.02361190e-01 -4.43018764e-01 -1.29869294e+00 -5.24177074e-01 -2.41164058e-01 -6.03179455e-01 -1.29086411e+00 -5.49221516e-01 -2.51595169e-01 4.38817054e-01 -2.95833535e-02 8.72018516e-01 -9.67679247e-02 -1.17173254e-01 3.53681535e-01 -4.14392412e-01 -6.31952345e-01 -7.18805969e-01 4.47719246e-01 -2.83703655e-01 -4.74889547e-01 1.53058976e-01 -5.18977165e-01 -4.23415780e-01 -3.92747015e-01 -7.55471766e-01 -4.49555703e-02 8.48098695e-01 6.20318413e-01 1.74140915e-01 -4.12406534e-01 1.10320616e+00 -1.52279305e+00 1.65568602e+00 -3.68696988e-01 -3.18694383e-01 7.20907807e-01 -6.86450958e-01 2.96398968e-01 6.39468491e-01 2.82680918e-03 -1.29843116e+00 -4.73184109e-01 1.53666645e-01 1.00959092e-03 9.67852175e-02 7.75385737e-01 2.01864079e-01 6.30707979e-01 1.20267189e+00 -1.09115452e-01 -4.17199433e-01 -2.74704009e-01 5.63695848e-01 9.56748128e-01 7.19248235e-01 -8.17874193e-01 7.04479814e-01 2.66496867e-01 -2.32034460e-01 -3.44247013e-01 -1.59568620e+00 -6.63227499e-01 -4.14890796e-01 -1.55135825e-01 6.18872404e-01 -7.86294699e-01 -5.15645444e-01 -4.60530818e-01 -1.56517386e+00 8.62940103e-02 -2.45006040e-01 2.50046879e-01 -2.47989371e-01 8.16931844e-01 -6.01371229e-01 -9.42817390e-01 -9.42646980e-01 -7.28984654e-01 1.15506935e+00 1.42197087e-01 -6.81124091e-01 -8.58985841e-01 4.26417977e-01 9.09821868e-01 2.56861914e-02 4.41004336e-01 1.30308962e+00 -1.42289364e+00 -1.75095201e-01 -5.13143897e-01 -1.92018047e-01 1.39029458e-01 2.62540221e-01 -1.18530810e-01 -8.03824544e-01 5.21086082e-02 -2.40859464e-01 -3.27661008e-01 1.14754009e+00 5.04489541e-01 8.88535261e-01 -1.04157341e+00 -4.98104990e-01 -3.23488832e-01 1.01510346e+00 3.15784723e-01 3.57219726e-01 2.41019294e-01 5.07822514e-01 6.19045973e-01 9.52778101e-01 5.80974102e-01 4.07999367e-01 3.13357413e-01 -3.85646492e-01 3.37573856e-01 -1.34122521e-01 -4.35188085e-01 -2.19038520e-02 4.57739919e-01 -2.72533000e-01 -4.35828418e-01 -9.20810759e-01 8.29882801e-01 -2.33266377e+00 -1.44189811e+00 9.99304205e-02 1.74636960e+00 1.04867470e+00 3.68710935e-01 1.68369144e-01 -3.06178387e-02 6.60001338e-01 2.87574649e-01 -4.86886352e-01 -5.97767115e-01 5.57615655e-03 2.71041453e-01 4.73933965e-01 6.36600971e-01 -1.10117590e+00 1.04195845e+00 6.56913185e+00 1.02529585e+00 -4.67812359e-01 1.48680940e-01 6.67515039e-01 -3.54151756e-01 -6.42519355e-01 2.05414489e-01 -8.37699592e-01 5.19803166e-01 9.37710762e-01 -7.57405400e-01 -3.61638337e-01 1.08023894e+00 4.62573975e-01 -2.22178966e-01 -1.10095978e+00 6.98632479e-01 3.97903442e-01 -1.91505682e+00 2.69482523e-01 2.17097715e-01 9.33349788e-01 -3.61919612e-01 -3.71083319e-01 5.54544747e-01 7.82387972e-01 -9.05207515e-01 6.94976568e-01 4.77565885e-01 6.36321485e-01 -5.86839318e-01 7.89216399e-01 5.16961932e-01 -6.95039928e-01 -3.96227390e-01 -3.46398443e-01 1.19991004e-01 6.45767450e-01 5.70483744e-01 -1.21076167e+00 7.31923938e-01 4.19139266e-02 7.14692712e-01 -6.16187990e-01 1.00739574e+00 -6.15398705e-01 6.19851649e-01 2.92139858e-01 -1.80546716e-01 3.46365839e-01 4.16180193e-02 7.70852804e-01 1.49567580e+00 4.46931422e-02 5.09979308e-01 3.92073542e-01 5.09697974e-01 -4.26327169e-01 2.93590516e-01 -6.78046942e-01 -4.58978675e-02 7.33714104e-01 1.17291796e+00 -4.38099831e-01 -9.84399855e-01 3.00697815e-02 5.83011150e-01 3.43618870e-01 2.13713288e-01 -7.52616465e-01 -4.57488239e-01 -6.64382577e-02 -6.03441261e-02 8.36378932e-02 1.65779814e-01 -5.96718729e-01 -1.16299117e+00 1.81869462e-01 -1.04281354e+00 7.44389534e-01 -5.39243877e-01 -1.16362727e+00 6.06357992e-01 5.04869878e-01 -1.14098668e+00 -6.30089998e-01 -4.02191132e-02 -8.37533474e-01 5.21432161e-01 -1.56278849e+00 -9.12859738e-01 1.22980863e-01 -5.51804788e-02 9.20575082e-01 -3.71943504e-01 4.63522732e-01 2.71293074e-01 -5.95530212e-01 4.70858008e-01 -3.20265740e-01 1.71633333e-01 9.38837886e-01 -1.28790760e+00 5.99256717e-02 1.09984648e+00 2.10431237e-02 8.72462153e-01 1.02056229e+00 -1.13977206e+00 -6.38707042e-01 -9.44148839e-01 1.29469740e+00 -6.66148305e-01 6.63262486e-01 1.77442998e-01 -6.10660553e-01 6.61498845e-01 6.67598724e-01 -8.70744586e-01 1.17019737e+00 1.66408941e-01 -3.80474716e-01 1.13648046e-02 -1.07039118e+00 6.98336482e-01 1.13398969e+00 -5.22640049e-01 -1.40702868e+00 7.75640786e-01 4.97846961e-01 -3.22780162e-01 -5.27617991e-01 2.33537421e-01 4.81158227e-01 -5.67810535e-01 6.42566323e-01 -1.12131357e+00 8.80581439e-01 -1.56973109e-01 1.61337450e-01 -1.09715986e+00 -2.56229281e-01 -7.83417583e-01 -8.70889723e-02 1.16731000e+00 9.04772162e-01 -3.40095252e-01 3.06184411e-01 9.21858966e-01 -4.86559689e-01 -6.35341644e-01 -8.60248566e-01 -5.09301066e-01 -1.20394476e-01 -1.45808309e-01 4.72453050e-02 8.22567344e-01 3.61356169e-01 9.14246380e-01 -4.62054670e-01 -1.89272568e-01 5.35939693e-01 3.78037095e-01 7.46972919e-01 -1.26040506e+00 -2.40116224e-01 -4.94964421e-01 5.63961454e-02 -8.38264585e-01 4.92574155e-01 -8.21189463e-01 -1.19540997e-01 -2.27041841e+00 6.46195769e-01 -7.45560825e-02 8.86480696e-03 7.01876223e-01 -4.80749696e-01 -4.60685074e-01 2.57544983e-02 4.84488964e-01 -1.04551983e+00 3.52848560e-01 9.94414508e-01 -3.33907306e-01 -2.50200748e-01 -9.19377580e-02 -1.59665453e+00 7.19987452e-01 6.23401463e-01 -7.60585368e-01 -5.96399963e-01 -3.85728300e-01 2.05717936e-01 2.96651423e-01 -1.51413143e-01 -7.34837592e-01 6.76596403e-01 -2.79073328e-01 6.16104305e-02 -8.12368810e-01 1.00557923e-01 -1.23275585e-01 -4.17548537e-01 2.95151174e-01 -1.11734509e+00 4.88078222e-02 3.23885605e-02 8.49930584e-01 -1.71206757e-01 -6.46160841e-01 1.81656897e-01 -3.37393731e-01 -3.71162929e-02 -9.89226103e-02 -5.83772063e-01 2.84072250e-01 7.85374701e-01 -2.62133796e-02 -9.93440628e-01 -6.08910441e-01 -3.96426320e-01 4.25056577e-01 1.37121445e-02 -3.01070381e-02 4.03597116e-01 -8.73846173e-01 -1.26493335e+00 -9.21968281e-01 1.80156216e-01 -3.23584765e-01 2.11288065e-01 8.23798478e-01 -3.29535931e-01 9.12699878e-01 2.19131261e-01 -1.13132544e-01 -1.41739714e+00 2.42227256e-01 -3.03331792e-01 -9.58757401e-01 -5.75384498e-01 3.93683225e-01 -3.64198834e-01 1.09362587e-01 5.95723614e-02 -3.51840556e-01 -7.56448090e-01 5.21549523e-01 7.01152027e-01 5.14404714e-01 3.73139307e-02 -3.10435742e-01 -2.00511098e-01 2.05464557e-01 -6.81682408e-01 -4.57538188e-01 1.42259669e+00 2.29301497e-01 -2.65064836e-02 5.62678576e-02 6.70054913e-01 6.91887379e-01 -4.73034054e-01 -2.53479809e-01 4.35897410e-01 -4.27123606e-01 -8.60517025e-02 -1.03288662e+00 -2.56070942e-01 3.49161148e-01 -3.47048193e-01 1.07288159e-01 8.61026585e-01 1.34813905e-01 6.05964482e-01 9.94435549e-01 -6.74030483e-02 -1.38458323e+00 1.70450583e-01 1.90762594e-01 1.17807400e+00 -1.12226319e+00 4.86250639e-01 -4.02671337e-01 -1.08843291e+00 9.03247714e-01 2.45767623e-01 -1.26449749e-01 7.23422021e-02 -1.05995111e-01 3.00860647e-02 -5.33493042e-01 -1.11540103e+00 -7.64338970e-02 6.03120208e-01 7.95729905e-02 3.81118894e-01 -2.31239706e-01 -1.10144281e+00 6.66999578e-01 -1.55989349e-01 -1.14908442e-01 9.12616909e-01 1.11946130e+00 -7.33173251e-01 -1.00463223e+00 -7.69043863e-02 9.89580214e-01 -8.06793272e-01 -3.29081118e-01 -7.36671209e-01 4.81597632e-01 -3.68546307e-01 1.32116377e+00 -4.90846485e-01 1.26763448e-01 4.45021540e-01 1.51392698e-01 2.43576869e-01 -1.04831302e+00 -7.38259435e-01 1.60026610e-01 1.13184154e+00 -2.52883703e-01 -6.31716251e-01 -8.23139369e-01 -9.08058643e-01 7.08291354e-03 -5.18611260e-02 5.94076574e-01 3.10099632e-01 1.19121027e+00 6.48683727e-01 6.17217362e-01 4.42643374e-01 -6.26984388e-02 -9.51942742e-01 -1.42340195e+00 7.76765496e-02 2.68549502e-01 3.35475385e-01 -3.67604136e-01 -1.58330247e-01 1.53363468e-02]
[12.09402847290039, 9.55251407623291]
464159bb-cbc0-463f-b610-2724b2f4434d
the-franz-parisi-criterion-and-computational
2205.09727
null
https://arxiv.org/abs/2205.09727v2
https://arxiv.org/pdf/2205.09727v2.pdf
The Franz-Parisi Criterion and Computational Trade-offs in High Dimensional Statistics
Many high-dimensional statistical inference problems are believed to possess inherent computational hardness. Various frameworks have been proposed to give rigorous evidence for such hardness, including lower bounds against restricted models of computation (such as low-degree functions), as well as methods rooted in statistical physics that are based on free energy landscapes. This paper aims to make a rigorous connection between the seemingly different low-degree and free-energy based approaches. We define a free-energy based criterion for hardness and formally connect it to the well-established notion of low-degree hardness for a broad class of statistical problems, namely all Gaussian additive models and certain models with a sparse planted signal. By leveraging these rigorous connections we are able to: establish that for Gaussian additive models the "algebraic" notion of low-degree hardness implies failure of "geometric" local MCMC algorithms, and provide new low-degree lower bounds for sparse linear regression which seem difficult to prove directly. These results provide both conceptual insights into the connections between different notions of hardness, as well as concrete technical tools such as new methods for proving low-degree lower bounds.
['Ilias Zadik', 'Alexander S. Wein', 'Tselil Schramm', 'Samuel B. Hopkins', 'Ahmed El Alaoui', 'Afonso S. Bandeira']
2022-05-19
null
null
null
null
['additive-models']
['methodology']
[ 5.61883271e-01 5.58884621e-01 1.76552787e-01 -3.25175852e-01 -9.91995752e-01 -5.70809841e-01 5.40741682e-01 2.95738757e-01 2.77208388e-02 5.72069228e-01 3.51969153e-02 -3.92083853e-01 -7.87295759e-01 -1.01446819e+00 -8.44800949e-01 -1.16422129e+00 -5.72631657e-01 4.76727903e-01 1.62625045e-01 -2.17294469e-01 3.66398573e-01 5.04030168e-01 -1.34934795e+00 -1.15751624e-01 7.38267243e-01 5.53088486e-01 -1.53656021e-01 9.26356673e-01 3.20110798e-01 1.18968531e-01 -3.22965570e-02 -4.91392165e-01 1.04203910e-01 -6.24603987e-01 -7.07765579e-01 -2.00440109e-01 5.27394116e-01 2.16960832e-01 -5.15496552e-01 1.19702005e+00 2.64772296e-01 -4.29527052e-02 8.24718714e-01 -1.28563941e+00 -7.30050504e-01 5.27333736e-01 -6.72960579e-01 1.25218660e-01 3.61005306e-01 -2.69328147e-01 1.27445602e+00 -5.09548783e-01 4.26729560e-01 1.24531519e+00 9.64171290e-01 4.60555196e-01 -1.73022485e+00 -1.40301555e-01 -2.45218679e-01 1.44813359e-01 -1.40158224e+00 -3.86261046e-01 7.33090937e-01 -4.14567977e-01 6.12627327e-01 7.18278885e-01 4.55028832e-01 8.48899722e-01 3.22654963e-01 5.70845246e-01 1.47875786e+00 -8.39526594e-01 5.17551124e-01 -1.80108264e-01 4.71757293e-01 9.35365736e-01 8.01668525e-01 1.48572385e-01 -5.82336009e-01 -6.74613357e-01 7.81998992e-01 -2.31956631e-01 -2.85739481e-01 -5.20441353e-01 -9.38251793e-01 1.05283523e+00 1.71834305e-01 4.37906832e-01 2.22981215e-01 5.04434466e-01 1.47439152e-01 1.02412775e-01 4.35712427e-01 1.86841816e-01 -4.05741751e-01 1.83725998e-01 -7.33941019e-01 3.87037635e-01 1.10158992e+00 7.85639882e-01 8.09834957e-01 -2.98257023e-01 2.27803871e-01 3.13950837e-01 5.14275610e-01 5.69328010e-01 -3.33426684e-01 -8.83144617e-01 2.35821426e-01 -1.82132930e-01 -3.20959724e-02 -1.47586048e+00 -4.32033569e-01 -3.97490203e-01 -1.11966419e+00 5.36224172e-02 5.82614362e-01 3.57722342e-01 -5.15136838e-01 2.07444882e+00 2.60973036e-01 3.58377934e-01 -8.71434435e-02 6.17877662e-01 2.20579714e-01 4.65661347e-01 -2.60277987e-01 -4.47153270e-01 9.57216024e-01 -2.80424297e-01 -3.62548113e-01 1.05220610e-02 7.19910443e-01 -3.68346214e-01 7.56924629e-01 3.79486710e-01 -1.12098539e+00 -8.83569382e-03 -1.05666459e+00 -1.64496273e-01 -1.35644317e-01 -4.98695225e-01 1.09055746e+00 9.71368134e-01 -1.15719211e+00 8.23558271e-01 -1.12183344e+00 -4.27036285e-01 1.15681179e-01 2.28162572e-01 -2.05359489e-01 -2.71079123e-01 -7.43058324e-01 6.01875365e-01 -1.00883126e-01 3.46785188e-01 -6.78239584e-01 -4.93874907e-01 -6.83863759e-01 6.54519582e-03 2.68614322e-01 -8.42305005e-01 7.00522125e-01 -3.36133033e-01 -8.81971776e-01 9.10508096e-01 -4.23964620e-01 -3.45862925e-01 2.88826376e-01 4.03762162e-02 -1.24712281e-01 1.55334905e-01 -1.83200076e-01 -1.58487499e-01 5.54637015e-01 -1.23830438e+00 7.23094568e-02 -8.34413707e-01 1.28073528e-01 -2.80733377e-01 4.27442268e-02 -3.66307169e-01 1.91626698e-01 -4.20694560e-01 6.88687146e-01 -1.11245883e+00 -5.81053972e-01 -1.47290140e-01 -6.52499795e-01 -2.30596766e-01 2.85463363e-01 -3.23468655e-01 1.02194047e+00 -1.87885153e+00 5.08931935e-01 7.39730716e-01 6.62927151e-01 -6.69011176e-01 -6.95420355e-02 4.96333897e-01 -8.13270956e-02 2.72108138e-01 -4.39845771e-01 -1.80446804e-01 5.35365403e-01 3.48928511e-01 -1.82189420e-01 1.13797200e+00 1.56919174e-02 8.26100945e-01 -9.32569504e-01 -3.06041569e-01 8.25635642e-02 4.73888695e-01 -8.95168364e-01 -3.92969847e-01 -2.26019457e-01 1.81451261e-01 -6.33090317e-01 4.45747018e-01 1.06720150e+00 -4.22967494e-01 3.82698059e-01 -8.08448419e-02 8.99115875e-02 1.89248413e-01 -1.20747554e+00 1.53551948e+00 4.52373363e-02 4.04035985e-01 4.70289022e-01 -1.39648223e+00 3.68384510e-01 -1.00532956e-01 2.46274352e-01 -2.08368823e-01 2.11181305e-02 3.16520512e-01 -2.68253237e-01 -2.35075906e-01 4.35350984e-01 -4.49255198e-01 -4.06304747e-01 3.30511242e-01 -1.86012253e-01 -2.35857263e-01 -4.93347012e-02 4.26993072e-01 1.50343716e+00 -2.80145276e-02 -7.18880817e-02 -7.86589265e-01 1.87243566e-01 -2.21256047e-01 2.01804981e-01 1.17828965e+00 1.33059219e-01 3.49224776e-01 6.44487381e-01 -6.93409741e-02 -1.21661484e+00 -1.40410781e+00 -4.01118577e-01 8.41738462e-01 1.19191244e-01 -6.82250679e-01 -7.02268720e-01 -8.30162689e-03 -1.19277991e-01 2.52556741e-01 -9.11294281e-01 -1.20116912e-01 -2.29163006e-01 -1.28408778e+00 6.06441081e-01 1.46372020e-01 6.50788918e-02 -4.99155112e-02 -9.05619934e-02 -7.22389743e-02 -3.52260843e-02 -1.02130926e+00 1.93280846e-01 4.81038004e-01 -1.17935348e+00 -1.16705430e+00 -3.31002593e-01 -4.59170431e-01 7.24705815e-01 3.13406467e-01 1.14542878e+00 2.83472657e-01 -5.28336644e-01 5.57011604e-01 -1.75499722e-01 -5.67831807e-02 -3.78019959e-01 -8.27748030e-02 2.84935683e-01 -3.27352732e-01 4.51003492e-01 -8.37896645e-01 -4.46867466e-01 4.66097370e-02 -1.11625886e+00 -1.31882995e-01 5.69255829e-01 7.44645119e-01 6.32538438e-01 3.28785896e-01 1.32920772e-01 -6.56588078e-01 3.34185660e-01 -6.85899138e-01 -6.90902591e-01 2.76663899e-01 -4.25435632e-01 4.85696673e-01 4.93860543e-01 2.11064573e-02 -4.48232144e-01 -1.02182277e-01 -1.00690342e-01 9.86397117e-02 1.16718329e-01 6.53208673e-01 -1.63166299e-01 -3.75294715e-01 6.77156329e-01 3.59797269e-01 -1.49243429e-01 -4.13537830e-01 5.61975002e-01 8.20647776e-02 5.34882545e-01 -1.18579721e+00 1.25596774e+00 7.43302226e-01 9.15711522e-01 -1.28329217e+00 -1.03110015e+00 -1.44371822e-01 -5.12181520e-01 2.60951608e-01 7.36043036e-01 -4.65200394e-01 -8.39797258e-01 1.86560005e-01 -8.33279252e-01 -3.15522969e-01 -1.14184275e-01 3.55451941e-01 -9.30758119e-01 7.92172968e-01 -6.59765303e-01 -1.29564953e+00 4.36138898e-01 -6.39025629e-01 1.16818929e+00 -1.12583652e-01 -1.80855487e-02 -1.27697563e+00 5.38446605e-01 3.60158890e-01 2.11762711e-01 6.18704319e-01 1.35196829e+00 -3.34338069e-01 -7.54583538e-01 -1.05322480e-01 -1.97725698e-01 -5.28361015e-02 -4.83097702e-01 1.56009972e-01 -9.30210173e-01 -3.28830272e-01 2.80499011e-01 -9.33502987e-02 1.06013155e+00 6.85596764e-01 1.13288414e+00 -4.16347206e-01 -4.76699919e-01 6.18824422e-01 1.89650083e+00 -6.95304513e-01 8.89387369e-01 -1.36063471e-01 5.67501247e-01 4.18419570e-01 -8.36282298e-02 4.60817248e-01 2.01482967e-01 5.52227080e-01 3.79861712e-01 1.81191191e-02 4.73222136e-01 -1.86236799e-01 1.84880883e-01 1.07215643e+00 -2.89543808e-01 -1.29637524e-01 -4.61321652e-01 2.80519247e-01 -1.88656259e+00 -1.37339354e+00 -8.01870286e-01 2.54837894e+00 9.51452732e-01 2.61324525e-01 1.62334248e-01 3.37483943e-01 7.56583691e-01 1.35285944e-01 -2.07647651e-01 -3.74837577e-01 -4.81569797e-01 5.58653235e-01 7.47339129e-01 9.19789493e-01 -9.63599384e-01 4.73021567e-01 7.47464037e+00 1.10783505e+00 -2.43937224e-01 3.51505995e-01 4.95632559e-01 6.81359768e-02 -8.56070697e-01 2.00252965e-01 -6.13412440e-01 3.87202978e-01 9.38251674e-01 -3.52902152e-02 5.58641374e-01 7.47927189e-01 1.16996150e-02 -3.90260220e-01 -1.30991888e+00 7.82161474e-01 -9.21809673e-02 -1.38613260e+00 -4.83263344e-01 6.36270583e-01 9.47489917e-01 3.61316614e-02 2.99802818e-03 -5.31936511e-02 5.76127350e-01 -1.31538141e+00 3.60532939e-01 6.44733191e-01 6.04317129e-01 -6.10492826e-01 3.44469637e-01 3.54437739e-01 -1.09875560e+00 3.77982557e-01 -6.91122711e-01 -3.15475881e-01 6.67679869e-03 1.21379375e+00 -2.66014356e-02 6.79577827e-01 3.74903738e-01 3.15867960e-01 -3.68388742e-01 8.90278876e-01 6.43759686e-03 5.77858329e-01 -9.72150028e-01 -1.49075896e-01 2.27519758e-02 -4.17175144e-01 5.44788182e-01 1.29068637e+00 3.39566529e-01 4.27439570e-01 -4.38525341e-02 1.03676856e+00 2.64213443e-01 -2.40794167e-01 -7.97467768e-01 -2.26300642e-01 2.11415127e-01 1.00082231e+00 -8.78683984e-01 9.52585563e-02 -3.43687564e-01 8.46246183e-01 1.29721776e-01 2.41458699e-01 -9.57448542e-01 -2.36017108e-01 6.44923747e-01 2.79682338e-01 1.96045548e-01 -9.01195705e-01 -5.26277483e-01 -1.33629203e+00 8.58703330e-02 -4.67390865e-01 1.90262765e-01 -4.76359725e-01 -1.58214676e+00 -7.84795806e-02 -7.06649795e-02 -5.08154035e-01 1.12477824e-01 -8.24424684e-01 -3.45882326e-01 7.27047503e-01 -1.11880457e+00 -8.38250518e-01 -1.49430083e-02 6.15310073e-01 -2.97847241e-01 5.43707728e-01 9.87069488e-01 -3.98428962e-02 -3.69316041e-01 3.79775643e-01 6.39676094e-01 -2.43067667e-01 1.18833654e-01 -1.50436640e+00 2.56608240e-02 8.38864565e-01 1.80315048e-01 7.76060224e-01 1.20746589e+00 -5.16889453e-01 -2.16361189e+00 -5.59711039e-01 8.43150079e-01 -9.38882530e-01 9.99954283e-01 -5.98038793e-01 -6.96364760e-01 4.91049677e-01 -3.10369164e-01 7.30472850e-03 7.75775969e-01 5.64621449e-01 -6.62209332e-01 2.80069709e-01 -1.21056914e+00 3.95158857e-01 1.28271818e+00 -7.16556966e-01 -4.62261021e-01 8.38544369e-01 4.58480984e-01 -1.67577956e-02 -7.67855465e-01 4.14859235e-01 5.15657187e-01 -1.09244275e+00 1.06298506e+00 -6.23696387e-01 5.65548718e-01 -2.10522324e-01 -7.79173851e-01 -6.78590119e-01 -4.35233653e-01 -9.28964615e-01 -3.72253954e-01 6.59646928e-01 3.06207865e-01 -5.81363618e-01 8.89959037e-01 5.39606512e-01 6.82826946e-03 -8.98468316e-01 -1.18912208e+00 -1.09197438e+00 4.19485420e-01 -7.94166625e-01 -1.03296369e-01 1.09673202e+00 2.12572798e-01 3.16515386e-01 -2.12795109e-01 4.62062448e-01 1.23083425e+00 -1.51830897e-01 5.70323288e-01 -1.65606356e+00 -8.40915561e-01 -6.49892926e-01 -8.75786126e-01 -1.12428796e+00 8.48602504e-02 -9.83323216e-01 -6.03600107e-02 -1.36650085e+00 6.76988184e-01 -7.16345012e-01 -1.73153892e-01 4.97376360e-03 1.99839830e-01 4.80028987e-01 -3.98798525e-01 1.17538162e-01 -6.43207014e-01 3.20012629e-01 6.57123804e-01 1.33000806e-01 1.94385946e-01 -1.67755708e-01 -8.91518950e-01 8.36640000e-01 4.56626147e-01 -5.83689451e-01 -1.77436709e-01 -1.43059462e-01 1.11946344e+00 -3.72492634e-02 9.30216193e-01 -7.86973536e-01 8.88404474e-02 -4.01907563e-01 3.22529495e-01 -4.23517853e-01 3.18834424e-01 -5.05089641e-01 1.36944875e-01 5.42561650e-01 -3.31132859e-01 -2.10566416e-01 -1.20772682e-01 9.76400077e-01 4.16869015e-01 -3.59011710e-01 9.05273914e-01 -1.12463720e-02 -1.95837200e-01 7.31040165e-02 -1.96872875e-02 1.96458369e-01 8.42467427e-01 -2.69500762e-01 -4.33881551e-01 -4.50916231e-01 -1.18814504e+00 -1.94189921e-01 6.51779115e-01 -3.85360301e-01 3.57568473e-01 -1.13858056e+00 -8.50800812e-01 1.59529280e-02 -1.33352265e-01 -3.40467840e-01 4.01839018e-01 1.31954658e+00 -3.80245507e-01 5.16070306e-01 3.74778479e-01 -6.88734710e-01 -9.76795077e-01 6.56276166e-01 1.67373762e-01 -1.91977248e-01 -3.77642065e-01 1.06099415e+00 3.11605513e-01 -1.36012271e-01 -7.53464401e-02 -3.95477653e-01 9.09958899e-01 -3.73758674e-01 3.78197581e-01 5.44166744e-01 -6.78278580e-02 -3.32978606e-01 -5.22905767e-01 7.44710684e-01 3.33256871e-01 -2.91404247e-01 1.46689498e+00 -2.67053038e-01 -5.35282314e-01 5.89709044e-01 1.16177869e+00 4.20612276e-01 -8.76943111e-01 -2.71284908e-01 3.84950601e-02 -3.76595587e-01 -1.12398937e-01 -4.54633147e-01 -4.48564351e-01 8.69673312e-01 2.51718581e-01 9.79111433e-01 1.02212524e+00 5.88634729e-01 4.38126564e-01 5.00181198e-01 7.76327729e-01 -1.01603639e+00 -7.28482082e-02 4.50771928e-01 6.46246970e-01 -8.80084693e-01 2.38640413e-01 -6.47660673e-01 4.32551533e-01 1.10974896e+00 -1.90141752e-01 -4.45490211e-01 6.78095102e-01 5.14788687e-01 -1.01750839e+00 -4.10798013e-01 -5.56547225e-01 -2.22325429e-01 1.43380493e-01 4.94057953e-01 4.73987550e-01 1.80627942e-01 -3.56667191e-01 4.38732862e-01 -4.09788847e-01 -2.72064060e-01 6.13012373e-01 6.54677451e-01 -7.36850500e-01 -9.85292315e-01 -2.79389679e-01 4.29790348e-01 -4.42781657e-01 -2.05457017e-01 -4.77585763e-01 6.94932818e-01 -9.38279405e-02 8.63837302e-01 -2.33317286e-01 -2.67981887e-01 -2.98351377e-01 -3.55316773e-02 1.28079391e+00 -5.01461864e-01 3.20385367e-01 6.94434494e-02 -4.27147411e-02 -5.05978167e-01 -5.22251487e-01 -6.43667161e-01 -8.04219246e-01 -1.02350223e+00 -5.16775548e-01 3.36246639e-01 6.79132700e-01 1.23853505e+00 1.11682154e-01 9.44478288e-02 4.94491845e-01 -8.81847084e-01 -8.89506698e-01 -5.92584908e-01 -9.00589705e-01 -1.29744008e-01 3.10841888e-01 -3.43125910e-01 -9.33691263e-01 -1.74023688e-01]
[6.871797561645508, 5.0019636154174805]
4055b9f4-82e8-486c-996d-59a762b788b2
an-intelligent-decision-support-ensemble
2210.14906
null
https://arxiv.org/abs/2210.14906v1
https://arxiv.org/pdf/2210.14906v1.pdf
An Intelligent Decision Support Ensemble Voting Model for Coronary Artery Disease Prediction in Smart Healthcare Monitoring Environments
Coronary artery disease (CAD) is one of the most common cardiac diseases worldwide and causes disability and economic burden. It is the world's leading and most serious cause of mortality, with approximately 80% of deaths reported in low- and middle-income countries. The preferred and most precise diagnostic tool for CAD is angiography, but it is invasive, expensive, and technically demanding. However, the research community is increasingly interested in the computer-aided diagnosis of CAD via the utilization of machine learning (ML) methods. The purpose of this work is to present an e-diagnosis tool based on ML algorithms that can be used in a smart healthcare monitoring system. We applied the most accurate machine learning methods that have shown superior results in the literature to different medical datasets such as RandomForest, XGboost, MLP, J48, AdaBoost, NaiveBayes, LogitBoost, KNN. Every single classifier can be efficient on a different dataset. Thus, an ensemble model using majority voting was designed to take advantage of the well-performed single classifiers, Ensemble learning aims to combine the forecasts of multiple individual classifiers to achieve higher performance than individual classifiers in terms of precision, specificity, sensitivity, and accuracy; furthermore, we have benchmarked our proposed model with the most efficient and well-known ensemble models, such as Bagging, Stacking methods based on the cross-validation technique, The experimental results confirm that the ensemble majority voting approach based on the top 3 classifiers: MultilayerPerceptron, RandomForest, and AdaBoost, achieves the highest accuracy of 88,12% and outperforms all other classifiers. This study demonstrates that the majority voting ensemble approach proposed above is the most accurate machine learning classification approach for the prediction and detection of coronary artery disease.
['El Houssine El Mazoudi', 'Noureddine Elalami', 'Jamila Elalami', 'Anas Maach']
2022-10-25
null
null
null
null
['disease-prediction']
['medical']
[-3.28484744e-01 -3.96321267e-01 -3.10830981e-01 -1.81293726e-01 -1.92789882e-01 -1.78533494e-01 2.42317110e-01 3.02079409e-01 -1.86641634e-01 1.06196702e+00 -3.04432958e-02 -8.92047286e-01 -6.37911916e-01 -7.25001276e-01 2.53803104e-01 -7.40218520e-01 1.33944929e-01 6.76491022e-01 1.47067532e-01 1.84979793e-02 4.24908400e-01 7.18768537e-01 -1.75793529e+00 4.56747025e-01 1.12275529e+00 1.06677389e+00 -2.83493459e-01 6.20481074e-01 -1.10981613e-01 1.01206398e+00 -3.48503292e-01 -2.95124501e-01 -1.52279675e-01 -6.20756924e-01 -5.30312300e-01 -4.64318931e-01 -2.29020581e-01 -1.65343851e-01 3.41679722e-01 2.80072331e-01 8.25956881e-01 -3.10306519e-01 1.05366421e+00 -9.77426589e-01 -4.39127609e-02 2.88083553e-01 -4.36844617e-01 4.06725615e-01 4.86364625e-02 -1.98331892e-01 2.26851210e-01 -7.31741428e-01 1.90041155e-01 1.11854672e+00 9.42454696e-01 2.53575891e-01 -8.04845929e-01 -7.75527358e-01 -2.54530966e-01 4.29927886e-01 -9.89978969e-01 -1.53167751e-02 4.33439553e-01 -8.08479071e-01 6.61829948e-01 5.47947586e-01 7.59007514e-01 5.10683239e-01 8.19229126e-01 2.05287442e-01 1.67176306e+00 -8.01785052e-01 2.60396242e-01 5.35670221e-01 6.40767097e-01 7.28812337e-01 5.79273343e-01 3.67871732e-01 -3.38182189e-02 -6.84087813e-01 3.68285775e-01 6.35032713e-01 -7.69340992e-02 -7.71711245e-02 -7.44046926e-01 9.36003625e-01 2.62266457e-01 5.75315297e-01 -6.56311274e-01 -3.20255131e-01 4.62866843e-01 1.39182448e-01 4.76715118e-01 1.58115253e-01 -7.00350642e-01 7.41840675e-02 -7.39359260e-01 7.02912882e-02 8.74843180e-01 -1.56120583e-01 3.70793417e-02 -1.18959330e-01 -1.37514532e-01 6.97332680e-01 5.87695539e-01 6.57057881e-01 9.03278947e-01 -5.43425679e-01 -3.14472951e-02 1.02917397e+00 -2.53537484e-03 -1.12796676e+00 -3.43501627e-01 -8.40794981e-01 -1.21749496e+00 6.59933805e-01 1.79160923e-01 -3.15834969e-01 -7.81564832e-01 8.47763598e-01 2.37316847e-01 2.22372070e-01 1.18870363e-01 4.39523041e-01 9.49393690e-01 2.03531086e-01 6.90138757e-01 -3.61261070e-01 1.45277894e+00 -5.93976557e-01 -5.36977649e-01 3.05795997e-01 5.56732535e-01 -8.40586543e-01 9.96435061e-02 5.89757204e-01 -4.56077456e-01 -6.46404088e-01 -9.23906982e-01 6.88459694e-01 -4.48206455e-01 4.70774502e-01 6.53416693e-01 1.12399280e+00 -4.81950849e-01 6.24435246e-01 -7.03147948e-01 -4.37144578e-01 4.11413580e-01 2.24420756e-01 -3.17347676e-01 -1.43751502e-01 -8.18375587e-01 1.40281439e+00 2.68478274e-01 2.90045273e-02 -1.03741199e-01 -5.16163349e-01 -1.65903553e-01 -1.56880111e-01 -2.52345771e-01 -1.18652761e+00 6.96699142e-01 -8.43840539e-01 -1.33051074e+00 7.21992850e-01 -1.43826380e-01 -5.55328369e-01 5.18256545e-01 -2.28442341e-01 -5.17761946e-01 -3.47428732e-02 -1.51429921e-01 -3.54538485e-02 2.11464688e-01 -8.99867773e-01 -8.61539006e-01 -8.14431429e-01 -6.23872161e-01 -3.74479555e-02 8.73832405e-02 7.43115097e-02 7.95933068e-01 -4.22697693e-01 3.16461951e-01 -7.97724664e-01 -3.05155963e-01 -5.34678459e-01 -1.38995498e-01 -2.08758384e-01 9.18586195e-01 -1.02526963e+00 1.51773143e+00 -1.90971243e+00 -2.34245539e-01 5.04330516e-01 5.24931103e-02 6.62047029e-01 1.03040242e+00 2.96607137e-01 -1.17148250e-01 3.03103805e-01 -4.79490198e-02 4.23136294e-01 -6.25096381e-01 2.51171321e-01 -1.75955564e-01 1.11013114e-01 -2.14450508e-01 4.31221008e-01 -4.66688722e-01 -6.92988992e-01 8.75505149e-01 5.13999104e-01 -5.67658544e-02 8.46518949e-02 5.36380470e-01 5.32572448e-01 -5.19923627e-01 7.36639261e-01 3.72571081e-01 -2.51403928e-01 3.76469582e-01 -2.49365598e-01 -1.19161360e-01 -4.20495421e-01 -1.29456210e+00 6.83866680e-01 -2.35727608e-01 3.42989713e-02 -6.90570951e-01 -1.30608165e+00 1.30922866e+00 6.77764952e-01 4.42633301e-01 -1.46947891e-01 2.81400919e-01 7.70527840e-01 1.24393731e-01 -8.07522893e-01 -6.19797826e-01 -7.06966519e-02 3.90438020e-01 1.04043357e-01 -2.04265743e-01 4.92340565e-01 -1.17043793e-01 -3.42992365e-01 7.41433620e-01 -8.49528536e-02 1.16066635e+00 -2.06479698e-01 9.41710651e-01 1.19639143e-01 4.98900861e-01 7.16655314e-01 -2.22890973e-01 1.88870206e-01 1.78844750e-01 -9.11385775e-01 -5.66315114e-01 -7.13410139e-01 -4.90222335e-01 6.09501719e-01 -2.87104756e-01 2.19774559e-01 -3.17590326e-01 -6.87330246e-01 2.72440344e-01 5.50432801e-01 -1.47780836e-01 1.01731814e-01 -5.04683614e-01 -1.42192531e+00 5.53315401e-01 3.59577388e-01 9.67761695e-01 -6.38886690e-01 -9.00637805e-01 5.39214730e-01 1.24278829e-01 -2.38832504e-01 8.09704304e-01 8.03515539e-02 -1.56860113e+00 -1.41982365e+00 -6.34778559e-01 -5.97779036e-01 2.32244626e-01 -1.73733339e-01 1.06112957e+00 4.11401600e-01 -4.20004129e-01 -7.79565200e-02 -4.96896386e-01 -7.68526673e-01 -4.65441167e-01 -6.53625503e-02 -4.28679764e-01 1.00713791e-02 5.48385978e-01 -6.25381231e-01 -8.49645078e-01 1.60535485e-01 -5.15285768e-02 -2.92332828e-01 9.19759989e-01 8.94326925e-01 3.99939239e-01 2.44802997e-01 8.96273017e-01 -1.22049463e+00 3.79241526e-01 -7.87821770e-01 -1.53367981e-01 2.97252536e-01 -1.07186162e+00 -1.99729189e-01 2.43541852e-01 -5.24072871e-02 -9.91505742e-01 -1.79878436e-02 -1.11118771e-01 1.76711231e-01 -4.12996441e-01 5.90727687e-01 2.03162804e-01 -1.97761342e-01 8.00872624e-01 -6.40176237e-02 2.59560883e-01 -6.74757123e-01 -1.82084441e-01 1.13328195e+00 3.17552388e-02 -1.79184988e-01 -8.93815979e-02 2.93187469e-01 4.84672338e-01 -4.72865939e-01 -5.45012414e-01 -6.33497417e-01 -4.07194287e-01 -4.61495191e-01 9.56939876e-01 -7.89261103e-01 -6.07563436e-01 5.62392950e-01 -8.94259155e-01 5.63855648e-01 4.07860279e-01 9.80194926e-01 -3.20582539e-02 1.08186387e-01 -1.91713795e-01 -1.22454667e+00 -8.10705602e-01 -9.02117789e-01 4.33263481e-01 2.71494418e-01 -4.13336456e-01 -1.02406394e+00 7.36969635e-02 3.63676846e-01 5.64536035e-01 7.96111524e-01 1.39298487e+00 -9.77055967e-01 -8.68469477e-02 -5.08341551e-01 8.38703196e-03 5.64557076e-01 1.53440356e-01 3.06812823e-01 -8.27896535e-01 1.66383013e-01 -1.74876880e-02 2.88394511e-01 8.62176955e-01 8.62588942e-01 9.76894915e-01 6.41079294e-03 -9.14838374e-01 9.39751342e-02 1.68734312e+00 8.95293057e-01 5.66976190e-01 5.07064760e-01 1.11892700e-01 2.29594156e-01 4.25070584e-01 2.31049612e-01 3.43488693e-01 2.87723601e-01 2.48363212e-01 -1.21371403e-01 -2.22995039e-02 2.41853490e-01 -1.84342921e-01 5.42372406e-01 -6.82255089e-01 1.68296658e-02 -1.14497209e+00 -3.10114734e-02 -1.98788869e+00 -1.28647459e+00 -5.07121563e-01 2.39026594e+00 2.25882769e-01 8.92536715e-02 3.35928559e-01 9.52966750e-01 7.24172652e-01 -6.90970957e-01 -6.51725233e-02 -4.97000277e-01 1.02750130e-01 6.16244197e-01 4.58677649e-01 8.80839899e-02 -1.32864797e+00 -4.03236039e-02 6.10096788e+00 2.83996880e-01 -1.30748987e+00 1.50614142e-01 1.15789557e+00 4.17233258e-01 4.05039996e-01 -2.29107328e-02 -7.00867653e-01 9.28031445e-01 1.05970216e+00 1.47776961e-01 -2.57004499e-01 1.01484501e+00 2.10031182e-01 -3.57910037e-01 -5.91412365e-01 7.35910714e-01 -5.37501241e-04 -1.20877290e+00 -2.54616767e-01 -6.69574738e-02 6.05470240e-01 -2.23243326e-01 -2.42263019e-01 2.71630019e-01 3.26057851e-01 -9.81089950e-01 -5.87141551e-02 8.88817072e-01 2.25581929e-01 -6.03953481e-01 1.62769759e+00 4.73580450e-01 -5.79017401e-01 -5.30458331e-01 1.62829250e-01 -4.61454391e-01 8.10393393e-02 1.18058205e+00 -8.46224189e-01 8.05660605e-01 9.54155087e-01 4.77789849e-01 -5.22465289e-01 1.46324944e+00 9.50232372e-02 8.49909186e-01 -1.91782445e-01 -2.59127855e-01 -2.64364392e-01 4.13286202e-02 1.58405468e-01 8.08980405e-01 7.82481313e-01 2.32580349e-01 2.74666011e-01 5.58488704e-02 7.85624027e-01 6.95905328e-01 -4.30409014e-01 6.69497848e-01 6.24563038e-01 8.53556156e-01 -6.81597710e-01 -6.71667695e-01 -1.89851150e-01 1.97077736e-01 -3.33363771e-01 -5.78407682e-02 -6.18799388e-01 -1.79221421e-01 -1.47972414e-02 1.77784801e-01 1.12701654e-01 5.74613988e-01 -8.16926956e-01 -6.64523840e-01 -2.53917187e-01 -9.11362886e-01 8.60826731e-01 -4.76557761e-01 -1.33045506e+00 5.89644432e-01 8.20629820e-02 -1.22513092e+00 -3.58054399e-01 -9.21777248e-01 -5.57338655e-01 1.20349324e+00 -1.21710408e+00 -1.01476216e+00 -5.32507956e-01 2.14472532e-01 1.12969287e-01 -6.32460058e-01 1.22042918e+00 3.65879565e-01 -6.39240086e-01 -7.08216652e-02 3.14521492e-01 -1.47236243e-01 6.83764458e-01 -1.06960475e+00 -7.31696784e-01 2.37163395e-01 -6.37644291e-01 3.30151141e-01 4.32989091e-01 -6.00157082e-01 -3.75168890e-01 -7.50810683e-01 1.16544890e+00 -2.40610883e-01 -1.14999451e-02 8.53913069e-01 -5.73367178e-01 3.22604239e-01 -1.00868352e-01 6.52955249e-02 1.12759447e+00 2.45094687e-01 -5.94224595e-02 -5.59462130e-01 -1.53872359e+00 -8.02673921e-02 1.29050255e-01 1.64944261e-01 -6.54271960e-01 2.41696373e-01 -3.81305933e-01 -1.56466708e-01 -1.14223468e+00 9.16880131e-01 1.15462792e+00 -1.49303067e+00 1.08598065e+00 -6.37155473e-01 3.50394219e-01 -3.41794163e-01 -1.36360541e-01 -1.11204267e+00 -4.12854850e-01 3.52648675e-01 -6.78066462e-02 1.10266590e+00 4.21706557e-01 -1.11451805e+00 5.75038671e-01 3.95708948e-01 3.14973533e-01 -1.09584534e+00 -8.75484943e-01 -3.74699384e-01 1.06722988e-01 5.73980138e-02 6.27098203e-01 8.49139571e-01 -2.30787873e-01 2.17389420e-01 -2.03945071e-01 -1.21826969e-01 6.06606960e-01 1.51957199e-01 5.94510555e-01 -1.98422539e+00 -3.89225066e-01 -4.75650162e-01 -8.48430455e-01 1.94056168e-01 -4.16127920e-01 -6.53832793e-01 -7.52841592e-01 -1.57024944e+00 3.02644879e-01 -8.42292249e-01 -5.88362932e-01 2.14928970e-01 -4.53245610e-01 1.36929467e-01 -1.22371046e-02 4.52294648e-01 2.67299682e-01 -4.69634801e-01 9.69425023e-01 1.34942636e-01 -2.75505245e-01 5.03697097e-01 -7.38714874e-01 9.72302616e-01 9.85377610e-01 -4.88926768e-01 -2.71162868e-01 4.22816575e-02 -1.41537338e-01 2.97289521e-01 5.08176208e-01 -1.13054848e+00 -4.50542606e-02 -1.53769955e-01 9.01263595e-01 -6.78963780e-01 -1.94516242e-01 -1.00582850e+00 8.11926186e-01 1.19841182e+00 1.19611651e-01 1.41330302e-01 -1.87944576e-01 3.57560962e-01 -2.68908530e-01 -2.47357681e-01 7.12078273e-01 -2.82863170e-01 -6.24716222e-01 -2.55714357e-01 -3.32643896e-01 -5.40029228e-01 1.52257681e+00 -3.28475207e-01 -3.75043631e-01 4.47066799e-02 -9.61915970e-01 -2.42388949e-01 -3.26020181e-01 -2.79321186e-02 2.63101041e-01 -1.15382361e+00 -1.00721371e+00 -8.60457420e-02 -9.78042036e-02 -4.68084216e-01 4.26641762e-01 1.17166996e+00 -1.01192164e+00 4.68115419e-01 -5.16176641e-01 -6.89481735e-01 -1.81615174e+00 2.65867233e-01 5.76841533e-01 -6.09812319e-01 -3.12718332e-01 4.68138188e-01 -6.38280988e-01 -1.29693076e-01 3.22065204e-02 -7.57538946e-03 -9.64806080e-01 2.79639773e-02 4.92968768e-01 1.23672080e+00 3.87070566e-01 -4.89686877e-01 -6.68558896e-01 7.51867115e-01 5.71346506e-02 2.93560743e-01 9.70806301e-01 2.73214519e-01 -4.27128494e-01 4.41029608e-01 6.65134728e-01 -2.99019128e-01 5.54627925e-02 3.58002990e-01 -2.68301535e-02 -1.48024842e-01 3.55340183e-01 -1.47682643e+00 -7.36100852e-01 7.47992992e-01 1.27991033e+00 1.21984124e-01 1.18639052e+00 -5.65542638e-01 2.39415362e-01 1.16574295e-01 3.50244939e-01 -6.28761053e-01 -6.14509046e-01 -1.60093412e-01 4.40558821e-01 -1.40004981e+00 2.77078390e-01 -5.07053196e-01 -4.37060684e-01 1.46353114e+00 2.33360201e-01 -2.29208499e-01 1.42622101e+00 2.74351751e-03 2.36944720e-01 2.41190329e-01 -8.29304397e-01 1.93623230e-01 2.58804232e-01 5.70130944e-01 7.88980603e-01 5.99960864e-01 -1.01478410e+00 7.68708289e-01 1.11147329e-01 7.52424002e-01 -2.03969240e-01 9.02627647e-01 -7.46689796e-01 -1.34427381e+00 -6.88344836e-01 1.01457858e+00 -8.60907614e-01 1.62237033e-01 -1.28187105e-01 8.36936831e-01 7.10208595e-01 1.06419945e+00 -1.80295974e-01 -3.02739352e-01 2.99162567e-01 5.48035681e-01 3.07779491e-01 -1.08489633e-01 -8.08058679e-01 -3.33495289e-01 1.12608947e-01 -6.02131821e-02 -7.63583839e-01 -5.60916364e-01 -7.45736003e-01 -4.89393800e-01 -3.70196700e-01 2.42386043e-01 8.89755845e-01 1.22810948e+00 4.40574855e-01 6.24188006e-01 6.04267001e-01 -2.65409797e-01 -6.42071724e-01 -1.16062820e+00 -3.21191341e-01 1.82258129e-01 -5.56987971e-02 -9.21428204e-01 -5.00646651e-01 1.64685965e-01]
[8.431498527526855, 4.857220649719238]
fa475c1e-76eb-42ab-9ea6-4b1cc4d57789
bi-lstm-price-prediction-based-on-attention
2212.03443
null
https://arxiv.org/abs/2212.03443v2
https://arxiv.org/pdf/2212.03443v2.pdf
Bi-LSTM Price Prediction based on Attention Mechanism
With the increasing enrichment and development of the financial derivatives market, the frequency of transactions is also faster and faster. Due to human limitations, algorithms and automatic trading have recently become the focus of discussion. In this paper, we propose a bidirectional LSTM neural network based on an attention mechanism, which is based on two popular assets, gold and bitcoin. In terms of Feature Engineering, on the one hand, we add traditional technical factors, and at the same time, we combine time series models to develop factors. In the selection of model parameters, we finally chose a two-layer deep learning network. According to AUC measurement, the accuracy of bitcoin and gold is 71.94% and 73.03% respectively. Using the forecast results, we achieved a return of 1089.34% in two years. At the same time, we also compare the attention Bi-LSTM model proposed in this paper with the traditional model, and the results show that our model has the best performance in this data set. Finally, we discuss the significance of the model and the experimental results, as well as the possible improvement direction in the future.
['Ye Li', 'Leyi Cui', 'Jiashu Lou']
2022-12-07
null
null
null
null
['feature-engineering']
['methodology']
[-9.02952552e-01 -4.33310449e-01 -9.06683579e-02 -8.90154913e-02 -4.65051793e-02 -4.60130185e-01 5.37928283e-01 -3.50363106e-01 -4.91508782e-01 7.29074895e-01 2.03297615e-01 -5.14624715e-01 -1.61180213e-01 -8.48432064e-01 -5.51108122e-01 -5.91613114e-01 -1.53564394e-01 1.58766046e-01 2.01456714e-02 -2.74580836e-01 5.47203660e-01 2.63339221e-01 -8.77916992e-01 9.70443189e-02 9.24580514e-01 1.82713056e+00 -1.63064733e-01 -2.08813162e-03 -2.49063686e-01 1.28665876e+00 -6.22556686e-01 -9.47717905e-01 6.49347305e-01 -4.29011226e-01 -3.78943712e-01 -5.01001298e-01 -3.24471176e-01 -7.25864589e-01 -3.75108123e-01 1.05047441e+00 3.78522784e-01 -2.27164961e-02 2.19318926e-01 -9.77125883e-01 -8.79711926e-01 1.17189801e+00 -5.80228090e-01 4.60676253e-01 -3.89616013e-01 -5.66576347e-02 1.13248837e+00 -8.36770415e-01 2.09879190e-01 7.40451813e-01 7.20074892e-01 -2.36756243e-02 -6.97883606e-01 -9.22292888e-01 1.09950036e-01 3.79563451e-01 -9.84955311e-01 -1.41713127e-01 9.79164779e-01 -4.23942953e-01 1.21597743e+00 2.65761334e-02 1.08558965e+00 7.76383996e-01 7.14131474e-01 7.47256696e-01 1.15037918e+00 -3.78365517e-01 -4.33937535e-02 1.95822775e-01 5.73245957e-02 2.67324179e-01 3.21065664e-01 2.87183881e-01 -3.42784107e-01 -1.36025622e-01 1.00925863e+00 3.27757746e-01 -1.43002450e-01 2.15173498e-01 -1.26968908e+00 9.63832259e-01 5.37030458e-01 9.24286067e-01 -7.63281703e-01 2.36028031e-01 3.64411533e-01 4.34919357e-01 6.26877546e-01 4.41306680e-01 -5.91350436e-01 -3.07599604e-01 -1.00278342e+00 1.72548503e-01 1.05449891e+00 4.65916067e-01 2.85094172e-01 5.45813024e-01 6.13208860e-02 6.52317703e-01 2.72180945e-01 3.34305257e-01 9.41778839e-01 -8.65374327e-01 5.47251165e-01 5.30487239e-01 1.26871631e-01 -1.17881119e+00 -2.33995378e-01 -1.09802198e+00 -8.52313995e-01 5.25473841e-02 3.34180117e-01 -4.49175864e-01 -6.22218192e-01 1.45188928e+00 -3.73912543e-01 -1.39127783e-02 -7.28045925e-02 8.80118906e-01 5.72643131e-02 7.56803036e-01 -2.77012527e-01 -4.44777071e-01 1.11942017e+00 -1.16332924e+00 -1.06097293e+00 2.42302403e-01 2.07761541e-01 -7.96898961e-01 5.68830848e-01 8.11482787e-01 -9.82233405e-01 -3.46458703e-01 -1.09515238e+00 3.25973064e-01 -3.33513141e-01 1.18473083e-01 9.14084613e-01 4.44712460e-01 -7.96177983e-01 9.61130142e-01 -8.18187058e-01 1.12675406e-01 1.06452256e-01 1.39806077e-01 3.33368391e-01 7.71842957e-01 -1.69346869e+00 9.47870195e-01 5.79335153e-01 3.95217240e-01 -4.79973137e-01 -4.63691592e-01 -5.31315245e-02 4.42897737e-01 -3.64197120e-02 -4.09957975e-01 1.42252743e+00 -1.27461922e+00 -1.82727289e+00 5.37914969e-02 4.17595476e-01 -1.05605221e+00 8.75329733e-01 -4.29599553e-01 -7.70869493e-01 -1.20285742e-01 -3.20670784e-01 8.97641256e-02 3.91056657e-01 -5.46890199e-01 -9.69974101e-01 -1.51043504e-01 5.76016903e-02 -1.13664530e-01 -6.96257770e-01 2.37319767e-01 -1.99930519e-02 -1.00102603e+00 4.87046093e-02 -6.74938798e-01 -5.15817329e-02 -5.11832118e-01 8.05101469e-02 -1.11849211e-01 3.91117185e-01 -1.25101840e+00 1.68763936e+00 -1.87013292e+00 -3.23025346e-01 3.63886267e-01 1.91470951e-01 2.66775727e-01 5.86012602e-01 3.19420576e-01 -1.92146525e-01 3.71348172e-01 5.93783408e-02 1.84812382e-01 6.65457547e-02 -2.74401873e-01 -6.91251218e-01 1.24092691e-01 -1.86543800e-02 1.07002926e+00 -4.96609360e-01 -2.01513425e-01 -2.93495506e-02 2.64073968e-01 -2.01047376e-01 2.31629848e-01 -5.73133193e-02 6.04433641e-02 -4.57088202e-01 6.22583866e-01 4.68407959e-01 -3.03747714e-01 1.09012589e-01 -1.84822306e-01 -4.98788208e-01 3.08744967e-01 -7.97886550e-01 1.26765871e+00 -3.17191094e-01 5.90174198e-01 -5.28876781e-01 -8.48367691e-01 1.25382459e+00 5.32671690e-01 7.08672643e-01 -1.30782807e+00 3.36889714e-01 9.53971982e-01 4.52122003e-01 -3.88492078e-01 3.73905718e-01 -3.50757062e-01 2.93513179e-01 7.21993089e-01 -3.80150217e-04 5.61306417e-01 3.24798822e-02 -1.51954740e-01 6.26949251e-01 1.00117356e-01 1.93125099e-01 -2.69056708e-01 2.69483924e-01 -2.84109980e-01 8.90754580e-01 2.24732757e-01 -1.64368093e-01 1.60212502e-01 6.00219369e-01 -7.96430945e-01 -1.07892406e+00 -4.56394076e-01 -3.26644108e-02 6.95300639e-01 -1.63054794e-01 3.72613706e-02 -4.69134957e-01 -4.13151443e-01 9.51913819e-02 8.12287807e-01 -4.67829883e-01 1.30834999e-02 -6.71125054e-01 -9.24231946e-01 5.03575087e-01 5.16311109e-01 1.16783273e+00 -1.18319416e+00 -8.25296581e-01 6.26369715e-01 -1.94689240e-02 -6.81763351e-01 -3.31316441e-01 8.36852491e-02 -1.03880608e+00 -7.27995932e-01 -1.19300413e+00 -4.51535255e-01 3.59704578e-03 -3.04670576e-02 8.73067737e-01 -6.59261644e-02 3.74510020e-01 -3.75973314e-01 -3.85120571e-01 -4.77852762e-01 6.89225495e-02 1.16235368e-01 1.97505131e-02 4.32074189e-01 3.52154315e-01 -6.21581376e-01 -8.09982300e-01 2.63732612e-01 -6.32662952e-01 -4.36157100e-02 7.09962189e-01 8.95252466e-01 7.15202466e-02 5.23205437e-02 7.84772158e-01 -5.87929547e-01 9.51790631e-01 -6.37876511e-01 -8.28724563e-01 3.93105149e-01 -1.33493626e+00 -2.24941894e-02 6.78771615e-01 -4.40095812e-01 -1.27573943e+00 -6.44006014e-01 2.61325926e-01 -7.05947101e-01 7.68561482e-01 1.00651860e+00 1.97324619e-01 1.05892494e-01 2.70211846e-01 1.43053651e-01 -1.10456444e-01 -7.70602763e-01 -1.26635298e-01 7.77090788e-01 1.02897212e-01 -1.61065429e-01 1.98983476e-01 7.13973492e-02 -4.75817382e-01 -9.53546837e-02 -4.12774056e-01 2.97145575e-01 -9.17207599e-02 -2.36209825e-01 4.90177482e-01 -7.35489070e-01 -8.93220484e-01 9.33575749e-01 -1.27202201e+00 -1.70719564e-01 -5.32567576e-02 1.00114930e+00 -1.75260022e-01 1.02521263e-01 -1.12263405e+00 -1.15506208e+00 -7.39072204e-01 -8.27793241e-01 -1.07251793e-01 3.26243818e-01 9.78208408e-02 -1.12728453e+00 -2.50062756e-02 2.03500584e-01 1.00767493e+00 2.76407778e-01 5.40879190e-01 -1.15753150e+00 -5.70761502e-01 -2.36509711e-01 -2.90975928e-01 7.79132426e-01 1.64100260e-01 6.30992949e-02 -8.01698625e-01 -1.74070820e-01 7.64132380e-01 4.99414988e-02 8.76140773e-01 3.27960938e-01 7.92530954e-01 -5.25199652e-01 1.60862118e-01 5.41963160e-01 1.31586993e+00 1.12433851e+00 6.63952112e-01 6.96605265e-01 3.51294339e-01 4.92418408e-01 6.03819489e-01 6.71819925e-01 2.34214351e-01 2.59276927e-01 2.60039955e-01 2.00602245e-02 2.29703739e-01 -3.87752652e-01 4.49029475e-01 1.52457511e+00 -4.49856013e-01 -8.16287100e-02 -9.84623134e-01 3.62097740e-01 -1.88992918e+00 -1.00344563e+00 1.02511711e-01 2.05591321e+00 5.94105661e-01 4.43439573e-01 8.11500847e-02 -5.51524647e-02 7.65836656e-01 2.00506896e-01 -6.72652960e-01 -3.95617306e-01 -2.78146684e-01 -5.85710183e-02 8.77067149e-01 2.59972692e-01 -8.05385053e-01 6.10481858e-01 6.00284863e+00 6.44127548e-01 -1.67948258e+00 1.82172254e-01 1.10070646e+00 -7.37747401e-02 -5.59174001e-01 2.30767637e-01 -4.33858782e-01 1.07770562e+00 1.12195659e+00 -6.30777180e-01 6.91845119e-01 4.81362104e-01 1.39026329e-01 3.49646986e-01 -7.26517260e-01 1.03657413e+00 -1.01656578e-01 -1.21240103e+00 2.88650542e-02 3.48624736e-01 6.74209177e-01 8.17278847e-02 1.72336504e-03 3.65573704e-01 1.04733780e-01 -8.62000585e-01 1.03595877e+00 9.11045849e-01 2.31117040e-01 -8.58443201e-01 1.28175652e+00 2.70986050e-01 -8.64642680e-01 -3.39454591e-01 -1.29790753e-01 -2.98860431e-01 1.19021468e-01 7.39748716e-01 -3.56254965e-01 8.70379865e-01 7.98062265e-01 6.96248829e-01 -1.75878957e-01 1.15540564e+00 -8.48194659e-02 7.65750468e-01 -3.40253890e-01 -2.60695517e-01 3.17947268e-01 -5.29774845e-01 5.03480695e-02 7.95397162e-01 8.75571430e-01 -1.22659784e-02 -2.80135155e-01 1.06911981e+00 -1.42535090e-01 1.92705795e-01 -1.93280816e-01 -2.48223990e-01 4.75101680e-01 1.03499317e+00 -4.24391240e-01 -3.53235245e-01 -4.98214304e-01 5.55344284e-01 1.31014287e-01 5.07454991e-01 -1.06690311e+00 -7.60756373e-01 1.46886364e-01 -1.51960701e-01 3.11788857e-01 -1.70032471e-01 -4.71372336e-01 -1.50961697e+00 3.23784441e-01 -7.50013530e-01 2.56336600e-01 -4.43475991e-01 -1.26522470e+00 7.41824448e-01 -3.94308537e-01 -1.44053173e+00 -2.89442122e-01 -5.58614075e-01 -6.04363263e-01 9.73630071e-01 -1.59715128e+00 -6.36173666e-01 1.59586281e-01 2.79485315e-01 2.33000755e-01 -6.20454609e-01 3.65134746e-01 6.29843473e-01 -5.89870989e-01 4.68470991e-01 5.10808170e-01 2.81838506e-01 5.61487019e-01 -9.26679134e-01 5.04267812e-01 8.67977560e-01 -2.28588656e-02 7.67669678e-01 3.28036398e-01 -6.52125120e-01 -1.05630040e+00 -5.18506467e-01 1.00279009e+00 2.41540462e-01 1.23282015e+00 2.96673272e-03 -9.20113683e-01 8.61801088e-01 6.78526521e-01 -2.82455415e-01 4.31510299e-01 -1.39858156e-01 -1.13832206e-01 -6.11956000e-01 -9.81605649e-01 3.05016816e-01 4.88145411e-01 -1.74829379e-01 -6.10907078e-01 -1.65679276e-01 8.36757660e-01 -7.97849298e-02 -1.23028028e+00 3.85030031e-01 9.35600281e-01 -1.23052227e+00 3.23925704e-01 -4.65454310e-01 3.17436814e-01 1.03801988e-01 -6.67124614e-02 -1.25225317e+00 -5.56479216e-01 -8.37748885e-01 -3.70609879e-01 1.46056890e+00 5.25079966e-01 -1.37924409e+00 5.16858220e-01 8.80003810e-01 9.37410891e-02 -9.30839777e-01 -1.05637622e+00 -8.85084629e-01 1.83056459e-01 -1.13635860e-01 1.00327837e+00 1.09740925e+00 1.32887945e-01 2.52196640e-01 -7.25966334e-01 -6.19649291e-01 3.31175983e-01 5.58957338e-01 1.93895683e-01 -1.14230061e+00 -4.56227720e-01 -7.57802784e-01 -1.11569621e-04 -1.07345831e+00 -2.86754698e-01 -4.87233460e-01 -6.29402995e-01 -1.04749572e+00 5.37844971e-02 -3.74729812e-01 -1.19366634e+00 2.37136647e-01 1.50205985e-01 -3.30645055e-01 3.72087926e-01 5.92934072e-01 -3.79985608e-02 7.41750538e-01 1.16882169e+00 -1.11572422e-01 -1.09570287e-01 -2.58611310e-02 -7.72832394e-01 5.19907057e-01 1.23442030e+00 -4.64822911e-02 3.19705270e-02 -6.40820861e-01 3.16505462e-01 2.59048909e-01 -4.77061048e-02 -6.09887898e-01 2.61585027e-01 -9.61986110e-02 4.52666849e-01 -3.59141290e-01 1.00860268e-01 -9.51960266e-01 3.41565549e-01 8.97448599e-01 -2.22302064e-01 4.29986507e-01 -2.39041463e-01 1.57750711e-01 -5.66902697e-01 -8.76667276e-02 3.81786346e-01 -1.59613326e-01 -3.27284038e-01 2.23635510e-01 -2.97698900e-02 -2.47623309e-01 8.67029786e-01 8.92703533e-02 -3.79336357e-01 -4.71626073e-01 -3.32981229e-01 3.54273856e-01 9.47267190e-03 4.46233034e-01 4.48259145e-01 -1.63209617e+00 -7.25610137e-01 4.49408293e-02 -5.96558154e-01 -5.72705805e-01 1.00081153e-02 1.11223996e+00 -7.94976175e-01 8.75400364e-01 -4.15076822e-01 1.80225372e-02 -4.79108274e-01 5.82169831e-01 5.94309688e-01 -4.72145736e-01 -3.19385409e-01 3.78398329e-01 -5.74817881e-02 1.77594364e-01 2.72383809e-01 -6.26309514e-01 -3.09756070e-01 3.82106125e-01 3.38040680e-01 6.90418661e-01 1.51925921e-01 -2.19304442e-01 -1.47652104e-01 2.98986822e-01 -1.09733276e-01 -3.34385574e-01 1.59662831e+00 1.89644858e-01 -4.43360031e-01 8.06697905e-01 9.89503384e-01 -9.40792486e-02 -1.02847087e+00 -2.36081913e-01 4.54679579e-01 -5.83029926e-01 -1.67786721e-02 -9.53023255e-01 -1.74294126e+00 9.00863528e-01 5.60912430e-01 7.19340086e-01 1.07448685e+00 -6.27386034e-01 1.25579917e+00 1.36536479e-01 5.08034110e-01 -1.26599205e+00 -2.53340185e-01 5.46202183e-01 9.08829927e-01 -8.28799427e-01 -1.98998719e-01 3.69669497e-01 -5.27115881e-01 1.32100034e+00 4.35451955e-01 -3.82674217e-01 7.85558760e-01 2.66288724e-02 2.38355309e-01 4.78542335e-02 -9.03220952e-01 4.95384932e-01 -1.37594178e-01 -3.57408494e-01 5.90824902e-01 1.16884150e-01 -8.94144595e-01 1.20547068e+00 -3.70371073e-01 3.44675779e-01 3.81595731e-01 5.22563159e-01 -2.97832161e-01 -1.00754583e+00 -3.35450500e-01 5.14793992e-01 -1.10539842e+00 -1.15165606e-01 -2.28285655e-01 6.19707584e-01 -8.45890716e-02 7.53002942e-01 8.95162597e-02 -7.09674358e-01 3.53296906e-01 1.73107579e-01 -1.19904898e-01 1.63366884e-01 -1.04112041e+00 4.32383150e-01 6.61879629e-02 -3.51870954e-01 -2.38031760e-01 -6.06986165e-01 -1.07271075e+00 -5.93466520e-01 -5.71305037e-01 4.63171899e-01 6.57695234e-01 6.23668253e-01 4.05596912e-01 5.59010625e-01 1.08035362e+00 -5.26919305e-01 -1.02257025e+00 -1.30494821e+00 -8.58461857e-01 1.42041102e-01 1.64963394e-01 -5.71933925e-01 -4.38780844e-01 -2.80819118e-01]
[4.440049171447754, 4.2337775230407715]
39389e25-ff4e-43b7-b420-741ba9939684
sisua-semi-supervised-generative-autoencoder
null
null
https://www.biorxiv.org/content/10.1101/631382v1
https://www.biorxiv.org/content/biorxiv/early/2019/05/08/631382.full-text.pdf
SISUA: Semi-Supervised Generative Autoencoder for Single Cell Data
Single-cell transcriptomics offers a tool to study the diversity of cell phenotypes through snapshots of the abundance of mRNA in individual cells. Often there is additional information available besides the single cell gene expression counts, such as bulk transcriptome data from the same tissue, or quantification of surface protein levels from the same cells. In this study, we propose models based on the Bayesian generative approach, where protein quantification available as CITE-seq counts from the same cells are used to constrain the learning process, thus forming a semi-supervised model. The generative model is based on the deep variational autoencoder (VAE) neural network architecture.
['Merja Heinäniemi', 'Ville Hautamäki', 'Gerardo González', 'Juha Mehtonen', 'Roger Kramer', 'Trung Ngo Trong']
2019-05-08
null
null
null
icml-workshop-on-computational-biology-2019
['single-cell-modeling']
['medical']
[-6.98349718e-03 -2.92278886e-01 1.58896834e-01 -1.40779331e-01 -6.30246401e-01 -6.32677853e-01 5.73927224e-01 1.88369751e-01 -5.88341951e-01 1.20560658e+00 2.13698059e-01 2.72713184e-01 8.40683281e-02 -8.65833819e-01 -5.44290125e-01 -1.55623198e+00 5.05407095e-01 9.23084676e-01 -3.06335330e-01 2.23567918e-01 -1.38969287e-01 6.00249231e-01 -1.11103821e+00 8.55087265e-02 3.58492643e-01 8.67432714e-01 -2.34624166e-02 7.50060201e-01 -4.37886029e-01 4.71320391e-01 -3.21036011e-01 -1.56127229e-01 5.57982288e-02 -5.20022333e-01 -8.10475200e-02 1.56808913e-01 -8.78421962e-02 -2.29225859e-01 -1.61241263e-01 1.07389092e+00 5.07936954e-01 4.58788127e-03 9.75367606e-01 -8.97552729e-01 -1.40162572e-01 3.68836164e-01 -2.03260481e-01 1.07154407e-01 -2.63690520e-02 3.21380109e-01 8.49860787e-01 -8.33760023e-01 9.61587310e-01 1.10371804e+00 2.40769491e-01 5.36783814e-01 -1.90909874e+00 -1.41005859e-01 -3.11871350e-01 -3.13233525e-01 -1.60091925e+00 -4.46205676e-01 5.02681255e-01 -8.90558124e-01 6.89554989e-01 -6.96058571e-02 1.08368361e+00 1.31094420e+00 6.61685288e-01 4.47969526e-01 1.31860721e+00 -2.12556913e-01 9.84866261e-01 -2.34673291e-01 -2.13884264e-01 5.37376642e-01 1.74671754e-01 -1.39947593e-01 -7.10398912e-01 -4.74221170e-01 7.54060805e-01 4.13381845e-01 -1.27777860e-01 -3.25028270e-01 -1.10938561e+00 7.89116442e-01 -3.18406582e-01 5.95593810e-01 -6.44675314e-01 3.77882928e-01 4.28786546e-01 -4.15899307e-02 6.48593903e-01 9.74331498e-02 -8.91825020e-01 -1.66275397e-01 -1.10140288e+00 9.02615413e-02 8.54080319e-01 4.72019374e-01 9.75605190e-01 2.78564189e-02 -1.67183861e-01 4.47422504e-01 7.47178078e-01 4.49718952e-01 4.73992318e-01 -1.20709002e+00 -4.82944816e-01 5.92925787e-01 1.57016322e-01 -3.92104596e-01 -2.49338269e-01 -9.42148790e-02 -8.85187924e-01 1.57002792e-01 7.94518650e-01 -5.31143785e-01 -9.71307278e-01 1.95800757e+00 4.26251382e-01 5.77048510e-02 6.10140078e-02 7.51039267e-01 5.02196968e-01 6.77509129e-01 2.00763792e-01 -8.45879734e-01 1.14000082e+00 -9.08609703e-02 -8.46448481e-01 3.16303551e-01 3.66090745e-01 -1.94350794e-01 4.38842028e-01 3.11531425e-01 -7.50255287e-01 1.70833189e-02 -6.27165675e-01 -1.15238152e-01 -7.62784660e-01 -8.66803527e-02 3.48965406e-01 4.96698350e-01 -9.25900042e-01 9.09287095e-01 -1.22830296e+00 -6.73999667e-01 6.25453055e-01 2.36806527e-01 -3.43294799e-01 8.53983164e-02 -7.54903138e-01 4.27540302e-01 1.48383290e-01 1.03929043e-01 -1.35306084e+00 -6.55273795e-01 -4.70910251e-01 2.84791023e-01 -5.87773100e-02 -1.16490948e+00 7.46343553e-01 -4.89315450e-01 -2.04412889e+00 1.09509814e+00 -2.77851045e-01 -7.99484849e-02 4.05532986e-01 3.81819159e-02 3.13962042e-01 2.96970643e-02 -3.06550503e-01 5.67246377e-01 5.58573902e-01 -9.62994039e-01 4.31671888e-02 -7.77098417e-01 -4.34742242e-01 -3.58177722e-01 2.10417241e-01 -2.08469674e-01 -1.90755501e-01 -2.17354205e-02 -1.02481872e-01 -9.10360694e-01 -1.35740012e-01 -9.12037343e-02 -4.43899989e-01 -9.42903012e-02 2.86845177e-01 -5.01330256e-01 5.93899071e-01 -2.33699775e+00 7.44314492e-01 -5.13562746e-02 3.98722529e-01 -3.71664613e-01 1.76750124e-01 6.84741795e-01 2.33487681e-01 1.78386047e-01 -5.00094414e-01 -3.47288758e-01 3.12658250e-02 4.05408829e-01 -7.95532949e-03 7.12771475e-01 1.55191213e-01 8.82955074e-01 -8.26631308e-01 -4.64668304e-01 1.13814764e-01 7.34045982e-01 -5.18462211e-02 2.23757550e-01 -6.03888392e-01 9.85734284e-01 -3.11737835e-01 6.23822510e-01 4.01893377e-01 -3.50340217e-01 6.98671639e-01 -1.06767274e-01 -5.79564422e-02 -2.98466116e-01 -7.87839532e-01 1.72118032e+00 -1.43482715e-01 3.80532444e-01 3.55517805e-01 -8.99016142e-01 6.40462101e-01 3.87656271e-01 5.17553091e-01 -9.24695954e-02 5.86830676e-01 2.03640945e-02 -8.68324786e-02 -2.21121803e-01 -3.33136261e-01 -7.16685474e-01 2.53241286e-02 2.13491052e-01 5.72562218e-01 -2.01426730e-01 4.61900860e-01 7.92990625e-02 1.13454938e+00 3.33351731e-01 4.32324976e-01 -5.35157800e-01 3.61128032e-01 -2.18379304e-01 8.19116890e-01 5.12799740e-01 -2.96989381e-01 4.75148886e-01 9.67205226e-01 -2.85768896e-01 -1.31326306e+00 -1.30612910e+00 -2.54916370e-01 7.26605892e-01 -4.04754579e-01 -2.12775037e-01 -8.84856582e-01 -9.58016887e-02 7.23485202e-02 2.21334010e-01 -9.30773914e-01 1.56809494e-01 8.22979435e-02 -1.27316940e+00 4.26106393e-01 2.70403117e-01 -1.24091394e-01 -6.23629868e-01 -4.55168039e-01 3.57470065e-01 8.53004232e-02 -9.20280933e-01 1.55387431e-01 6.00632906e-01 -1.21602082e+00 -1.09764922e+00 -6.69222116e-01 -5.56550659e-02 6.08370066e-01 -5.10642648e-01 8.76336336e-01 -2.26253510e-01 -4.73780751e-01 3.44334662e-01 1.38164386e-01 -5.18488228e-01 -4.14378434e-01 -2.53517359e-01 3.30942005e-01 2.18699247e-01 3.68236870e-01 -9.82752025e-01 -4.37975407e-01 -4.43851352e-01 -8.95152390e-01 -2.50308037e-01 4.67268616e-01 1.10236442e+00 1.31382656e+00 -2.98045397e-01 4.09561515e-01 -9.86084282e-01 3.74661773e-01 -7.40532994e-01 -7.91668236e-01 6.88870475e-02 -5.10545611e-01 1.82393879e-01 8.21697772e-01 -2.47560129e-01 -9.95735288e-01 1.78070918e-01 -2.58472562e-01 -4.31602031e-01 -4.99469817e-01 5.74790895e-01 -3.26606184e-01 4.26970214e-01 2.16864929e-01 4.97045130e-01 3.16067845e-01 -4.29390848e-01 2.90778995e-01 3.90236557e-01 1.74256235e-01 -4.46466595e-01 7.07181310e-03 9.33124125e-01 6.04710400e-01 -7.44358003e-01 -7.90514410e-01 -3.30927402e-01 -9.35072303e-01 -1.83562145e-01 1.00691080e+00 -1.03221083e+00 -9.94933784e-01 6.63700044e-01 -7.99663007e-01 -6.85925782e-01 -2.48796061e-01 4.84813184e-01 -1.05435038e+00 9.63905603e-02 -1.16104805e+00 -9.35035348e-01 -2.88258970e-01 -8.49877119e-01 1.42624843e+00 3.28805357e-01 -1.29303440e-01 -9.96495545e-01 8.60845149e-01 -6.86228648e-02 8.26456174e-02 5.03665745e-01 9.68166292e-01 -6.77343249e-01 -6.56697452e-01 -1.31490454e-01 1.88215837e-01 4.59215678e-02 5.42744100e-02 4.76476789e-01 -1.08311510e+00 -1.11350089e-01 1.33377444e-02 -1.06234975e-01 8.75499308e-01 8.28374445e-01 7.86486626e-01 2.37710103e-02 -4.36936051e-01 5.51994324e-01 1.85954559e+00 -2.70718392e-02 5.90236247e-01 -3.37083668e-01 5.07506669e-01 5.01151323e-01 -1.57965302e-01 6.70108020e-01 -3.16837668e-01 2.11048245e-01 2.25862116e-01 6.09500051e-01 2.18006983e-01 -8.03340599e-02 1.13474138e-01 9.09846246e-01 1.84110533e-02 -5.12047410e-01 -7.15003371e-01 4.40580249e-01 -1.85678232e+00 -8.65794837e-01 -2.02851355e-01 1.89907968e+00 6.73068523e-01 -3.60656619e-01 2.61314213e-02 -2.62045592e-01 6.29417896e-01 -4.22419235e-03 -6.16109908e-01 -1.54970750e-01 -3.30661774e-01 1.40602827e-01 1.55475155e-01 4.90881652e-01 -6.20299697e-01 5.15257716e-01 7.34896517e+00 6.89850688e-01 -9.93496478e-01 3.40026140e-01 6.91206455e-01 -3.37435305e-01 -1.60329103e-01 2.27434546e-01 -6.89930499e-01 8.00786853e-01 1.25486171e+00 -1.50164276e-01 4.00690585e-01 5.33703148e-01 3.80320817e-01 -6.58099711e-01 -1.14767551e+00 9.04114723e-01 -3.77075195e-01 -1.39218307e+00 -1.71916351e-01 5.06161869e-01 5.80079854e-01 3.83534342e-01 -2.75763899e-01 1.16859920e-01 6.24378502e-01 -7.35990763e-01 2.61497438e-01 1.26714313e+00 6.06849909e-01 -5.29163539e-01 1.08128428e+00 7.62156487e-01 -4.49614882e-01 2.54425615e-01 -4.01806861e-01 -1.73561335e-01 4.68006790e-01 1.34174335e+00 -6.38540208e-01 1.32726580e-01 4.11430776e-01 5.22145569e-01 -3.37437205e-02 5.57787180e-01 -4.23547737e-02 7.95128942e-01 -7.07349002e-01 -2.58344561e-01 -3.22788537e-01 -7.84974754e-01 5.15432537e-01 1.17026889e+00 4.74966317e-01 2.64124930e-01 -5.52144311e-02 1.31247771e+00 -2.75276810e-01 -2.95414850e-02 -6.08514309e-01 -5.84335446e-01 3.38788748e-01 1.69082010e+00 -9.73836362e-01 -3.85459572e-01 -2.58630902e-01 8.44050646e-01 5.81379712e-01 4.53789353e-01 -2.02343374e-01 2.44772092e-01 9.08248961e-01 7.82047305e-03 2.28689089e-01 -4.16917324e-01 -8.89807940e-02 -1.14442611e+00 -6.22330308e-01 -1.83194295e-01 2.24839717e-01 -8.01592767e-01 -1.58115041e+00 -2.47458279e-01 -2.33300641e-01 -5.25509059e-01 -1.43312365e-01 -9.73685682e-01 -4.85011786e-01 9.70243156e-01 -8.25251162e-01 -6.14116728e-01 -3.76866832e-02 -1.01032900e-03 1.93051994e-01 5.77955358e-02 1.06074154e+00 1.29009634e-02 -1.06810939e+00 -2.23815411e-01 7.66806245e-01 8.20442215e-02 3.03010941e-01 -1.39238834e+00 -2.84708053e-01 5.29404461e-01 1.16561837e-01 8.75417650e-01 8.61250997e-01 -7.82966137e-01 -1.50389731e+00 -8.23736429e-01 5.99423587e-01 -6.00900054e-01 5.71653485e-01 -4.22372818e-01 -7.64854670e-01 6.38584495e-01 7.43473992e-02 2.04676583e-01 1.23660481e+00 1.44280404e-01 6.31634220e-02 5.29253297e-02 -1.10768235e+00 3.57816190e-01 4.95673239e-01 -7.35334873e-01 -1.32778019e-01 2.46369332e-01 2.33075887e-01 -1.91104785e-01 -1.42063844e+00 6.53361157e-03 6.71314120e-01 -7.32233644e-01 4.35899228e-01 -6.79624140e-01 4.99098897e-01 -5.46026587e-01 -5.79449356e-01 -1.41965926e+00 -5.00917256e-01 -2.15175465e-01 -4.87128586e-01 1.11316669e+00 2.46117458e-01 -2.83561856e-01 7.64313459e-01 3.79758090e-01 2.24313706e-01 -9.00359154e-01 -1.25610542e+00 -4.94242400e-01 9.49371830e-02 1.44873247e-01 2.88398087e-01 7.22119391e-01 6.59112260e-02 4.32588220e-01 -1.01997107e-01 -1.72724038e-01 5.95721900e-01 -2.68619880e-03 6.49303973e-01 -1.49135232e+00 -4.31687891e-01 -1.50481716e-01 -2.56169945e-01 -6.48649573e-01 3.81729186e-01 -6.94505155e-01 3.40749115e-01 -1.38220000e+00 8.39949548e-01 4.81806129e-01 -3.00341040e-01 1.69787705e-02 -3.43016423e-02 6.19758032e-02 -2.07852036e-01 -1.13148773e-02 -4.99879122e-01 8.17962766e-01 7.91512430e-01 2.13822220e-02 1.40557095e-01 -6.22309864e-01 -3.85738254e-01 6.07227027e-01 6.38900995e-01 -6.74710095e-01 2.27776393e-01 -4.00802381e-02 5.32673061e-01 3.01125735e-01 2.53266245e-01 -6.98734939e-01 1.88535392e-01 -2.63726652e-01 7.54778743e-01 -5.59059918e-01 3.81625682e-01 -7.05715716e-01 6.11151338e-01 3.44807774e-01 -7.29037896e-02 -2.79514432e-01 2.76066847e-02 1.09682572e+00 -2.49232799e-01 -2.29365081e-01 8.07875156e-01 -5.91865778e-01 -1.61342740e-01 3.32494348e-01 -9.82945740e-01 -2.34019950e-01 5.88698566e-01 -6.45388588e-02 -2.01359063e-01 -5.84061034e-02 -1.16478181e+00 -1.22604623e-01 7.74745822e-01 -4.91834998e-01 1.86779320e-01 -1.08674943e+00 -5.89709818e-01 -2.14205742e-01 5.58354519e-02 4.84105609e-02 5.47527730e-01 1.08636689e+00 -4.42981482e-01 2.40202472e-01 -3.89038891e-01 -7.89099634e-01 -6.37028039e-01 7.53389239e-01 4.52328801e-01 -3.80386174e-01 -1.70184255e-01 6.86165392e-01 2.43664905e-01 -4.37568724e-01 -6.29538536e-01 1.60805538e-01 -1.26050249e-01 4.53790903e-01 2.48336837e-01 3.96809399e-01 -2.27454200e-01 -4.84229654e-01 -1.77173913e-01 4.29902375e-01 3.02012593e-01 -2.23842800e-01 1.37466359e+00 -3.46644729e-01 -7.24204421e-01 1.40035725e+00 1.25848246e+00 9.82773863e-03 -1.47694874e+00 1.60735041e-01 -2.69349545e-01 -2.04288572e-01 1.71459377e-01 -5.88059187e-01 -7.82873392e-01 7.27881968e-01 5.15303373e-01 -2.06155330e-01 7.78881729e-01 8.06243047e-02 4.72537220e-01 3.00451428e-01 5.35750449e-01 -1.18188620e+00 -1.78106308e-01 3.52085829e-01 5.08909464e-01 -1.02819324e+00 -6.73587024e-02 -1.56123847e-01 -7.50424787e-02 1.08748877e+00 1.67308003e-01 -1.54537961e-01 7.70080328e-01 5.40718675e-01 3.61067690e-02 -2.90395677e-01 -1.07048833e+00 2.57380418e-02 -4.55882251e-01 4.95954454e-01 6.90903664e-01 2.21406311e-01 -5.01311123e-01 7.56208241e-01 3.15486789e-01 4.45579290e-01 6.21877313e-01 6.36331081e-01 -3.66487414e-01 -9.51888680e-01 2.91322172e-02 6.63970292e-01 -7.68478990e-01 9.48235113e-03 -5.34778416e-01 2.99036413e-01 9.05436203e-02 6.56290531e-01 2.74920821e-01 1.46429211e-01 -1.57231644e-01 9.66533124e-01 7.30938852e-01 -6.75742745e-01 1.40148643e-02 4.26209390e-01 -2.75700718e-01 -5.00744581e-01 -4.68536735e-01 -1.08488595e+00 -1.14923775e+00 -2.07464159e-01 -1.84085280e-01 7.36142471e-02 6.54165506e-01 1.15213192e+00 5.51614165e-01 6.02473497e-01 2.81555176e-01 -7.43358850e-01 -1.14349775e-01 -1.14871001e+00 -1.30100596e+00 2.03452334e-01 2.43799388e-01 -6.27482235e-01 -5.98565102e-01 5.07661760e-01]
[6.688995361328125, 5.09923791885376]
b6d770b3-0a12-4a9d-9460-0e23fc6084ef
dstcgcn-learning-dynamic-spatial-temporal
2307.00518
null
https://arxiv.org/abs/2307.00518v1
https://arxiv.org/pdf/2307.00518v1.pdf
DSTCGCN: Learning Dynamic Spatial-Temporal Cross Dependencies for Traffic Forecasting
Traffic forecasting is essential to intelligent transportation systems, which is challenging due to the complicated spatial and temporal dependencies within a road network. Existing works usually learn spatial and temporal dependencies separately, ignoring the dependencies crossing spatial and temporal dimensions. In this paper, we propose DSTCGCN, a dynamic spatial-temporal cross graph convolution network to learn dynamic spatial and temporal dependencies jointly via graphs for traffic forecasting. Specifically, we introduce a fast Fourier transform (FFT) based attentive selector to choose relevant time steps for each time step based on time-varying traffic data. Given the selected time steps, we introduce a dynamic cross graph construction module, consisting of the spatial graph construction, temporal connection graph construction, and fusion modules, to learn dynamic spatial-temporal cross dependencies without pre-defined priors. Extensive experiments on six real-world datasets demonstrate that DSTCGCN achieves the state-of-the-art performance.
['Ling Chen', 'Binqing Wu']
2023-07-02
null
null
null
null
['graph-construction']
['graphs']
[-1.43842995e-01 -3.61264765e-01 -2.57546842e-01 -7.38382280e-01 -3.24609101e-01 -3.28938305e-01 6.29718542e-01 -5.05796432e-01 -1.76576942e-01 5.71188271e-01 9.86381769e-02 -9.55868006e-01 -5.88720083e-01 -1.12385547e+00 -6.65300012e-01 -7.49877274e-01 -5.34115851e-01 2.98380375e-01 7.36528397e-01 -2.19428450e-01 -3.78535688e-02 7.29600906e-01 -1.40802085e+00 1.19555667e-01 1.17449450e+00 9.57331359e-01 1.82443969e-02 7.56744742e-01 -2.63908207e-01 7.59723186e-01 -1.26604751e-01 -3.74268234e-01 3.20718624e-02 -1.53661877e-01 -4.77034688e-01 -5.42941280e-02 3.75154674e-01 4.64991778e-02 -1.02302229e+00 5.49210906e-01 1.54723167e-01 6.84152305e-01 3.85531902e-01 -1.72363353e+00 -6.06797993e-01 4.99534935e-01 -4.90759104e-01 6.37807548e-01 -2.70616084e-01 6.20262802e-01 7.15115249e-01 -2.54375637e-01 2.22892925e-01 1.48609960e+00 6.11623824e-01 2.76113451e-01 -1.09555590e+00 -9.99118268e-01 9.97024953e-01 5.93409538e-01 -1.40351796e+00 -2.86786407e-01 1.02862060e+00 -4.46347177e-01 1.02237201e+00 4.13777828e-02 6.28449678e-01 7.36770391e-01 3.95643234e-01 5.03374875e-01 6.74729884e-01 2.53102273e-01 -1.90912470e-01 -5.98747194e-01 3.54378462e-01 7.91243911e-01 -3.41477543e-01 4.56620485e-01 -2.27306008e-01 1.92493379e-01 6.54220343e-01 1.19639702e-01 8.08474049e-02 7.06777200e-02 -8.03229094e-01 6.56995237e-01 8.40624034e-01 8.73130187e-02 -3.78415376e-01 7.16134250e-01 3.60591203e-01 2.60609895e-01 6.74142897e-01 -5.41762531e-01 -4.45827663e-01 -1.32659122e-01 -5.87087393e-01 -1.15869157e-01 3.75032842e-01 1.06665254e+00 9.76101756e-01 9.21253785e-02 -3.80606890e-01 5.58948398e-01 2.25707054e-01 5.96970797e-01 -3.26753199e-01 -6.81953609e-01 9.17061448e-01 5.95683634e-01 -2.68257499e-01 -1.26920629e+00 -6.34397328e-01 -4.23148811e-01 -9.20709968e-01 -8.08607414e-03 4.26504791e-01 -3.21671516e-01 -1.28209662e+00 1.77399027e+00 4.14840519e-01 1.15943015e+00 -2.86877751e-01 8.32873046e-01 6.73526525e-01 8.31028581e-01 5.29258668e-01 8.60260427e-03 9.42682981e-01 -1.06328082e+00 -6.77904665e-01 -8.76141116e-02 7.69321620e-01 -3.60777676e-01 8.58876050e-01 -2.28120878e-01 -5.98803818e-01 -6.33176804e-01 -4.61688131e-01 -2.68161800e-02 -7.14863777e-01 5.60766011e-02 8.85759711e-01 3.58923733e-01 -9.86794472e-01 3.25117379e-01 -9.95650828e-01 -5.48267737e-02 4.64383811e-01 3.94607306e-01 -1.13084383e-01 -2.10035130e-01 -1.48893380e+00 6.86697185e-01 1.71145201e-01 6.16774201e-01 -5.74858129e-01 -7.56945848e-01 -9.09111321e-01 1.53322807e-02 5.83410919e-01 -4.78840798e-01 7.32347250e-01 -4.67992634e-01 -1.33213437e+00 1.77015215e-01 -5.00906169e-01 -5.89554965e-01 5.21874964e-01 8.35286453e-02 -1.29175520e+00 -1.24559827e-01 9.32682678e-02 2.71345407e-01 6.62805676e-01 -8.94948006e-01 -8.82866502e-01 -1.46282941e-01 1.01879455e-01 -2.22343847e-01 2.97900289e-01 -2.65401214e-01 -1.11092854e+00 -5.78279436e-01 -7.24838823e-02 -9.64300334e-01 -4.71755594e-01 -2.58210361e-01 -4.54455823e-01 -5.61384737e-01 1.28953695e+00 -6.91238046e-01 1.63437068e+00 -1.97234094e+00 -2.90163636e-01 5.91147244e-01 2.35159472e-01 3.00069869e-01 -3.19039732e-01 2.88496703e-01 -1.74071923e-01 -7.42475390e-02 -6.98217154e-02 -1.95501521e-02 4.27320600e-03 4.57488179e-01 -2.70377904e-01 3.63158554e-01 3.27092558e-01 1.12977207e+00 -1.13838625e+00 -3.41160953e-01 6.85667574e-01 8.51567328e-01 -2.62568444e-01 -1.35494143e-01 -3.14094514e-01 6.33174002e-01 -7.53058434e-01 1.37460694e-01 8.44526052e-01 -2.57887930e-01 -1.15218744e-01 -2.98751771e-01 -4.35227662e-01 5.03909528e-01 -8.13902497e-01 1.18778980e+00 -7.09769487e-01 9.66656625e-01 -3.42637748e-01 -9.65763688e-01 8.44251275e-01 1.14699014e-01 5.95048428e-01 -1.27487421e+00 5.46179339e-02 -1.66961461e-01 -8.69191885e-02 -6.71606183e-01 1.41784295e-01 2.68214941e-01 -1.67033095e-02 2.43771181e-01 -1.35253087e-01 4.27664071e-01 4.26105022e-01 3.39642823e-01 1.22461283e+00 -1.17872834e-01 -5.93301713e-01 -2.63168514e-02 6.87488139e-01 -1.08078979e-01 5.76720834e-01 3.64928037e-01 -1.26301840e-01 2.37587094e-01 7.91712165e-01 -7.99385786e-01 -3.83897781e-01 -9.87738669e-01 2.74924785e-01 1.13291407e+00 1.87775716e-01 -3.29474658e-01 -4.56810802e-01 -1.01766014e+00 1.13604695e-01 8.63215387e-01 -7.93275774e-01 -2.43842199e-01 -1.12655449e+00 -5.02360404e-01 2.60182530e-01 6.43039167e-01 5.99806666e-01 -7.62151182e-01 4.98773120e-02 3.80767822e-01 -3.43301035e-02 -1.24689436e+00 -1.05894089e+00 -2.10294470e-01 -5.00218630e-01 -1.16531408e+00 -1.79969043e-01 -5.98175049e-01 6.61439717e-01 6.66712761e-01 9.78087783e-01 2.16550946e-01 1.17358409e-01 1.56837344e-01 -3.56673487e-02 1.14557236e-01 1.41550168e-01 3.16878825e-01 -4.65568095e-01 4.41563010e-01 3.36723357e-01 -9.24305260e-01 -6.33789062e-01 6.49293184e-01 -5.30159235e-01 3.25590014e-01 3.04662436e-01 4.92809802e-01 5.21007061e-01 5.29429436e-01 5.16942263e-01 -9.08558905e-01 4.50943917e-01 -6.09129608e-01 -9.99920309e-01 3.49939972e-01 -4.52006519e-01 3.95074002e-02 6.20332718e-01 -4.80953872e-01 -1.06415582e+00 -1.25475287e-01 -6.59560263e-02 -4.60285097e-01 -6.04164042e-02 7.93553114e-01 -3.03668082e-01 -1.48739383e-01 1.55666992e-01 1.54212520e-01 -2.79960275e-01 -2.72404402e-01 7.03314126e-01 2.39162549e-01 4.92946804e-01 -4.81255174e-01 8.40312541e-01 5.21157622e-01 1.35610476e-01 -6.16294980e-01 -9.11372304e-01 -3.84367943e-01 -8.34730864e-01 -6.72141373e-01 1.00597012e+00 -7.19682515e-01 -1.03938246e+00 4.75400507e-01 -1.05368531e+00 -8.60716343e-01 1.88097671e-01 6.28630757e-01 -2.36614376e-01 -2.40029357e-02 -3.82804215e-01 -6.50620759e-01 1.80452466e-01 -9.07151341e-01 9.68805373e-01 3.85793269e-01 5.51791012e-01 -1.40232456e+00 -1.70425922e-01 1.69399276e-01 5.66791832e-01 3.27437490e-01 8.85453761e-01 -2.14485582e-02 -9.92531359e-01 -1.31315559e-01 -7.13216960e-01 6.38759583e-02 2.94333071e-01 2.79366076e-01 -4.77275491e-01 5.79725318e-02 -7.66855061e-01 6.03746414e-01 1.25725198e+00 7.26459444e-01 1.51697290e+00 -3.34360421e-01 -6.64350152e-01 8.24271441e-01 9.90570843e-01 4.49792325e-01 7.26737261e-01 -1.85985968e-01 1.25612402e+00 7.29990661e-01 4.48377520e-01 -1.77449803e-03 9.70604420e-01 5.92120290e-01 1.90235719e-01 -1.78799942e-01 -3.54513705e-01 -3.57831687e-01 1.76827461e-01 6.89308226e-01 -1.56578109e-01 -6.47491157e-01 -1.07526922e+00 6.28705561e-01 -2.21546340e+00 -1.13482594e+00 -5.82922518e-01 1.96444106e+00 2.35424891e-01 4.46171820e-01 2.57945210e-01 -3.53594422e-01 1.01786268e+00 4.04053688e-01 -7.40076423e-01 3.28488201e-02 5.62499054e-02 6.12647347e-02 9.35842037e-01 7.57507801e-01 -1.13633883e+00 1.30538356e+00 5.81709480e+00 9.20944810e-01 -1.41617477e+00 -4.49796990e-02 6.76994681e-01 8.12898725e-02 -3.78409475e-01 2.63581853e-02 -6.93387330e-01 8.14637005e-01 1.26524687e+00 -4.05563600e-02 5.96550345e-01 3.44060004e-01 5.79504371e-01 1.84634686e-01 -5.53040564e-01 6.83637679e-01 -4.69744951e-01 -1.40305686e+00 -3.99005152e-02 -3.92276645e-02 6.12708986e-01 3.78357351e-01 1.10879354e-01 4.70425308e-01 9.25655365e-01 -1.04026306e+00 2.27783158e-01 8.80369902e-01 5.81341326e-01 -1.06506598e+00 2.95495927e-01 -4.84734587e-02 -1.99779105e+00 8.63513816e-03 2.14550287e-01 1.83013916e-01 7.72282481e-01 7.68482506e-01 -3.66446495e-01 6.39205873e-01 4.18816328e-01 1.19342971e+00 -4.77335274e-01 1.07661974e+00 -3.56690526e-01 1.07965195e+00 -4.00474250e-01 1.69450298e-01 5.52753925e-01 -5.57890654e-01 3.06763828e-01 1.25134706e+00 2.67354518e-01 2.54078954e-01 2.40514740e-01 6.16593182e-01 2.41612673e-01 -4.90851313e-01 -4.51203346e-01 3.85322571e-02 5.18864930e-01 9.53467607e-01 -6.88127339e-01 -2.81055540e-01 -7.36361146e-01 5.07563949e-01 3.55624855e-01 1.06679749e+00 -1.46643817e+00 -4.97126848e-01 8.83234382e-01 2.06061900e-01 5.63542843e-01 -7.22138464e-01 -9.59080532e-02 -8.52268159e-01 9.71671641e-02 1.10299680e-02 7.14100063e-01 -6.41916990e-01 -1.36319423e+00 5.56639791e-01 1.60161749e-01 -1.13819051e+00 1.08394600e-01 -4.32431161e-01 -1.04899752e+00 9.94093657e-01 -1.78925657e+00 -1.51767206e+00 -3.62813681e-01 8.54787171e-01 2.44126454e-01 5.37397563e-02 -2.52381414e-02 6.44531846e-01 -9.16704535e-01 4.03266519e-01 -3.31041098e-01 3.35850269e-01 2.18406722e-01 -9.55396235e-01 1.01914537e+00 9.31134105e-01 -3.36633682e-01 3.54810774e-01 2.93023646e-01 -7.72768617e-01 -1.22042215e+00 -1.84425461e+00 1.01698279e+00 -2.57076323e-01 9.99700606e-01 -2.69644320e-01 -1.00201344e+00 8.77645969e-01 -6.35176748e-02 5.67554414e-01 2.65885055e-01 2.68476695e-01 -4.78147030e-01 -6.44347370e-01 -7.76971698e-01 7.31368661e-01 1.53835106e+00 -6.28547847e-01 1.93444595e-01 3.40156823e-01 1.04837155e+00 -3.76486152e-01 -5.63648820e-01 3.29477131e-01 4.40236986e-01 -4.46134567e-01 9.57283974e-01 -7.42580056e-01 -1.53408110e-01 -6.91562355e-01 2.27218404e-01 -1.30531716e+00 -5.71815073e-01 -6.85423732e-01 -1.49346530e-01 1.17726123e+00 5.73588133e-01 -1.12358367e+00 6.93393767e-01 7.86598682e-01 -5.35966158e-01 -6.13531590e-01 -1.12194204e+00 -8.53746176e-01 -1.87348813e-01 -9.75998163e-01 1.08375895e+00 8.68497610e-01 -5.34684062e-01 5.42914271e-01 -4.74376827e-01 5.46444893e-01 4.30408567e-01 2.42226362e-01 9.03320253e-01 -1.15386724e+00 1.63342163e-01 -6.50086582e-01 -5.10410666e-01 -1.31672764e+00 3.76258224e-01 -6.90072894e-01 1.35080088e-02 -1.60665178e+00 -4.02054697e-01 -7.29986966e-01 -4.21222627e-01 3.47465008e-01 -2.39772484e-01 -3.38343561e-01 -1.13952696e-01 -2.75516123e-01 -7.66097128e-01 7.80289471e-01 1.48833609e+00 -2.75972962e-01 -3.60954493e-01 3.65126133e-01 -3.37333441e-01 1.95599973e-01 8.02371860e-01 -3.50371689e-01 -9.39951897e-01 -6.27510786e-01 -6.26217127e-02 2.12651223e-01 4.20867682e-01 -1.04755867e+00 5.74002087e-01 -6.94668353e-01 -1.83432117e-01 -1.13043785e+00 1.75249174e-01 -8.59144449e-01 3.79489124e-01 6.78155720e-02 -2.13350594e-01 2.98022330e-01 3.23508382e-01 9.85672653e-01 -1.47510558e-01 8.14395368e-01 4.22952384e-01 3.41519326e-01 -9.68741000e-01 1.08921051e+00 -4.69948262e-01 -1.41547859e-01 9.85206962e-01 -1.21395988e-02 -5.67645609e-01 -3.89987230e-01 -8.42528939e-01 9.41013336e-01 -1.50623649e-01 6.24021232e-01 4.19114441e-01 -1.63365960e+00 -4.53559339e-01 1.92142710e-01 7.56586194e-02 2.98283156e-02 8.60456467e-01 1.07654095e+00 -1.81975007e-01 4.92446184e-01 1.35125086e-01 -6.57296836e-01 -9.48420346e-01 6.40706480e-01 4.24250782e-01 -3.01334858e-01 -7.24189579e-01 7.35798001e-01 3.03367645e-01 -5.03260195e-01 7.30302706e-02 -6.82672024e-01 -2.88228542e-01 -1.38045892e-01 3.57079268e-01 4.61770803e-01 -2.02921964e-02 -8.93655419e-01 -3.34814698e-01 6.26122415e-01 2.10846901e-01 1.07787624e-01 1.15139842e+00 -3.69344026e-01 2.95258556e-02 2.30190337e-01 1.30684996e+00 -4.30900127e-01 -1.70662069e+00 -4.66604501e-01 -6.14357814e-02 -5.14017820e-01 2.33676642e-01 -6.39041603e-01 -1.79037821e+00 9.47233260e-01 3.99023592e-01 3.69906902e-01 1.18089581e+00 -1.75396651e-01 1.25391996e+00 1.11941516e-01 1.43800065e-01 -9.37232018e-01 -3.22875887e-01 1.00409186e+00 6.58762038e-01 -1.09482408e+00 -4.23193663e-01 -7.84964085e-01 -4.82742727e-01 1.06808305e+00 8.61880124e-01 -1.87844157e-01 1.35451174e+00 5.45375682e-02 -1.46989495e-01 -5.01934886e-01 -1.10588586e+00 -7.72040546e-01 6.74944341e-01 6.83756948e-01 1.98759899e-01 3.76576662e-01 2.61837803e-03 1.48983076e-01 3.72205898e-02 -8.31823349e-02 -2.18883917e-01 5.32215238e-01 -1.29893824e-01 -9.19769704e-01 2.41204292e-01 5.21445334e-01 2.07075238e-01 6.84467927e-02 1.17492884e-01 8.88582647e-01 1.84788987e-01 1.09665215e+00 4.64694679e-01 -9.05679166e-01 5.59253216e-01 -3.18586618e-01 1.76491201e-01 -1.59854740e-01 -3.23752135e-01 -1.05866134e-01 3.75780821e-01 -8.12975049e-01 -3.78534675e-01 -5.97334087e-01 -1.48424172e+00 -7.33436882e-01 -1.23347320e-01 1.61193103e-01 4.29036468e-01 1.12204015e+00 8.24806690e-01 9.36451972e-01 8.06922615e-01 -7.77179182e-01 6.02974296e-01 -7.28384078e-01 -2.39620537e-01 2.90063918e-01 6.99657977e-01 -9.94153976e-01 -1.05153464e-01 1.05937840e-02]
[6.459216594696045, 2.068203926086426]
5f776c56-bac5-47ad-8154-e40dfb442827
learning-enhancement-from-degradation-a
2303.04603
null
https://arxiv.org/abs/2303.04603v1
https://arxiv.org/pdf/2303.04603v1.pdf
Learning Enhancement From Degradation: A Diffusion Model For Fundus Image Enhancement
The quality of a fundus image can be compromised by numerous factors, many of which are challenging to be appropriately and mathematically modeled. In this paper, we introduce a novel diffusion model based framework, named Learning Enhancement from Degradation (LED), for enhancing fundus images. Specifically, we first adopt a data-driven degradation framework to learn degradation mappings from unpaired high-quality to low-quality images. We then apply a conditional diffusion model to learn the inverse enhancement process in a paired manner. The proposed LED is able to output enhancement results that maintain clinically important features with better clarity. Moreover, in the inference phase, LED can be easily and effectively integrated with any existing fundus image enhancement framework. We evaluate the proposed LED on several downstream tasks with respect to various clinically-relevant metrics, successfully demonstrating its superiority over existing state-of-the-art methods both quantitatively and qualitatively. The source code is available at https://github.com/QtacierP/LED.
['Xiaoying Tang', 'Wenhan Luo', 'Huaqing He', 'Yijin Huang', 'Li Lin', 'Puijin Cheng']
2023-03-08
null
null
null
null
['image-enhancement']
['computer-vision']
[ 2.70030469e-01 -1.50698513e-01 3.58088762e-02 -3.75992000e-01 -7.98642635e-01 -4.96512443e-01 4.05493528e-01 -6.37694895e-02 -3.90093476e-01 7.53643751e-01 4.72127587e-01 -1.74316719e-01 -2.55736887e-01 -5.17126024e-01 -4.34717417e-01 -9.12774682e-01 5.56348152e-02 -3.74488682e-01 2.08374169e-02 1.21427394e-01 8.12232196e-02 2.53515422e-01 -1.30042887e+00 1.91313073e-01 1.49839246e+00 9.19940650e-01 2.29748502e-01 6.57034814e-01 5.97823799e-01 8.19706976e-01 -9.24917459e-02 -4.13463444e-01 3.38361502e-01 -6.83006227e-01 -6.09769583e-01 2.14014649e-01 2.74421155e-01 -7.38578081e-01 -4.70587373e-01 1.32939768e+00 8.03986907e-01 -1.99295375e-02 4.69909281e-01 -7.93603003e-01 -9.75917995e-01 7.31728785e-03 -6.28825963e-01 4.16976005e-01 -1.67776607e-02 6.15847707e-01 5.25578022e-01 -7.28964508e-01 6.19200230e-01 9.54244912e-01 2.41863444e-01 3.93413037e-01 -1.34017539e+00 -4.96628433e-01 -5.06531745e-02 1.41568437e-01 -1.02431059e+00 -5.26427567e-01 6.89991951e-01 -5.74502647e-01 4.21903312e-01 7.03214183e-02 6.92740798e-01 8.45954359e-01 4.74651247e-01 7.28420198e-01 1.63451374e+00 -1.99322000e-01 3.09970398e-02 -2.73442026e-02 -2.64597505e-01 6.91376984e-01 2.64467448e-01 4.41060692e-01 -2.36740038e-01 1.03532769e-01 8.02027047e-01 -6.69650128e-03 -6.62496328e-01 -1.91293195e-01 -1.11727059e+00 5.20191550e-01 7.56289780e-01 1.03851415e-01 -7.34424412e-01 1.47830293e-01 1.18904114e-01 2.40438834e-01 6.99482441e-01 3.04133177e-01 -9.59202722e-02 -6.51716813e-02 -8.15663815e-01 3.70673947e-02 2.02110425e-01 4.18685824e-01 4.99912530e-01 -2.66080350e-01 -6.06646240e-01 7.69328654e-01 4.83633250e-01 4.77302939e-01 3.12999964e-01 -1.42691469e+00 1.23942539e-01 4.15838033e-01 4.12416875e-01 -7.89945960e-01 -3.13065439e-01 -7.42684543e-01 -9.88070726e-01 5.72156370e-01 3.35648268e-01 -3.04148197e-01 -8.49230528e-01 1.78887141e+00 4.55946773e-01 3.33179504e-01 -7.76274875e-03 1.16835153e+00 7.38104403e-01 3.89217287e-01 2.03670055e-01 -3.64864588e-01 1.34712446e+00 -1.13454163e+00 -9.48541522e-01 -8.62812102e-02 4.10117805e-01 -7.60138631e-01 1.00719512e+00 4.76858199e-01 -1.42449617e+00 -4.74523485e-01 -8.82062495e-01 -4.19633538e-01 1.65806353e-01 5.25725067e-01 6.02787197e-01 4.00727719e-01 -1.27888572e+00 7.69624054e-01 -1.27079439e+00 -5.91103137e-02 8.01245809e-01 1.58949643e-01 -2.65070856e-01 -5.24846494e-01 -1.08117867e+00 9.28447843e-01 -5.04706688e-02 2.88612515e-01 -1.11450422e+00 -9.44424391e-01 -7.21228480e-01 -1.67018846e-01 1.53418193e-02 -1.33194077e+00 1.15606618e+00 -8.10956717e-01 -1.51150680e+00 7.78764963e-01 -3.07971746e-01 -3.11975211e-01 8.55655074e-01 -3.27853143e-01 -4.00685400e-01 5.14270723e-01 -3.69512080e-03 7.62679875e-01 9.03425753e-01 -1.22820616e+00 -4.75601971e-01 -4.55635875e-01 1.76180631e-01 2.73892224e-01 -3.54391932e-01 6.78704903e-02 -5.81426442e-01 -7.62541354e-01 -2.70416886e-01 -8.16047132e-01 -1.87953711e-01 8.10082793e-01 -1.99770108e-01 1.94496423e-01 2.34648392e-01 -1.07859862e+00 1.34714782e+00 -2.09600186e+00 2.94262618e-01 -1.31296337e-01 6.70256197e-01 4.31957126e-01 -2.34900549e-01 3.42374505e-03 1.28631383e-01 7.60191679e-02 -5.40297270e-01 -5.30610919e-01 -2.03753576e-01 -1.68732196e-01 1.71667293e-01 7.00737238e-01 4.61651236e-01 1.05982673e+00 -1.18406093e+00 -2.90443391e-01 2.90120065e-01 1.00195312e+00 -4.24868971e-01 2.71879345e-01 1.19927540e-01 9.05208170e-01 -3.72255921e-01 6.58896089e-01 7.83453584e-01 -4.44207847e-01 4.43044640e-02 -4.21422869e-01 -1.61393598e-01 -7.51464888e-02 -6.55568421e-01 1.85936046e+00 -4.28517312e-01 6.60599709e-01 -7.60645792e-02 -5.07470727e-01 5.96883893e-01 1.99083671e-01 4.25225735e-01 -7.62188375e-01 2.75281399e-01 2.42579937e-01 1.55362502e-01 -6.81043088e-01 2.82593928e-02 -2.53066003e-01 6.58878148e-01 4.03606534e-01 -7.21289916e-03 1.32554948e-01 1.93583176e-01 -1.43246846e-02 9.38906372e-01 2.94204712e-01 2.19702840e-01 -2.65717432e-02 4.51740086e-01 -3.69958252e-01 4.75148588e-01 4.41254467e-01 -5.89867651e-01 7.38635421e-01 5.81518590e-01 1.74798429e-01 -9.63447809e-01 -1.14927268e+00 -3.26508343e-01 3.17156851e-01 2.82407999e-01 -4.79584709e-02 -6.82114780e-01 -4.40189898e-01 -9.46887732e-02 3.07428062e-01 -7.42933810e-01 -3.51425171e-01 1.85838994e-02 -1.01765537e+00 3.15872461e-01 3.93660873e-01 6.48474634e-01 -7.14023650e-01 -4.57200944e-01 9.86868516e-02 -2.73738563e-01 -8.60046208e-01 -6.01112723e-01 -3.67674083e-01 -1.08923507e+00 -9.48529959e-01 -1.33093464e+00 -6.13544405e-01 8.44853580e-01 3.26966077e-01 8.09962928e-01 1.08475983e-01 -3.88695389e-01 1.66731551e-01 -2.42013320e-01 -2.51061410e-01 -3.60891104e-01 -4.32440311e-01 -9.80478749e-02 3.78538162e-01 7.09598977e-03 -6.55060589e-01 -1.52254343e+00 4.97943573e-02 -1.15812719e+00 1.34230733e-01 9.62887228e-01 8.29599082e-01 8.26749504e-01 1.65316880e-01 4.97200191e-01 -6.26494050e-01 7.45703161e-01 -4.35671479e-01 -5.33441246e-01 1.76120445e-01 -8.92060101e-01 -1.16549887e-01 1.85403973e-01 -3.02466869e-01 -1.26064622e+00 -1.93316236e-01 -3.79606374e-02 -5.09641409e-01 -5.21603674e-02 4.82431442e-01 1.67352147e-02 -3.36148798e-01 6.38263047e-01 1.10683613e-01 2.85997301e-01 -5.57429850e-01 4.50303435e-01 5.78241885e-01 6.41978979e-01 -1.77668914e-01 5.77148139e-01 8.68360281e-01 -1.02418102e-01 -3.32273781e-01 -8.77811491e-01 -3.98747057e-01 -4.26378191e-01 -4.06737745e-01 7.84804702e-01 -1.15603077e+00 -4.60381895e-01 9.74784017e-01 -9.06762838e-01 -6.34062827e-01 -1.38864517e-01 7.16424167e-01 -4.92512167e-01 5.75858176e-01 -8.22394788e-01 -5.93907773e-01 -4.22201514e-01 -1.19748092e+00 8.65954459e-01 5.65994203e-01 1.89197809e-01 -1.29458368e+00 1.06461801e-01 3.75283450e-01 5.65621674e-01 4.89203095e-01 8.02441001e-01 2.97094196e-01 -5.25016189e-01 7.52128959e-02 -5.84266067e-01 8.72652352e-01 4.10547942e-01 -1.05943933e-01 -7.98320949e-01 -5.10414481e-01 7.90613666e-02 -1.90971300e-01 1.06474471e+00 8.85367215e-01 1.03011882e+00 -1.53618887e-01 -1.41740024e-01 9.16268885e-01 1.56631982e+00 -9.31794643e-02 1.05509543e+00 2.40653515e-01 3.05081487e-01 5.39702535e-01 6.51935637e-01 4.79470581e-01 3.85207832e-01 4.86520857e-01 3.22097242e-01 -7.32285917e-01 -6.90671802e-01 -1.41804442e-01 2.49539524e-01 4.94532257e-01 -3.07529598e-01 -2.57172555e-01 -6.93836868e-01 6.32223248e-01 -1.68243289e+00 -5.85484266e-01 -2.59046733e-01 2.06974554e+00 1.05870235e+00 -1.01572126e-01 -9.93672386e-02 -2.99653023e-01 6.40545547e-01 -1.14483591e-02 -8.94628644e-01 2.32872263e-01 -1.65352911e-01 6.10915348e-02 2.48594567e-01 6.16724789e-01 -1.06606352e+00 5.74436367e-01 5.98375034e+00 3.72404188e-01 -1.08086061e+00 3.78788650e-01 8.55610073e-01 -2.57572949e-01 -3.85832578e-01 -4.15787883e-02 -1.97615877e-01 5.99229395e-01 7.57263362e-01 -3.20700943e-01 4.67300177e-01 1.97975159e-01 8.58227193e-01 -2.10748121e-01 -6.86830938e-01 9.55484390e-01 -1.71790957e-01 -1.10910141e+00 -1.39882654e-01 1.21041991e-01 9.50511217e-01 5.43045886e-02 5.09379089e-01 -2.02397183e-01 -3.06789409e-02 -8.62931848e-01 2.82251388e-01 1.11531949e+00 1.05852032e+00 -3.70113641e-01 7.60456145e-01 -1.55794859e-01 -6.22314632e-01 -1.93720926e-02 -2.02829570e-01 1.69145167e-01 3.57499331e-01 9.28504169e-01 -1.28514603e-01 6.31963491e-01 7.62558579e-01 1.16343796e+00 -6.99158013e-01 1.64067030e+00 -5.61502993e-01 5.20494342e-01 1.51260763e-01 5.43621540e-01 -2.38629319e-02 -3.56128067e-01 5.63000619e-01 9.61644888e-01 5.27047336e-01 2.15492040e-01 -3.87802809e-01 1.15307426e+00 -1.02568656e-01 -1.05180174e-01 -2.49180809e-01 9.13072005e-02 7.86746889e-02 1.38256037e+00 -3.76767695e-01 -2.60167122e-01 -4.95450497e-01 1.15453589e+00 5.44782355e-02 6.87944591e-01 -7.72084415e-01 -3.91640604e-01 6.57013416e-01 5.91386743e-02 8.18297267e-02 1.58215538e-02 -6.59192875e-02 -1.39116657e+00 2.73957461e-01 -7.84485161e-01 1.48391649e-01 -1.17015588e+00 -1.37794793e+00 6.35051012e-01 -2.63992071e-01 -1.55825245e+00 2.93314397e-01 -4.91165042e-01 -4.80883658e-01 1.08904994e+00 -2.25798035e+00 -1.03974378e+00 -5.56830645e-01 5.83460033e-01 1.87037680e-02 1.84168398e-01 4.19072002e-01 5.59788108e-01 -6.02210522e-01 3.23318392e-01 3.40418726e-01 -8.19435865e-02 1.04225087e+00 -1.24187219e+00 -7.67432228e-02 1.14169145e+00 -3.91899914e-01 5.60360074e-01 5.54808736e-01 -5.34913361e-01 -9.57117498e-01 -1.32061338e+00 6.54499412e-01 -2.84797430e-01 7.05073714e-01 2.12645561e-01 -9.31708813e-01 3.90480727e-01 3.22287947e-01 3.48779261e-01 6.27272964e-01 -4.44721371e-01 -6.68607280e-02 -3.41808349e-01 -1.05173361e+00 5.48627198e-01 9.78903890e-01 -4.18913871e-01 -2.91402012e-01 3.65389675e-01 5.41013539e-01 -4.90007102e-01 -1.24155807e+00 3.18124413e-01 5.32407224e-01 -9.91524339e-01 7.79680550e-01 -4.77790952e-01 7.74394333e-01 -5.00811338e-01 2.15765938e-01 -1.50187933e+00 -4.24484938e-01 -7.37209976e-01 -3.53618473e-01 1.09409010e+00 3.31666380e-01 -8.67098272e-01 1.94428191e-01 4.99160677e-01 -1.25214338e-01 -8.45130146e-01 -6.76074266e-01 -6.68240964e-01 1.18595317e-01 -1.19083233e-01 2.05691755e-01 6.35430634e-01 -2.88592815e-01 -6.39922991e-02 -4.43372697e-01 3.65795702e-01 7.61670172e-01 1.11139432e-01 1.95811197e-01 -8.37314487e-01 -4.10585195e-01 -5.16006947e-01 -3.62248421e-01 -9.67949748e-01 -1.84395775e-01 -9.13619399e-01 -1.17109567e-01 -1.81509304e+00 4.00264919e-01 -2.82028943e-01 -6.81140780e-01 4.65259463e-01 -5.13154030e-01 5.10957301e-01 2.68630292e-02 3.88056099e-01 -5.04094064e-01 7.82342136e-01 1.85278463e+00 -1.08727351e-01 -2.12276176e-01 -1.02942884e-01 -1.09330428e+00 3.50429922e-01 8.23109925e-01 -4.06924695e-01 -4.69016403e-01 -5.94784558e-01 2.90156491e-02 1.26269788e-01 6.48431957e-01 -9.18733358e-01 6.80955425e-02 1.51519701e-01 3.80648375e-01 1.86327130e-01 1.10267930e-01 -3.90342206e-01 3.17459479e-02 5.98399699e-01 -3.93789113e-01 -3.15072119e-01 2.45210737e-01 6.02401018e-01 -4.13823873e-01 1.02232791e-01 1.05391395e+00 1.81188181e-01 -3.57514828e-01 6.57329500e-01 -1.15211566e-04 -1.23170195e-02 9.27793920e-01 2.21452907e-01 -5.90284586e-01 -4.65058774e-01 -8.09240043e-01 3.24983120e-01 4.99868363e-01 3.26131195e-01 7.43267655e-01 -1.28219926e+00 -1.00621676e+00 -1.69606552e-01 1.51803955e-01 -3.29682678e-01 8.52737129e-01 1.73777997e+00 -4.18161809e-01 1.81877643e-01 -2.87760466e-01 -4.98025000e-01 -1.00247693e+00 6.51421607e-01 7.35846996e-01 -2.15572670e-01 -6.18035495e-01 6.03831053e-01 3.15294832e-01 3.54769006e-02 -1.30423401e-02 -4.36372459e-01 -1.08968236e-01 -2.98337013e-01 8.18779826e-01 3.22565407e-01 -6.51102066e-02 -3.08117181e-01 -1.17265277e-01 6.55456841e-01 -1.15162684e-02 -3.07731163e-02 1.31617844e+00 -6.28295898e-01 -1.68984339e-01 3.63916382e-02 1.11613798e+00 -2.17075273e-01 -1.86162961e+00 -3.08150619e-01 -4.34401363e-01 -7.22230256e-01 6.76675439e-01 -1.31492150e+00 -1.36452389e+00 9.03088510e-01 1.21000397e+00 -1.70023337e-01 1.67359483e+00 -1.68125033e-01 6.97829604e-01 -3.27028573e-01 -8.36418495e-02 -6.68663859e-01 8.61904994e-02 -3.03536147e-01 1.07558191e+00 -1.41044831e+00 -6.12400286e-03 -4.18663681e-01 -7.53309131e-01 8.48633528e-01 2.96527892e-01 -8.25472102e-02 6.60373271e-01 3.76398526e-02 3.49086493e-01 -7.06377774e-02 -7.58596003e-01 -3.24343592e-01 5.42112470e-01 6.38367295e-01 4.27644759e-01 -1.47959813e-01 -5.07293224e-01 4.45952982e-01 4.16211665e-01 6.08627617e-01 6.18290782e-01 5.42205453e-01 -1.08213678e-01 -1.08167589e+00 7.30590224e-02 4.55259770e-01 -5.31724155e-01 -2.78202057e-01 1.21850938e-01 4.58212197e-01 8.37724060e-02 1.04677558e+00 -2.03022197e-01 -9.86754894e-02 2.97785074e-01 -4.10242647e-01 4.87533718e-01 -3.35623682e-01 -1.33162335e-01 3.56499553e-01 -1.45371944e-01 -7.02413261e-01 -8.97531509e-01 -7.47840166e-01 -8.93212974e-01 -1.22893788e-01 3.47968913e-03 -3.55960876e-01 3.23978841e-01 6.62338734e-01 7.60561943e-01 6.82383776e-01 7.34448373e-01 -6.08700216e-01 -3.70934308e-01 -8.07372570e-01 -7.40032792e-01 4.21209544e-01 6.78884327e-01 -6.69335008e-01 -4.65493649e-01 2.92527139e-01]
[15.70450210571289, -3.8908276557922363]
affaa7db-12bf-4421-beb4-2be5d8b34069
a-deep-face-identification-network-enhanced
1805.00324
null
http://arxiv.org/abs/1805.00324v1
http://arxiv.org/pdf/1805.00324v1.pdf
A Deep Face Identification Network Enhanced by Facial Attributes Prediction
In this paper, we propose a new deep framework which predicts facial attributes and leverage it as a soft modality to improve face identification performance. Our model is an end to end framework which consists of a convolutional neural network (CNN) whose output is fanned out into two separate branches; the first branch predicts facial attributes while the second branch identifies face images. Contrary to the existing multi-task methods which only use a shared CNN feature space to train these two tasks jointly, we fuse the predicted attributes with the features from the face modality in order to improve the face identification performance. Experimental results show that our model brings benefits to both face identification as well as facial attribute prediction performance, especially in the case of identity facial attributes such as gender prediction. We tested our model on two standard datasets annotated by identities and face attributes. Experimental results indicate that the proposed model outperforms most of the current existing face identification and attribute prediction methods.
['Nasser M. Nasrabadi', 'Jeremy Dawson', 'Fariborz Taherkhani']
2018-04-20
null
null
null
null
['gender-prediction']
['computer-vision']
[ 1.85302824e-01 1.23272650e-01 -2.87813753e-01 -1.09272552e+00 -3.05534601e-01 -4.27899778e-01 6.99821651e-01 -4.16010112e-01 -2.01664820e-01 4.41295117e-01 8.78949165e-02 1.13386869e-01 2.94067971e-02 -7.12585807e-01 -5.20281434e-01 -7.24295676e-01 3.33589464e-01 5.74093759e-01 -3.51932853e-01 6.45806715e-02 5.67745864e-02 5.35266340e-01 -1.96054506e+00 4.86729234e-01 6.03324175e-01 1.62182999e+00 -5.52130997e-01 1.73484892e-01 -1.98503621e-02 9.14462864e-01 7.19785038e-03 -9.04805899e-01 4.44219261e-01 -9.08080954e-03 -8.72545302e-01 -8.32820758e-02 6.18834794e-01 -7.12839305e-01 -1.15420841e-01 9.92795765e-01 6.33061767e-01 -2.58545041e-01 5.49865127e-01 -1.60351145e+00 -7.09605694e-01 4.32925552e-01 -5.99718034e-01 -2.35923991e-01 2.73331314e-01 1.44686699e-02 7.28658736e-01 -1.17663527e+00 2.68892258e-01 1.54767692e+00 1.02627897e+00 8.50119650e-01 -9.37618315e-01 -1.24866092e+00 -9.05730296e-03 1.97713643e-01 -1.57910359e+00 -1.04743946e+00 5.89020848e-01 -4.86615211e-01 5.13814151e-01 -4.52688783e-02 1.08375646e-01 1.26343572e+00 -4.76199716e-01 8.16008270e-01 1.01117361e+00 -1.89648792e-01 -2.19629332e-01 2.75434107e-01 -4.62541059e-02 8.33610594e-01 -1.84039533e-01 2.23375484e-01 -6.19656205e-01 -2.91021258e-01 4.54646081e-01 -2.74181692e-03 1.67457312e-01 -3.30046900e-02 -9.44345117e-01 8.53080332e-01 2.10755125e-01 -1.15229003e-01 -4.71809894e-01 -4.53323536e-02 4.68428224e-01 1.82238773e-01 7.00258493e-01 -2.04364240e-01 -3.74157697e-01 1.12633049e-01 -8.64062428e-01 8.54243636e-02 6.88039720e-01 7.73207545e-01 9.99581814e-01 8.56534287e-04 -3.72429967e-01 1.17490876e+00 4.92913872e-01 3.69227499e-01 3.19588274e-01 -7.79083252e-01 1.17244266e-01 8.54089975e-01 -2.07598150e-01 -8.17872643e-01 -1.79084241e-01 -1.87929317e-01 -8.23342025e-01 1.52442619e-01 2.44057447e-01 -1.69330135e-01 -8.10020089e-01 1.83017194e+00 3.60663950e-01 5.86953282e-01 -1.53198525e-01 7.32146740e-01 1.24773204e+00 2.24081203e-01 5.92676163e-01 1.83053002e-01 1.32882345e+00 -8.52768540e-01 -7.94463217e-01 3.08312699e-02 3.91926467e-01 -9.39381659e-01 6.35637283e-01 -4.23345938e-02 -9.82856095e-01 -9.45143342e-01 -6.92760527e-01 2.83269156e-02 -4.23349053e-01 6.65374577e-01 7.33885407e-01 9.64431822e-01 -1.27829194e+00 3.83407593e-01 -3.50850195e-01 -3.61256361e-01 1.06075954e+00 8.87643754e-01 -7.51834214e-01 8.55672657e-02 -1.07239985e+00 7.27917373e-01 3.49094152e-01 2.13569686e-01 -9.97500122e-01 -7.37452984e-01 -8.91336858e-01 2.21052215e-01 2.14492589e-01 -6.13291919e-01 1.18846416e+00 -1.42999065e+00 -1.63218784e+00 1.30701697e+00 -4.34679210e-01 -1.27078280e-01 3.68252039e-01 -2.92492241e-01 -4.42321002e-01 -1.26818314e-01 3.75609100e-02 1.03226006e+00 9.16773438e-01 -1.15367126e+00 -6.96967065e-01 -6.41167164e-01 -1.01485483e-01 -1.47912968e-02 -1.00532913e+00 4.73829925e-01 -3.33457440e-01 -3.94991994e-01 -2.92986155e-01 -9.09355819e-01 2.96799153e-01 1.43901661e-01 -4.65655297e-01 -5.65754116e-01 1.12944400e+00 -6.28761232e-01 9.66020226e-01 -2.06450462e+00 -1.76011026e-02 2.92288512e-01 1.88571438e-01 4.13083851e-01 -2.19409347e-01 1.20718293e-01 -3.64041775e-01 8.82198215e-02 1.46012148e-02 -9.20651913e-01 8.33294094e-02 1.33389995e-01 -5.51105738e-02 3.31485361e-01 4.65525657e-01 8.63905013e-01 -3.77793849e-01 -5.99049568e-01 2.99062021e-02 7.32728899e-01 -4.65241402e-01 4.37940687e-01 -2.61819223e-03 4.95795786e-01 -4.56741422e-01 1.14761448e+00 1.09638977e+00 -5.41044101e-02 1.44258961e-01 -4.10580426e-01 8.40190351e-02 -2.49396160e-01 -1.05055535e+00 1.31024003e+00 -3.52527857e-01 1.47524506e-01 2.85422444e-01 -8.04239511e-01 1.22534537e+00 5.14485300e-01 7.04360366e-01 -3.39688003e-01 3.92674834e-01 2.40665928e-01 -1.00458667e-01 -5.13836801e-01 1.97064191e-01 -9.26664378e-03 2.23951295e-01 4.18920517e-01 4.70445216e-01 6.66137874e-01 -1.88262999e-01 -1.51380897e-01 4.63319331e-01 4.25428361e-01 1.85212925e-01 -1.92262903e-01 1.17514300e+00 -5.89015901e-01 8.33259106e-01 2.24878937e-01 -3.61663908e-01 4.43107307e-01 3.94915670e-01 -9.32768464e-01 -9.87232566e-01 -6.49336040e-01 -1.87359959e-01 1.68605077e+00 -1.99089542e-01 -4.00870293e-01 -9.25377071e-01 -1.06823063e+00 2.84346908e-01 5.53234108e-02 -9.16268051e-01 -1.48684040e-01 -3.79859507e-01 -7.36437082e-01 8.32339644e-01 7.62337089e-01 7.65163660e-01 -1.12420964e+00 1.41076818e-01 -1.98174521e-01 -2.13833928e-01 -1.43237758e+00 -3.90872985e-01 -3.00361395e-01 -3.04788888e-01 -1.11425090e+00 -3.57968658e-01 -7.88781345e-01 6.09124005e-01 -2.37742230e-01 1.00550044e+00 3.93606395e-01 -4.50989790e-02 3.58764887e-01 -2.95822799e-01 -6.85968399e-01 -3.67129862e-01 2.15737388e-01 2.59972751e-01 1.07307458e+00 8.32656384e-01 -4.83908683e-01 -5.89124084e-01 4.35129732e-01 -6.41325235e-01 -1.69384390e-01 7.15774953e-01 8.68625164e-01 3.45043510e-01 -4.37838048e-01 8.51841450e-01 -9.69884932e-01 2.75029510e-01 -7.34476268e-01 -3.14261258e-01 3.51475805e-01 -5.35265863e-01 -1.43279761e-01 4.91367847e-01 -2.59248823e-01 -1.18035007e+00 7.24156260e-01 -4.27948624e-01 -4.55638498e-01 -4.26497489e-01 2.30866626e-01 -4.59746182e-01 -4.50832993e-01 1.11239269e-01 1.69802293e-01 5.06163538e-01 -5.71738303e-01 -2.09898339e-03 1.05797613e+00 6.10940933e-01 -6.72064900e-01 6.89826250e-01 3.63080561e-01 2.02996746e-01 -4.85718489e-01 -8.30736935e-01 -3.94068420e-01 -1.00667238e+00 -4.52313811e-01 9.28257823e-01 -1.11798906e+00 -1.46125710e+00 7.71571755e-01 -1.06304836e+00 3.03069860e-01 3.30058813e-01 2.39588097e-01 -5.50217748e-01 1.49737507e-01 -5.69928467e-01 -8.78270745e-01 -6.02847934e-01 -1.26115000e+00 1.35454714e+00 3.80369276e-01 -4.30683121e-02 -8.79858255e-01 -3.04935127e-01 5.95957398e-01 7.21266389e-01 1.74117818e-01 6.82166100e-01 -1.14473009e+00 -2.72536606e-01 -1.47786424e-01 -4.97758448e-01 2.37627387e-01 1.41265780e-01 8.99413154e-02 -1.59986198e+00 -1.66105226e-01 -3.44117850e-01 -8.25673819e-01 8.13395619e-01 9.93104577e-02 1.69792104e+00 -2.21052080e-01 -3.12910289e-01 1.04081416e+00 1.24963033e+00 -9.35787894e-03 7.51178384e-01 2.56832749e-01 8.50454092e-01 7.81953633e-01 3.67736369e-01 6.50010049e-01 6.19696259e-01 7.85426736e-01 6.46532536e-01 -2.78387696e-01 1.92888081e-02 -1.55951247e-01 1.22715391e-01 2.57900059e-01 -3.70665103e-01 5.31868935e-02 -1.01621819e+00 2.72301227e-01 -1.85576534e+00 -1.02223301e+00 2.25411698e-01 1.95570171e+00 8.07148635e-01 -5.01404107e-01 4.46488678e-01 -1.00895032e-01 9.30842042e-01 -5.72100878e-02 -4.98728096e-01 -3.93730044e-01 -9.09212753e-02 4.34084028e-01 2.23352551e-01 2.77947068e-01 -1.57477343e+00 1.03928721e+00 6.26022625e+00 6.33202195e-01 -1.04980552e+00 8.14307332e-02 1.11219287e+00 8.06984666e-04 7.82764181e-02 -3.16200227e-01 -1.18678951e+00 5.28123140e-01 8.91613364e-01 1.43301636e-01 1.34149909e-01 9.05427575e-01 -2.05853090e-01 4.96154696e-01 -1.23222911e+00 1.05856550e+00 2.02530712e-01 -9.81609523e-01 3.26615244e-01 1.79058328e-01 4.14448828e-01 -4.87936199e-01 3.79525483e-01 3.43622983e-01 2.56538510e-01 -1.62439001e+00 4.82370704e-01 5.87238908e-01 9.98582184e-01 -8.10073078e-01 1.03466403e+00 7.92085528e-02 -1.28888905e+00 -4.02907997e-01 -1.39235169e-01 2.34923605e-02 -2.86154211e-01 1.23053782e-01 -8.79528046e-01 6.09754920e-01 7.86614537e-01 8.63106549e-01 -6.58279061e-01 7.52858460e-01 1.57991216e-01 4.25086766e-01 4.46467921e-02 4.18749809e-01 1.25143886e-01 -1.72941342e-01 1.34583833e-02 1.02210355e+00 4.11716849e-01 1.27389714e-01 3.72626424e-01 7.57430196e-01 -4.54499125e-01 1.97752848e-01 -4.89534557e-01 2.33723801e-02 7.08638906e-01 1.72221363e+00 -1.75645545e-01 -2.25096077e-01 -7.65851736e-01 7.20940113e-01 3.45152408e-01 6.51943982e-02 -8.07470918e-01 -4.40101791e-03 9.55896616e-01 -9.02533531e-02 2.20894530e-01 4.23873395e-01 -3.18864822e-01 -8.06213200e-01 -5.90913594e-02 -8.70447218e-01 5.72822213e-01 -3.27220142e-01 -1.48183012e+00 7.77510464e-01 -2.66510516e-01 -1.13503253e+00 -3.37565720e-01 -8.05189729e-01 -5.43016195e-01 1.16495776e+00 -1.83382428e+00 -2.11501932e+00 -6.04224026e-01 8.54485035e-01 2.91825801e-01 -8.12739670e-01 1.15536726e+00 6.31608188e-01 -8.35044622e-01 1.23472440e+00 -1.91854209e-01 5.41913807e-01 1.01712465e+00 -8.82854879e-01 1.03728153e-01 3.64990771e-01 -2.57854432e-01 6.27819598e-01 6.86487108e-02 -4.98871922e-01 -1.15514743e+00 -1.17335737e+00 9.26258922e-01 -3.70834768e-01 3.09383482e-01 -2.66711861e-01 -8.20299983e-01 7.86329508e-01 2.53792673e-01 4.46302116e-01 1.21412873e+00 2.49274939e-01 -7.20096588e-01 -3.46129835e-01 -1.33236933e+00 -7.21766353e-02 1.00832582e+00 -6.41442120e-01 2.07229909e-02 1.30874127e-01 -2.45172977e-02 -1.28706202e-01 -1.14947176e+00 8.57380927e-01 1.02083802e+00 -8.68294418e-01 1.17993641e+00 -7.94389546e-01 6.63986146e-01 4.33213487e-02 -1.91243291e-01 -1.04958642e+00 -4.09745723e-01 -2.95086354e-01 -5.83013147e-02 1.66401041e+00 2.52088159e-01 -5.41724741e-01 9.34245229e-01 6.79071009e-01 6.93216920e-03 -9.13308680e-01 -7.43965328e-01 -5.11352777e-01 -7.81113729e-02 2.36321121e-01 1.20711970e+00 1.22020185e+00 -6.20987892e-01 2.76871175e-01 -6.46854877e-01 8.30102488e-02 7.15446234e-01 7.08331987e-02 5.73731124e-01 -1.73008144e+00 1.99387059e-01 -5.32641351e-01 -4.84159350e-01 -3.25357050e-01 9.14233685e-01 -1.02207494e+00 -3.04260135e-01 -8.09966028e-01 6.98471069e-01 -5.39053440e-01 -4.35165733e-01 9.53542292e-01 -2.99710840e-01 6.69794917e-01 1.10995874e-01 1.12049535e-01 -4.37557727e-01 6.52829528e-01 8.26662958e-01 -9.78325531e-02 1.86130255e-01 1.74933761e-01 -8.11104655e-01 8.47337663e-01 7.17663407e-01 -1.05303556e-01 -2.58266240e-01 -4.76603925e-01 -1.77735269e-01 -1.15955912e-01 4.51408476e-01 -8.80246878e-01 2.93387920e-01 4.43123952e-02 9.83900011e-01 -4.28279221e-01 5.19294322e-01 -8.78177822e-01 -3.97771373e-02 1.40277818e-01 -2.86491722e-01 9.94183272e-02 2.45707020e-01 1.52046666e-01 -4.57268804e-01 1.84258103e-01 9.97603714e-01 1.08513005e-01 -8.72067273e-01 9.43484902e-01 1.56862274e-01 -5.90587974e-01 1.26536262e+00 -3.03032249e-01 -4.17784482e-01 -3.71573895e-01 -7.76683748e-01 3.49692315e-01 1.29150182e-01 6.48492813e-01 6.55511677e-01 -1.79231358e+00 -1.09555519e+00 5.17115772e-01 2.52013087e-01 -6.64987922e-01 1.53750643e-01 7.20045507e-01 6.84082229e-03 2.47510746e-01 -7.70572484e-01 -5.91361880e-01 -1.77441025e+00 5.01179516e-01 4.51061964e-01 1.09441891e-01 1.44573063e-01 1.04217839e+00 2.46744007e-01 -8.03418338e-01 4.24815834e-01 7.71829665e-01 -8.00769389e-01 2.72033304e-01 6.58737659e-01 2.41762578e-01 -4.95651662e-02 -1.34934223e+00 -5.85588753e-01 7.12932110e-01 -2.41464928e-01 3.64760846e-01 1.44769359e+00 -6.30128309e-02 -5.15241444e-01 -1.36662915e-01 1.36621594e+00 -3.08966845e-01 -1.08946538e+00 -4.51138258e-01 -4.48953286e-02 -6.89373791e-01 -1.02434210e-01 -9.67656195e-01 -1.42847741e+00 8.51145566e-01 8.19379449e-01 -1.61850661e-01 1.41428339e+00 -8.92308876e-02 7.11261570e-01 2.35409290e-02 1.35843292e-01 -9.74181354e-01 -3.96565571e-02 5.36497474e-01 6.23165905e-01 -1.61042392e+00 -1.61503017e-01 -7.81332016e-01 -6.85089111e-01 1.31607008e+00 1.12438953e+00 3.53778929e-01 7.60197759e-01 1.64836481e-01 3.99460606e-02 -6.51462153e-02 -5.99008799e-01 -2.87601560e-01 5.06282151e-01 5.79838157e-01 6.29788995e-01 -1.26604453e-01 3.53298225e-02 8.95526826e-01 -8.10484886e-02 1.99815840e-01 -6.81692585e-02 5.15339494e-01 -1.31908581e-01 -1.54279232e+00 -3.88961464e-01 5.10820568e-01 -7.74244368e-01 6.89750910e-02 -6.41550064e-01 5.10335863e-01 5.80491543e-01 8.57786357e-01 1.66486934e-01 -8.73497784e-01 3.30983885e-02 4.18943584e-01 3.33108068e-01 -4.55979466e-01 -8.71978104e-01 -4.04591173e-01 6.79204240e-02 -7.31169641e-01 -6.03952229e-01 -5.76436400e-01 -7.02826321e-01 -6.26341045e-01 3.20207551e-02 -2.37195507e-01 5.70656300e-01 9.64341521e-01 5.68145633e-01 2.14678511e-01 1.02842247e+00 -7.21430182e-01 -6.61986172e-01 -1.16793847e+00 -2.84476191e-01 6.68774724e-01 3.20841163e-01 -9.58778739e-01 2.59373579e-02 2.62637347e-01]
[13.468602180480957, 0.8603034019470215]
0aa71286-89de-4a95-b34f-282641e0eeae
towards-enhancing-health-coaching-dialogue-in
null
null
https://aclanthology.org/2022.coling-1.58
https://aclanthology.org/2022.coling-1.58.pdf
Towards Enhancing Health Coaching Dialogue in Low-Resource Settings
Health coaching helps patients identify and accomplish lifestyle-related goals, effectively improving the control of chronic diseases and mitigating mental health conditions. However, health coaching is cost-prohibitive due to its highly personalized and labor-intensive nature. In this paper, we propose to build a dialogue system that converses with the patients, helps them create and accomplish specific goals, and can address their emotions with empathy. However, building such a system is challenging since real-world health coaching datasets are limited and empathy is subtle. Thus, we propose a modularized health coaching dialogue with simplified NLU and NLG frameworks combined with mechanism-conditioned empathetic response generation. Through automatic and human evaluation, we show that our system generates more empathetic, fluent, and coherent responses and outperforms the state-of-the-art in NLU tasks while requiring less annotation. We view our approach as a key step towards building automated and more accessible health coaching systems.
['Shweta Yadav', 'Nikolaos Agadakos', 'Ben Gerber', 'Bing Liu', 'Lisa Sharp', 'Brian Ziebart', 'Barbara Di Eugenio', 'Yue Zhou']
null
null
null
null
coling-2022-10
['empathetic-response-generation']
['natural-language-processing']
[ 2.74456572e-02 8.60300124e-01 -4.40217465e-01 -4.92508680e-01 -5.99888384e-01 -2.30034649e-01 2.38395393e-01 5.00493050e-01 -1.15297221e-01 9.54503834e-01 8.32186341e-01 2.43414983e-01 9.45952488e-04 -5.41443348e-01 1.72074184e-01 -2.44569734e-01 3.87960702e-01 8.40630352e-01 -6.53781295e-01 -4.38398778e-01 9.64753795e-04 1.58158407e-01 -8.53971362e-01 4.62344855e-01 1.29334426e+00 3.80524039e-01 -2.68290877e-01 7.66379237e-01 1.11699045e-01 1.44591057e+00 -5.00175893e-01 -5.70898116e-01 -3.94566119e-01 -1.18974423e+00 -1.39996731e+00 -1.42779827e-01 -1.65554762e-01 -5.39485812e-01 3.06273609e-01 5.55487990e-01 8.92167091e-01 1.57715127e-01 3.79557431e-01 -1.11744678e+00 -7.15438247e-01 5.09642243e-01 1.82708114e-01 -6.51540577e-01 1.01126873e+00 3.04114491e-01 8.16303432e-01 -3.43744643e-02 5.89290142e-01 1.12217164e+00 8.50059032e-01 1.55647779e+00 -1.25260448e+00 -3.48391324e-01 -4.38977450e-01 -1.78263038e-02 -5.37829041e-01 -6.92433834e-01 6.95302546e-01 -3.55577826e-01 9.24377441e-01 5.15589952e-01 9.80360627e-01 1.37495303e+00 -2.77047753e-01 7.72960842e-01 1.07785201e+00 -2.94785321e-01 2.42333859e-01 4.08362836e-01 1.16852738e-01 7.58676469e-01 -2.96136767e-01 -2.08463490e-01 -4.94087070e-01 -4.67093021e-01 7.28829265e-01 -1.04305305e-01 -4.90092933e-01 1.51678190e-01 -1.30325460e+00 1.11378396e+00 4.10954893e-01 4.19115096e-01 -8.91333044e-01 1.42541066e-01 4.73769933e-01 2.61378020e-01 5.10606825e-01 9.58434582e-01 -1.15270339e-01 -6.68231428e-01 -6.88049912e-01 4.00793329e-02 1.45316303e+00 4.91346151e-01 1.70256868e-01 -4.40786242e-01 -4.76967007e-01 8.67823482e-01 5.07529266e-02 2.18084723e-01 3.28660607e-01 -1.59341228e+00 -2.16764063e-01 7.51305640e-01 3.45555216e-01 -8.40396702e-01 -8.32167804e-01 -2.53408164e-01 -7.22900569e-01 -1.55626118e-01 4.15879935e-01 -7.14109480e-01 9.00945365e-02 1.93915844e+00 6.61420405e-01 -2.02854931e-01 4.98136312e-01 9.40833092e-01 1.25782895e+00 3.10129076e-01 6.28585696e-01 -4.61706311e-01 1.55519700e+00 -1.25829375e+00 -1.01709390e+00 -1.27810538e-01 1.06856239e+00 -6.07625008e-01 1.22136450e+00 2.55950212e-01 -1.41936576e+00 -5.63846789e-02 -1.75842464e-01 -1.70104787e-01 3.39514762e-01 1.33088604e-01 8.43539178e-01 6.50452256e-01 -9.23530161e-01 7.40097642e-01 -4.79854673e-01 -1.06634974e+00 5.03668785e-01 2.20186338e-01 -4.76556480e-01 1.90263897e-01 -1.26002347e+00 1.07978535e+00 -6.07316243e-03 -3.47928792e-01 -2.97079116e-01 -9.72881258e-01 -7.51835465e-01 8.01310018e-02 2.51895159e-01 -1.43558180e+00 1.54971433e+00 -1.10821486e+00 -2.08370399e+00 1.15439749e+00 1.72313437e-01 -2.59647131e-01 7.05101848e-01 -3.60763311e-01 -1.62413478e-01 5.72808087e-01 -1.29780665e-01 1.12216532e+00 1.48511514e-01 -8.32096517e-01 -1.19264126e-01 -6.41688555e-02 1.41064316e-01 5.39204121e-01 -5.46011627e-01 3.80635679e-01 3.20409238e-01 -1.20287143e-01 -7.88972914e-01 -7.32274830e-01 -4.36117053e-01 1.79620385e-01 -3.21825147e-01 -2.43964180e-01 2.05956951e-01 -6.70738995e-01 9.95347142e-01 -1.98637986e+00 8.05572495e-02 -3.61329913e-01 5.06534755e-01 4.25946593e-01 -3.97545062e-02 9.41515088e-01 1.59030780e-01 4.36348579e-04 8.40949491e-02 -1.46885633e-01 2.19254401e-02 1.98391423e-01 1.54872432e-01 3.49758901e-02 6.14384934e-02 1.19230008e+00 -1.37524569e+00 -6.13558888e-01 9.40539613e-02 6.01410210e-01 -8.64569187e-01 6.48847103e-01 -3.50949824e-01 1.00321567e+00 -6.12441063e-01 3.10203969e-01 1.34435622e-02 -5.85534096e-01 2.16012180e-01 6.69101924e-02 7.85814896e-02 3.53759617e-01 -3.37595254e-01 1.65836906e+00 -5.96566796e-01 -4.14459556e-02 1.18188731e-01 -6.34865403e-01 9.64667976e-01 7.77413189e-01 7.93800116e-01 -4.32422847e-01 3.12122732e-01 -8.68886802e-03 -1.25055350e-02 -1.08417952e+00 1.53934687e-01 -5.87286770e-01 -1.42118022e-01 8.55038702e-01 -4.33353901e-01 -2.85047442e-01 -4.00635689e-01 3.93751264e-01 1.08149230e+00 3.96269793e-03 8.70685577e-01 1.20729301e-02 5.37822485e-01 4.03340161e-01 3.97770405e-01 3.78325880e-01 -7.46148705e-01 1.49666637e-01 5.92988670e-01 -2.98828244e-01 -4.81569171e-01 -4.41864789e-01 4.83244359e-01 9.97435451e-01 -1.62241757e-01 -2.50236303e-01 -1.07267153e+00 -5.57396233e-01 -3.42259109e-01 1.00188076e+00 -5.88420510e-01 -3.25557351e-01 -4.37770128e-01 -3.32312614e-01 9.17542577e-01 4.68316793e-01 5.34918785e-01 -1.45624626e+00 -9.91205037e-01 4.57859576e-01 -9.11535323e-01 -9.59878743e-01 -6.03742003e-01 -5.05341887e-01 -7.63373733e-01 -1.19263268e+00 -7.56850600e-01 -7.53963411e-01 5.19685090e-01 -1.79356206e-02 1.20423210e+00 1.47612393e-01 -1.87679082e-01 6.10441387e-01 -4.32880253e-01 -1.72964796e-01 -8.60289574e-01 -1.71381950e-01 -2.35287532e-01 -4.16857153e-01 2.11921811e-01 -4.64602023e-01 -9.67843354e-01 2.55529433e-01 -4.45430994e-01 5.89646995e-01 2.22709000e-01 8.52986693e-01 4.91440333e-02 -7.07728148e-01 9.99246120e-01 -1.29473794e+00 1.16811967e+00 -5.48523426e-01 4.80675846e-01 1.95712164e-01 -4.26563144e-01 -3.40307832e-01 7.34264195e-01 -6.10899925e-01 -1.23172843e+00 8.17059502e-02 -3.86718392e-01 2.44689807e-01 -3.86389375e-01 4.02170897e-01 7.65630156e-02 -3.46319415e-02 9.71661627e-01 -3.72517675e-01 3.07539463e-01 -3.12712520e-01 5.37896454e-01 7.83825099e-01 6.05099022e-01 -5.61853886e-01 1.22931458e-01 3.88399094e-01 -3.31330359e-01 -6.86682105e-01 -9.67993915e-01 -3.76302123e-01 4.01681401e-02 -4.95246500e-01 1.03107202e+00 -7.65102684e-01 -1.39019895e+00 1.54786646e-01 -1.20744514e+00 -7.31348455e-01 -5.02787948e-01 5.02620161e-01 -8.33294332e-01 4.25522864e-01 -8.63611341e-01 -7.74582684e-01 -1.05835676e+00 -5.84886670e-01 7.70739794e-01 4.68822777e-01 -1.37599695e+00 -1.16106462e+00 5.47284961e-01 9.86369610e-01 6.38861179e-01 5.54255605e-01 9.13342118e-01 -6.71856642e-01 3.27256769e-01 -1.51223198e-01 -1.43499106e-01 -1.31847262e-01 1.83713421e-01 -2.96419024e-01 -7.19381809e-01 2.33793885e-01 2.64980167e-01 -1.11697781e+00 -2.74943262e-02 1.62604004e-01 6.86656177e-01 -7.69919634e-01 -1.34264171e-01 2.30784789e-01 6.60243154e-01 -8.15628096e-02 6.12616241e-01 -1.24486394e-01 3.13676596e-01 1.32886541e+00 8.33055615e-01 9.04585540e-01 8.65365803e-01 4.85739380e-01 1.50050849e-01 -4.50279772e-01 7.44735003e-02 -2.01623201e-01 3.74393463e-01 6.27294958e-01 -8.61476809e-02 -6.95606768e-02 -7.03636706e-01 5.94658554e-01 -2.13018894e+00 -1.09818017e+00 -3.70609134e-01 1.73136783e+00 1.33563638e+00 -6.66676104e-01 4.47924018e-01 -3.14718992e-01 3.94426227e-01 -3.46804947e-01 -4.69817251e-01 -6.71231925e-01 2.88173705e-01 2.69531220e-01 -3.52636069e-01 5.89584351e-01 -5.48940957e-01 1.05803728e+00 5.92371225e+00 2.09782779e-01 -8.50358605e-01 2.87250221e-01 7.04350471e-01 -6.03389069e-02 -3.35199237e-01 -2.75855511e-01 3.21069025e-02 3.79363671e-02 9.08867955e-01 -7.47961327e-02 3.66500646e-01 6.12501562e-01 6.89081907e-01 -4.19451669e-02 -1.43535447e+00 9.23284471e-01 2.86953393e-02 -1.05700803e+00 -6.36342943e-01 -2.38951877e-01 4.30295140e-01 -5.21584809e-01 -3.04675728e-01 1.95542768e-01 5.64986765e-01 -1.22044599e+00 1.53426573e-01 7.33093202e-01 6.53116345e-01 -5.17976403e-01 6.26118779e-01 5.52704036e-01 -4.20634449e-01 2.65358597e-01 2.34436437e-01 -3.40810061e-01 4.67737585e-01 2.05971316e-01 -9.26823676e-01 2.54796118e-01 2.66493618e-01 7.16163814e-01 1.99705616e-01 7.02916741e-01 -4.66863662e-01 4.81421530e-01 1.22259576e-02 -3.87954324e-01 8.03003311e-02 -1.33738592e-01 2.10619867e-01 1.09290528e+00 5.22751212e-02 7.56936073e-01 3.87020260e-01 9.15761530e-01 -8.55653733e-02 6.38345063e-01 -2.71408141e-01 -4.35704976e-01 3.53863299e-01 1.61530614e+00 -2.62723505e-01 -4.01940018e-01 8.20673257e-03 1.20348048e+00 4.26392138e-01 -8.70055184e-02 -8.15112591e-01 -6.73560947e-02 5.25982797e-01 -6.26755804e-02 -6.17183089e-01 4.81376350e-01 -3.43605816e-01 -1.03973448e+00 -6.10706449e-01 -1.31325626e+00 5.09160995e-01 -7.23878562e-01 -1.39785242e+00 3.84487540e-01 -4.66758579e-01 -8.66536379e-01 -6.01426780e-01 2.70971693e-02 -8.09735179e-01 5.44930518e-01 -1.24154758e+00 -1.41161180e+00 -7.03671932e-01 5.96030712e-01 1.24307305e-01 3.63935083e-01 1.46292460e+00 1.95660040e-01 -5.39568543e-01 5.53310037e-01 -6.23274267e-01 -1.04476012e-01 1.23931944e+00 -9.85472858e-01 -2.68582374e-01 -1.06966682e-01 -7.79203236e-01 5.05169630e-01 7.95564234e-01 -6.15546167e-01 -1.07315147e+00 -8.67042720e-01 1.53087819e+00 -1.29926085e-01 5.64760625e-01 9.34987664e-02 -8.54952693e-01 3.03310871e-01 4.93178815e-01 -5.24728179e-01 1.46454787e+00 2.79646385e-02 7.09925266e-03 4.50904131e-01 -1.68696713e+00 8.55561078e-01 1.00437152e+00 -4.74031568e-01 -5.38871050e-01 8.88748467e-01 7.42182672e-01 -2.00527817e-01 -1.28851688e+00 -1.03722766e-01 4.97827232e-01 -9.72483337e-01 7.64999211e-01 -9.39487159e-01 8.04977417e-01 3.45171899e-01 3.62436801e-01 -1.33850598e+00 -2.45956868e-01 -1.23005176e+00 9.51142237e-02 1.19749928e+00 5.24293371e-02 -6.71946526e-01 8.59404564e-01 1.28192496e+00 -2.13850122e-02 -7.36012220e-01 -3.32020402e-01 -2.06057563e-01 -9.52064469e-02 1.31859064e-01 3.56897831e-01 1.53843498e+00 1.05079639e+00 6.66843176e-01 -8.22677433e-01 -4.23896253e-01 2.43721560e-01 7.10063726e-02 8.80894661e-01 -1.22490728e+00 -6.95605516e-01 -5.10772526e-01 3.50209564e-01 -5.57441890e-01 1.64894491e-01 -6.40894413e-01 -9.87045467e-03 -1.74910259e+00 4.46024030e-01 -3.72236460e-01 4.29196686e-01 1.00104153e+00 -2.51562536e-01 2.78039519e-02 5.12639321e-02 2.80504525e-01 -6.86869383e-01 5.25770366e-01 1.41951144e+00 2.68975705e-01 -5.61386585e-01 -1.20461136e-01 -1.24998939e+00 6.36205316e-01 1.03812826e+00 -3.66964221e-01 -2.67231524e-01 7.93106947e-03 2.90852129e-01 6.87902033e-01 3.05573285e-01 -5.00936151e-01 1.47763371e-01 -4.05255288e-01 -1.72813967e-01 8.60984549e-02 3.52237046e-01 -4.87104684e-01 4.29038763e-01 1.05835843e+00 -7.87779152e-01 -4.48698968e-01 -1.98025018e-01 1.80040039e-02 1.12036698e-01 -3.20394546e-01 9.24945593e-01 -3.26991171e-01 1.21900868e-02 -2.27037996e-01 -4.25013006e-01 2.39450067e-01 1.24605155e+00 7.88460374e-02 -6.64792597e-01 -1.15869141e+00 -7.72920132e-01 5.60447991e-01 6.90928102e-01 -8.60703215e-02 5.14791369e-01 -1.08450925e+00 -8.45613301e-01 -3.94208312e-01 1.13207065e-01 -3.25751066e-01 5.16172051e-01 1.23515010e+00 -6.25029802e-01 1.82487756e-01 -3.94578576e-01 -1.01569155e-02 -1.57500935e+00 3.01727802e-01 3.91239136e-01 -6.11607492e-01 -5.30315757e-01 7.04757214e-01 2.11876556e-02 -6.73317134e-01 2.15658873e-01 1.83822080e-01 -5.96359491e-01 9.51748118e-02 5.59324980e-01 5.46541154e-01 -4.89403665e-01 -4.55186039e-01 -1.03935741e-01 1.46514565e-01 1.55117393e-01 -9.58983451e-02 1.34814107e+00 -1.09419256e-01 -2.53054321e-01 -6.21830113e-02 7.42891610e-01 -1.16580985e-01 -8.38266671e-01 1.09668057e-02 -2.15616018e-01 -9.95031968e-02 -2.41704434e-01 -1.25223076e+00 -5.42299449e-01 8.27289343e-01 4.30957451e-02 1.03764556e-01 1.15464377e+00 4.87429313e-02 1.38231361e+00 3.67724508e-01 1.10416934e-02 -1.05149293e+00 5.62026143e-01 1.05375156e-01 8.09055567e-01 -1.20205605e+00 -2.47161239e-01 -5.08712351e-01 -1.30335248e+00 9.50452149e-01 6.42631531e-01 2.61206388e-01 -6.04340397e-02 -1.54197618e-01 5.28882921e-01 -4.37022865e-01 -9.79888618e-01 -6.00409843e-02 -5.40104546e-02 6.82927787e-01 8.31147492e-01 3.11652035e-01 -6.95819318e-01 7.65564263e-01 -2.30130300e-01 3.41123998e-01 5.17098546e-01 8.22125196e-01 -3.11745495e-01 -1.44893694e+00 -1.86996862e-01 4.71760295e-02 -4.12744820e-01 4.97766398e-03 -1.07396042e+00 4.61421430e-01 -1.87636256e-01 1.25206983e+00 -4.03956383e-01 1.50996279e-02 3.18006068e-01 4.65782583e-02 4.76208210e-01 -8.38192999e-01 -1.43395543e+00 -2.54096445e-02 7.79680669e-01 -7.62133062e-01 -6.55519426e-01 -5.82060158e-01 -1.63346422e+00 -5.21780074e-01 -4.26217727e-03 1.78015411e-01 2.83186346e-01 8.85650039e-01 6.19584858e-01 3.46160889e-01 5.64446449e-01 -3.98344934e-01 -5.60166061e-01 -8.35024476e-01 1.69182971e-01 8.55909467e-01 -6.78585917e-02 1.35067264e-02 1.94463849e-01 -1.08026259e-01]
[13.108696937561035, 7.64446496963501]
87fb4865-d8b8-4d56-885b-c4cbf3c87059
cca-mdd-a-coupled-cross-attention-based
2111.08191
null
https://arxiv.org/abs/2111.08191v2
https://arxiv.org/pdf/2111.08191v2.pdf
CoCA-MDD: A Coupled Cross-Attention based Framework for Streaming Mispronunciation Detection and Diagnosis
Mispronunciation detection and diagnosis (MDD) is a popular research focus in computer-aided pronunciation training (CAPT) systems. End-to-end (e2e) approaches are becoming dominant in MDD. However an e2e MDD model usually requires entire speech utterances as input context, which leads to significant time latency especially for long paragraphs. We propose a streaming e2e MDD model called CoCA-MDD. We utilize conv-transformer structure to encode input speech in a streaming manner. A coupled cross-attention (CoCA) mechanism is proposed to integrate frame-level acoustic features with encoded reference linguistic features. CoCA also enables our model to perform mispronunciation classification with whole utterances. The proposed model allows system fusion between the streaming output and mispronunciation classification output for further performance enhancement. We evaluate CoCA-MDD on publicly available corpora. CoCA-MDD achieves F1 scores of 57.03% and 60.78% for streaming and fusion modes respectively on L2-ARCTIC. For phone-level pronunciation scoring, CoCA-MDD achieves 0.58 Pearson correlation coefficient (PCC) value on SpeechOcean762.
['Xiao Chen', 'Qun Liu', 'Xin Jiang', 'Yasheng Wang', 'Yuanyuan Guo', 'Baohua Xu', 'Yu Ting Yeung', 'Wenyong Huang', 'Liqun Deng', 'Nianzu Zheng']
2021-11-16
null
null
null
null
['phone-level-pronunciation-scoring']
['speech']
[ 1.97598599e-02 -7.48424232e-02 1.70394868e-01 -4.82723743e-01 -1.48964846e+00 -3.31347913e-01 4.31018591e-01 1.58842504e-01 -4.04128551e-01 3.78948122e-01 4.54981744e-01 -4.81808007e-01 2.88802356e-01 -1.76496565e-01 -6.10793948e-01 -4.09135848e-01 2.44320109e-01 -6.85094018e-03 -1.59917444e-01 7.11577609e-02 -5.49791455e-02 1.72595009e-01 -1.52688074e+00 8.88487041e-01 1.01960325e+00 1.02291656e+00 6.08828664e-01 1.26966929e+00 -5.64121045e-02 5.06930828e-01 -9.08948302e-01 -7.59481251e-01 -2.93176651e-01 -2.59282142e-01 -4.34089214e-01 -3.46630439e-02 4.33037728e-01 -2.80226141e-01 -3.53382975e-01 1.13647294e+00 9.96381402e-01 -7.75215924e-02 6.09231174e-01 -1.11937070e+00 -5.14057577e-01 5.39020181e-01 -2.84273446e-01 3.74537408e-01 3.23229641e-01 1.34127840e-01 6.80078387e-01 -1.55559325e+00 6.00772686e-02 1.31941068e+00 6.22250676e-01 6.33743823e-01 -7.75800765e-01 -7.51293123e-01 1.63871162e-02 5.38661242e-01 -1.31562686e+00 -1.09506404e+00 5.77371836e-01 -2.18540788e-01 1.36578178e+00 4.12853718e-01 3.49853605e-01 1.22492456e+00 2.52259105e-01 1.24169052e+00 5.72823524e-01 -4.03917789e-01 1.29228637e-01 2.52154201e-01 5.36978915e-02 1.31540358e-01 -2.12910414e-01 8.30712467e-02 -9.00920331e-01 1.79670691e-01 3.80918771e-01 -3.96700680e-01 -2.57698745e-01 7.65739202e-01 -1.16503549e+00 3.25652272e-01 -8.42816532e-02 3.16004783e-01 -5.40997088e-01 -1.53869972e-01 7.57243037e-01 3.46649349e-01 4.92952317e-01 1.43544627e-02 -4.78300691e-01 -6.88738644e-01 -9.64389205e-01 3.65069360e-02 3.08349222e-01 1.05921316e+00 -1.09481722e-01 3.84841204e-01 -3.93672377e-01 1.54995978e+00 4.51833934e-01 8.74938965e-01 7.60237515e-01 -6.79872990e-01 8.26052666e-01 -1.95380628e-01 -9.40034539e-02 -7.23154247e-01 2.63477992e-02 -6.45781875e-01 -7.47668743e-01 -1.49587616e-01 -8.56479555e-02 -4.00806338e-01 -6.06500566e-01 1.65009546e+00 8.41461793e-02 3.29229206e-01 6.56998694e-01 7.24868298e-01 9.71924186e-01 1.17307258e+00 7.85346702e-02 -3.70112360e-01 1.34137785e+00 -1.15091848e+00 -1.19852424e+00 -1.74911276e-01 7.92404771e-01 -1.13464952e+00 8.87271047e-01 5.92500389e-01 -1.36738122e+00 -9.25082207e-01 -8.81634235e-01 4.39181253e-02 6.21420369e-02 6.94203734e-01 -1.49919733e-01 4.78544056e-01 -1.07180929e+00 2.65968263e-01 -7.67217159e-01 1.55795282e-02 3.50069791e-01 2.48390451e-01 -5.28624177e-01 -7.46629387e-02 -1.26903474e+00 5.95868468e-01 5.12972847e-02 2.83298880e-01 -8.45233977e-01 -7.22834289e-01 -9.74667728e-01 1.37981456e-02 -2.22155616e-01 -1.29337758e-01 1.76179349e+00 -8.17924440e-01 -1.73142922e+00 5.31719446e-01 -6.40620112e-01 -5.98542511e-01 3.79225373e-01 -5.27185738e-01 -9.80628371e-01 2.15337798e-01 2.42912415e-02 8.76172543e-01 7.70110488e-01 -7.74164975e-01 -8.24201167e-01 -9.92401466e-02 -7.27563024e-01 6.26846254e-01 -4.39463347e-01 2.34301075e-01 -4.20225739e-01 -9.12273467e-01 1.36738289e-02 -6.12042308e-01 3.31237912e-01 -3.47392023e-01 -4.88793284e-01 -5.82795203e-01 8.78079295e-01 -1.12936246e+00 1.53387976e+00 -2.47117496e+00 -2.64882356e-01 -4.65997040e-01 -1.47870898e-01 8.41145396e-01 -1.97400212e-01 3.12271297e-01 2.51481868e-02 -1.91831380e-01 -9.11618695e-02 -8.63362849e-01 -2.22505108e-02 -1.26943976e-01 -2.07475442e-02 2.55346149e-01 4.48342472e-01 6.40737295e-01 -7.58520782e-01 -4.46916401e-01 3.98268729e-01 7.23400533e-01 -7.11028934e-01 6.27755105e-01 1.61274582e-01 1.91601694e-01 2.65250746e-02 5.61120927e-01 8.20096135e-01 3.15699399e-01 -2.91642159e-01 -8.67263749e-02 -1.64709806e-01 7.31485069e-01 -8.56866896e-01 1.71945369e+00 -8.10219586e-01 8.97251666e-01 2.11627632e-01 -8.19205523e-01 1.02312291e+00 1.02415812e+00 1.55461207e-01 -6.16596520e-01 3.12628388e-01 5.27583778e-01 2.33993813e-01 -6.92444980e-01 5.26184320e-01 5.58587373e-04 9.23515856e-02 -1.80491179e-01 2.52962857e-01 1.40967607e-01 -2.69675404e-01 1.03182480e-01 1.00756192e+00 -3.87747407e-01 2.19393410e-02 -3.96639034e-02 6.18814170e-01 -4.94186401e-01 4.43612784e-01 3.69175524e-01 -5.43936789e-01 8.97975683e-01 1.78434104e-02 2.80145228e-01 -9.81312215e-01 -1.09404790e+00 -2.70193994e-01 6.88274086e-01 -2.82962233e-01 -4.15667415e-01 -9.12048578e-01 -5.96895337e-01 -2.13856056e-01 8.85919392e-01 -1.31053746e-01 -3.22012454e-01 -3.06896508e-01 -1.32916451e-01 8.00309718e-01 6.53910160e-01 5.24415433e-01 -1.10227108e+00 -6.15541190e-02 4.90224928e-01 -4.15473193e-01 -1.34856188e+00 -8.04553270e-01 1.24000736e-01 -4.68300849e-01 -3.29900891e-01 -1.10623491e+00 -1.08626390e+00 2.26337656e-01 1.71643361e-01 7.44906902e-01 -5.69111824e-01 1.24282226e-01 2.68633720e-02 -4.90862101e-01 -4.02736098e-01 -6.82450414e-01 -1.84400529e-01 3.73959661e-01 2.16602117e-01 6.70103610e-01 -3.58581990e-01 -5.48558116e-01 1.99615255e-01 -3.36268544e-01 2.32047752e-01 5.94321311e-01 1.04920912e+00 5.77184558e-01 -2.46061563e-01 1.26590967e+00 -4.78107870e-01 6.69824481e-01 -7.66146004e-01 -2.79751658e-01 1.85190793e-02 -3.25595260e-01 -4.06695753e-01 7.56937742e-01 -5.94905794e-01 -1.13282323e+00 -8.08737874e-02 -1.01329565e+00 -1.01435196e+00 -4.66601253e-01 3.79351169e-01 -6.57385290e-01 7.24901140e-01 2.05709085e-01 3.67918164e-01 -4.96543199e-02 -5.90506434e-01 5.36684133e-02 1.75911927e+00 1.05576360e+00 -1.76856890e-01 -4.15265337e-02 -4.98705894e-01 -8.44201624e-01 -1.07920015e+00 -3.28043342e-01 -5.59641063e-01 -4.26220000e-02 -2.04836830e-01 8.71285737e-01 -1.23025346e+00 -6.53608620e-01 8.05047214e-01 -1.54285991e+00 1.06008396e-01 1.69789955e-01 9.49912190e-01 -3.96483749e-01 7.35329017e-02 -6.34956956e-01 -1.12564754e+00 -5.65578818e-01 -1.40754378e+00 1.18924952e+00 1.14561565e-01 -2.94458449e-01 -5.90592027e-01 -2.46621802e-01 5.15962720e-01 4.06390667e-01 -3.87360632e-01 3.60194474e-01 -9.67880011e-01 2.70652622e-01 -3.21997464e-01 -9.61291194e-02 9.33279216e-01 2.45360449e-01 -1.95793808e-01 -1.44573653e+00 -1.27701148e-01 1.34184659e-01 -9.15388688e-02 3.45022410e-01 6.60602510e-01 1.27061784e+00 -5.19310474e-01 -1.28961310e-01 2.74021655e-01 8.64040554e-01 7.51700282e-01 6.10538304e-01 -4.02553290e-01 5.19564748e-01 5.65652251e-01 8.89677405e-01 3.61693799e-01 3.16125512e-01 7.64989555e-01 -1.57446936e-02 7.95136467e-02 -5.35821259e-01 -4.41873193e-01 7.83092916e-01 1.70463884e+00 7.36088872e-01 -6.34057760e-01 -9.11359966e-01 9.37680006e-01 -1.41954362e+00 -8.31308126e-01 -1.02455884e-01 1.89691997e+00 9.50985074e-01 1.31852210e-01 -1.56405807e-01 5.57657957e-01 1.07948375e+00 -1.80178657e-01 -6.07531965e-01 -8.06798637e-01 -1.01144865e-01 2.27819994e-01 2.18460485e-02 7.17140555e-01 -8.76572013e-01 6.94350839e-01 4.92449665e+00 1.29484522e+00 -1.18675184e+00 3.31390649e-01 7.88723588e-01 -2.11490914e-01 -6.96988031e-02 -5.89350224e-01 -1.01585376e+00 8.56127620e-01 1.64627647e+00 -1.58061162e-01 -7.05448240e-02 8.00616682e-01 4.40475941e-01 6.93654642e-02 -1.09047163e+00 1.47218847e+00 1.44202456e-01 -1.07306778e+00 -1.04803614e-01 -2.05268949e-01 4.23530549e-01 6.14956245e-02 3.32568079e-01 3.56964052e-01 -2.92344987e-01 -8.92782152e-01 8.22789848e-01 1.86444968e-01 1.23325789e+00 -1.00394499e+00 8.24210882e-01 3.05007279e-01 -1.19332707e+00 5.50501160e-02 -1.52138531e-01 2.84926325e-01 2.85128027e-01 5.41401625e-01 -1.13720930e+00 4.52283204e-01 5.35478711e-01 7.84930885e-01 -9.97946411e-02 1.00083470e+00 1.00020319e-01 1.06758916e+00 -2.23801851e-01 -8.08854178e-02 7.74565861e-02 3.99714053e-01 7.02332675e-01 1.66911972e+00 6.50484025e-01 8.66569877e-02 -3.97061199e-01 2.69214541e-01 -2.58244723e-01 2.20636636e-01 -4.82599616e-01 -2.19634980e-01 1.03729856e+00 7.23848403e-01 3.32318470e-02 -4.91177857e-01 -4.32754338e-01 1.14794934e+00 -7.45997280e-02 2.20427528e-01 -8.21986198e-01 -8.70660365e-01 1.00872922e+00 -2.36202642e-01 3.47441345e-01 5.37439212e-02 -1.10979222e-01 -1.05927098e+00 2.15738326e-01 -1.09908187e+00 -1.69285819e-01 -8.77165437e-01 -1.13438070e+00 1.04307926e+00 -3.31950277e-01 -1.49690366e+00 -3.80255640e-01 -5.26641130e-01 -5.10620534e-01 1.03959763e+00 -1.40512896e+00 -7.50320733e-01 8.17661881e-02 5.21621108e-01 1.45393646e+00 -4.65636134e-01 8.99185598e-01 8.46702516e-01 -6.90111399e-01 1.25067413e+00 1.39484629e-01 7.89483637e-02 7.07643330e-01 -9.77179646e-01 5.23574650e-01 1.07412720e+00 9.76149738e-02 2.18183428e-01 6.01320744e-01 -5.36512315e-01 -1.15858555e+00 -1.40030479e+00 1.38157451e+00 -6.35404661e-02 1.45035014e-01 -2.67250627e-01 -1.00803101e+00 1.73693955e-01 4.77691948e-01 1.37365669e-01 6.39514863e-01 -2.62610972e-01 -4.60858084e-02 -1.58468619e-01 -9.73914802e-01 3.33731949e-01 6.59681499e-01 -8.90106201e-01 -4.67200160e-01 2.83430647e-02 1.12282121e+00 -3.48056406e-01 -8.02346766e-01 4.39189970e-01 4.11313742e-01 -7.33232737e-01 5.32016516e-01 -9.26937684e-02 3.80283117e-01 -8.15325379e-02 -5.96022189e-01 -1.38331032e+00 1.96842924e-01 -7.08178282e-01 -2.04294294e-01 1.53790450e+00 6.67078137e-01 -3.12015325e-01 3.35886300e-01 1.38962895e-01 -8.45779955e-01 -9.30636823e-01 -1.06459737e+00 -8.25073838e-01 9.60580483e-02 -1.14461660e+00 4.59022254e-01 5.82949340e-01 1.96432233e-01 3.48170847e-01 -3.60586911e-01 3.88275206e-01 2.49784783e-01 -6.49213314e-01 3.24113309e-01 -5.22255838e-01 -2.92960018e-01 -3.24332476e-01 -3.50901723e-01 -1.34207857e+00 1.04840793e-01 -9.18699086e-01 3.54707241e-01 -1.24476278e+00 -3.42547834e-01 -2.56733358e-01 -2.92079359e-01 1.82092756e-01 -1.88995481e-01 -5.65155819e-02 4.68948372e-02 -3.89056168e-02 -2.85124987e-01 8.45006704e-01 8.49367857e-01 -7.66716748e-02 -2.19585821e-01 2.50562072e-01 -2.91435629e-01 4.08948332e-01 8.62017453e-01 -4.21421975e-01 -4.03462678e-01 -7.44393408e-01 -4.98917371e-01 6.26298845e-01 -2.06178904e-01 -1.24634159e+00 3.35926294e-01 3.55779469e-01 1.38702169e-01 -8.82375777e-01 6.05283797e-01 -4.60099429e-01 -1.06950821e-02 3.78666461e-01 -4.79708523e-01 1.88643068e-01 4.72434431e-01 3.49845648e-01 -7.53654718e-01 -5.71309403e-02 7.98778594e-01 4.09526139e-01 -2.54526585e-01 1.18213557e-01 -6.30946159e-01 -1.62949622e-01 6.92309558e-01 -4.02994640e-02 -1.38019279e-01 -4.12803352e-01 -8.37191701e-01 8.27934369e-02 -3.02943945e-01 7.94868171e-01 1.00397742e+00 -1.55617380e+00 -1.06637144e+00 5.89288533e-01 1.76296324e-01 -1.00847445e-02 5.41712403e-01 7.76849449e-01 -2.47338250e-01 6.38907492e-01 1.69656813e-01 -7.65933156e-01 -1.56301332e+00 1.07158974e-01 3.15524191e-01 3.90934169e-01 -4.83075202e-01 1.32870603e+00 1.55497313e-01 -1.37604013e-01 7.03979611e-01 -3.35446805e-01 -2.02689603e-01 -4.39099632e-02 7.75421023e-01 3.52383226e-01 4.90113676e-01 -9.26393867e-01 -5.13749659e-01 1.97075326e-02 -3.46128315e-01 -5.42445123e-01 9.49606895e-01 -3.92416835e-01 5.06715357e-01 4.93038237e-01 1.54818881e+00 5.20248078e-02 -1.09418297e+00 -1.50087297e-01 -3.36313725e-01 -2.71910608e-01 3.06238830e-01 -8.98162961e-01 -9.14156914e-01 1.45829427e+00 8.23854804e-01 -7.13244602e-02 1.22290635e+00 -8.04051235e-02 1.20761180e+00 4.96794358e-02 -1.29385188e-01 -9.39983606e-01 -7.15430826e-02 5.54853499e-01 1.04986620e+00 -1.19244087e+00 -9.10368383e-01 -1.50190920e-01 -9.41594124e-01 9.42434669e-01 6.89352691e-01 1.05286159e-01 7.56098747e-01 4.83096689e-01 2.73276657e-01 3.98135364e-01 -1.10691059e+00 1.23219699e-01 2.57722676e-01 5.10812342e-01 5.98135948e-01 2.21269101e-01 -1.17809519e-01 1.12690234e+00 -5.41243136e-01 -2.85258472e-01 2.18544856e-01 6.28612220e-01 -2.88896948e-01 -8.82499337e-01 -3.92766744e-01 3.54283124e-01 -7.08002865e-01 -3.85224193e-01 1.52111396e-01 6.28492050e-03 6.03581630e-02 1.55544245e+00 3.42636317e-01 -7.50250936e-01 3.81356418e-01 2.58660495e-01 5.60548343e-02 -6.04724705e-01 -6.07271433e-01 6.03699088e-01 1.47671342e-01 -2.78239280e-01 4.24353518e-02 -7.93997943e-01 -1.22793543e+00 -8.28250051e-02 -5.19059002e-01 4.54862028e-01 1.00377524e+00 8.75454068e-01 7.12509751e-01 8.61882746e-01 9.03131604e-01 -5.26851892e-01 -2.65005946e-01 -1.47611046e+00 -3.24209750e-01 1.39600178e-02 8.04527581e-01 -2.58171141e-01 -2.77191520e-01 1.65774494e-01]
[14.619193077087402, 6.228481769561768]
49ed58cf-e31f-4acf-95fd-d715a60627fe
multilingual-dependency-parsing-for-low-1
null
null
https://aclanthology.org/2021.iwpt-1.9
https://aclanthology.org/2021.iwpt-1.9.pdf
Multilingual Dependency Parsing for Low-Resource African Languages: Case Studies on Bambara, Wolof, and Yoruba
This paper describes a methodology for syntactic knowledge transfer between high-resource languages to extremely low-resource languages. The methodology consists in leveraging multilingual BERT self-attention model pretrained on large datasets to develop a multilingual multi-task model that can predict Universal Dependencies annotations for three African low-resource languages. The UD annotations include universal part-of-speech, morphological features, lemmas, and dependency trees. In our experiments, we used multilingual word embeddings and a total of 11 Universal Dependencies treebanks drawn from three high-resource languages (English, French, Norwegian) and three low-resource languages (Bambara, Wolof and Yoruba). We developed various models to test specific language combinations involving contemporary contact languages or genetically related languages. The results of the experiments show that multilingual models that involve high-resource languages and low-resource languages with contemporary contact between each other can provide better results than combinations that only include unrelated languages. As far genetic relationships are concerned, we could not draw any conclusion regarding the impact of language combinations involving the selected low-resource languages, namely Wolof and Yoruba.
['Cheikh M. Bamba Dione']
null
null
null
null
acl-iwpt-2021-8
['multilingual-word-embeddings']
['methodology']
[-5.45044601e-01 -1.86261218e-02 -2.46166483e-01 -4.02095675e-01 -6.26771331e-01 -4.92954344e-01 4.95010614e-01 3.57172698e-01 -8.68911564e-01 1.26896584e+00 5.55923522e-01 -4.91831809e-01 -4.89239255e-03 -5.95664740e-01 -6.79915011e-01 -4.12589580e-01 -1.58106610e-01 7.24951625e-01 1.83298618e-01 -6.63280964e-01 -6.09823652e-02 3.26695502e-01 -9.64367330e-01 1.93139225e-01 1.08270061e+00 -3.06682676e-01 4.75555509e-01 5.89237273e-01 -1.17521740e-01 5.72652102e-01 -5.36132157e-01 -7.14209139e-01 8.41630995e-03 -4.08226192e-01 -8.84824276e-01 -6.72962427e-01 -9.70590487e-02 3.04144472e-02 1.03776723e-01 9.90694642e-01 5.25863588e-01 -3.52914333e-01 5.57181060e-01 -5.41653931e-01 -1.24884641e+00 1.20231891e+00 -5.79200685e-01 5.32260418e-01 4.25194830e-01 -1.44600779e-01 1.23053539e+00 -8.43820870e-01 9.84887660e-01 1.62301266e+00 7.44150341e-01 7.36697391e-02 -8.99923325e-01 -6.18739903e-01 2.18391582e-01 1.10036276e-01 -1.50531828e+00 -3.07214558e-01 3.99822056e-01 -4.15969253e-01 1.51119626e+00 -1.92792207e-01 2.73096174e-01 9.55924094e-01 3.84168148e-01 1.34703770e-01 1.30132151e+00 -9.84549284e-01 -4.86243308e-01 5.48267245e-01 1.69822633e-01 8.79578650e-01 5.23670733e-01 -4.65635629e-03 -5.47136605e-01 -5.79434112e-02 4.98145819e-01 -7.40379095e-01 4.31562774e-02 3.52265745e-01 -1.22073162e+00 1.11745167e+00 5.62944375e-02 1.06940043e+00 -1.88950837e-01 -2.27016106e-01 5.86688578e-01 4.84713554e-01 8.77641439e-01 1.92295685e-01 -1.05187094e+00 2.01051742e-01 -2.36085176e-01 -2.38936365e-01 8.37030649e-01 8.21377575e-01 9.86355722e-01 2.21929818e-01 2.61187285e-01 1.35914683e+00 3.52566153e-01 5.26267827e-01 5.32660246e-01 4.11756150e-02 4.74491507e-01 1.84632912e-01 -2.05238938e-01 -6.09997630e-01 -5.05132854e-01 6.22340441e-02 -2.43549302e-01 -2.23001599e-01 3.22284549e-01 -7.23751247e-01 -7.48099923e-01 1.87879694e+00 3.04380596e-01 -4.21261638e-01 6.89407706e-01 6.41960859e-01 7.41109133e-01 7.27073848e-01 3.84867102e-01 -1.12785645e-01 1.64790606e+00 -9.55223739e-01 -6.41797841e-01 -2.35244736e-01 8.49886417e-01 -9.81703758e-01 1.00644338e+00 -8.64861533e-02 -7.99854338e-01 -7.16275990e-01 -9.25791800e-01 -2.17573524e-01 -1.04961598e+00 1.81444466e-01 8.63170385e-01 9.21360075e-01 -1.00835240e+00 2.94552952e-01 -6.57003224e-01 -8.92536521e-01 -3.14920783e-01 2.69200116e-01 -7.48773873e-01 -1.60377309e-01 -1.71792674e+00 1.54472375e+00 9.04779017e-01 1.69151034e-02 -6.18806124e-01 -4.81270790e-01 -1.11003423e+00 -3.48370314e-01 -4.95549366e-02 -5.91916591e-02 4.89578456e-01 -1.09307909e+00 -1.22247958e+00 1.14189899e+00 1.97571129e-01 -3.56432647e-02 4.12037261e-02 -4.68515515e-01 -1.06634474e+00 -3.70830417e-01 3.16475600e-01 2.96489000e-01 3.03306766e-02 -7.89331019e-01 -8.16118777e-01 -2.97049582e-01 -1.32250667e-01 1.47859886e-01 -3.21812332e-01 1.06686223e+00 -2.21735686e-01 -7.26319730e-01 -6.39507353e-01 -9.02535677e-01 -1.41510352e-01 -7.46705949e-01 -1.69598714e-01 -2.85252839e-01 3.32303733e-01 -1.31611252e+00 8.62010062e-01 -1.83267355e+00 2.76676774e-01 -1.08628727e-01 -6.68486059e-01 3.55413258e-01 -4.01777774e-01 5.98376870e-01 -1.68695241e-01 4.26009536e-01 8.20249543e-02 2.48270765e-01 -9.49585959e-02 7.28970885e-01 4.62444127e-01 4.65942055e-01 7.93464005e-01 6.99071765e-01 -9.86450434e-01 -4.76605326e-01 2.98118014e-02 7.34005570e-01 -4.51703846e-01 -3.40203270e-02 -3.88232656e-02 4.61505353e-01 -2.14874640e-01 7.07687140e-01 3.82168710e-01 5.66477954e-01 5.49523890e-01 -1.42835230e-01 -5.83054245e-01 3.06530148e-01 -6.76999211e-01 1.83977664e+00 -8.69031370e-01 4.07165557e-01 -6.37210533e-02 -6.59728229e-01 9.62803245e-01 2.48268083e-01 -1.16641700e-01 -6.56394303e-01 -1.21562451e-03 6.33092165e-01 6.63028657e-01 -6.13228500e-01 5.04501402e-01 -2.45563224e-01 -5.16859829e-01 1.64018691e-01 6.55712605e-01 3.86202820e-02 4.22052830e-01 -2.68195063e-01 7.34503269e-01 6.34637415e-01 7.46879756e-01 -6.93084121e-01 7.09619403e-01 -4.30995226e-02 9.91593003e-01 9.07315388e-02 -1.00007132e-01 3.65906842e-02 5.41454852e-01 -5.14597237e-01 -1.15427780e+00 -1.01177454e+00 -5.02865553e-01 1.78665960e+00 -3.96740168e-01 -3.53077233e-01 -1.41992196e-01 -7.15628207e-01 -1.08398788e-01 7.27017045e-01 -6.28406227e-01 2.13463217e-01 -8.14481735e-01 -1.31400061e+00 8.87924254e-01 2.87603587e-01 1.10924333e-01 -1.23725510e+00 -6.58820868e-02 2.18978629e-01 1.57761022e-01 -1.22375166e+00 -1.57187343e-01 4.52520460e-01 -2.32026681e-01 -8.20799112e-01 -8.78027260e-01 -1.15303540e+00 3.80902410e-01 -6.69932842e-01 1.11591327e+00 -1.67640537e-01 -2.84262270e-01 -8.53408501e-02 -6.31358027e-01 -6.33222222e-01 -7.19225228e-01 1.48665816e-01 2.46401325e-01 -3.69496614e-01 6.35573089e-01 -1.64638802e-01 3.49993110e-01 -4.95780446e-02 -6.27987325e-01 -3.60683680e-01 7.57795393e-01 8.26430023e-01 4.52388287e-01 -3.47836882e-01 9.00871158e-01 -1.21005023e+00 3.60610366e-01 -7.96903074e-01 -3.84804457e-01 6.92000449e-01 4.82180975e-02 3.18260729e-01 6.66194201e-01 -3.42255592e-01 -1.37784016e+00 -1.72663689e-01 -5.32396853e-01 3.58431607e-01 -2.58796871e-01 7.62973547e-01 -3.28590900e-01 7.16658980e-02 4.65499490e-01 -2.11434260e-01 -6.95017517e-01 -6.58660829e-01 5.89227140e-01 6.92100704e-01 2.39366680e-01 -9.12671804e-01 5.12216151e-01 -3.10620099e-01 -4.89892811e-01 -1.40338516e+00 -6.39368773e-01 -1.00670733e-01 -1.33904171e+00 1.99924722e-01 1.48341727e+00 -1.14807129e+00 1.85733801e-03 3.48529935e-01 -1.27334547e+00 -3.09452981e-01 5.57761863e-02 7.85249531e-01 1.27732471e-01 3.18577439e-02 -9.94696736e-01 -4.16714877e-01 -2.07862765e-01 -9.47294533e-01 6.28799438e-01 -2.16122884e-02 -2.00290427e-01 -1.58941805e+00 5.87254047e-01 1.17138423e-01 2.57798702e-01 3.00665855e-01 1.51049912e+00 -1.12601805e+00 1.88472152e-01 3.38976592e-01 -1.61050215e-01 3.69369417e-01 4.83389676e-01 2.73632377e-01 -6.32646203e-01 -3.24028045e-01 -6.19511843e-01 -6.68510318e-01 4.49549437e-01 -5.61568066e-02 -5.48827983e-02 1.66228741e-01 -5.28239422e-02 4.67872441e-01 1.67421174e+00 1.79206565e-01 1.48028672e-01 2.99597740e-01 8.36156189e-01 8.53519738e-01 5.83891630e-01 4.76852320e-02 6.17503047e-01 3.42133731e-01 -3.17289859e-01 -2.08209023e-01 -1.72711298e-01 -1.00804642e-01 8.48317266e-01 1.56993413e+00 -2.96316296e-01 -1.71670228e-01 -1.30126286e+00 1.00025952e+00 -1.33864701e+00 -2.80323207e-01 -1.46978050e-01 2.02722454e+00 1.11654758e+00 -1.05611004e-01 -1.20790713e-02 -6.49667621e-01 6.95475876e-01 7.05137327e-02 -1.41179726e-01 -1.08314252e+00 -4.64537948e-01 6.58510447e-01 5.31827152e-01 7.00807869e-01 -1.01274812e+00 1.59920025e+00 5.97269583e+00 4.44694042e-01 -8.96795273e-01 4.34753776e-01 2.36049429e-01 2.27229163e-01 -4.10029262e-01 1.07052743e-01 -1.01508391e+00 1.71516791e-01 1.52163064e+00 -1.70500636e-01 4.85005528e-02 3.47187340e-01 -1.78404316e-01 -2.43344121e-02 -8.89917552e-01 2.60628015e-01 2.50708669e-01 -7.65593410e-01 6.39183149e-02 -1.48967400e-01 6.19323969e-01 6.02975786e-01 -4.04232651e-01 4.81534421e-01 1.09086192e+00 -9.44655120e-01 4.54738110e-01 2.38339975e-01 1.00018179e+00 -1.12719393e+00 1.08072329e+00 1.22036815e-01 -1.10285819e+00 3.17443192e-01 -8.15512121e-01 2.31964700e-02 3.07380497e-01 3.70646566e-01 -5.71110368e-01 1.03831506e+00 7.70726025e-01 5.18128037e-01 -6.38345659e-01 1.90963283e-01 -6.73341036e-01 5.70770919e-01 -2.54611224e-01 -1.34124637e-01 2.57385939e-01 -2.09344029e-01 2.23623514e-01 1.84800243e+00 3.35953265e-01 -2.91824132e-01 2.85186589e-01 3.19446683e-01 1.74288183e-01 9.27172840e-01 -8.12592149e-01 -2.46299073e-01 1.54917628e-01 1.11919129e+00 -5.85689723e-01 -1.22739643e-01 -1.04501200e+00 1.01571476e+00 9.10565078e-01 1.46976084e-01 -6.19067609e-01 -6.16633415e-01 7.38686800e-01 -3.90321553e-01 3.05782408e-01 -2.58835644e-01 4.09473658e-01 -1.21174991e+00 -4.78181511e-01 -8.24770451e-01 4.78028744e-01 -4.76029634e-01 -1.49788964e+00 9.25825775e-01 7.77208284e-02 -2.44095385e-01 -2.36504570e-01 -7.78134704e-01 -2.86774814e-01 1.38079107e+00 -1.57281470e+00 -1.86828351e+00 5.80687404e-01 4.61185634e-01 4.80212361e-01 -6.41986549e-01 1.35755980e+00 4.31875169e-01 -8.27282429e-01 5.72523773e-01 4.96943370e-02 4.73796457e-01 1.06420875e+00 -1.16681361e+00 3.61806571e-01 1.00082719e+00 3.80077779e-01 6.01790130e-01 4.00394499e-01 -8.81060719e-01 -9.34582293e-01 -9.03509855e-01 1.46534646e+00 -1.79640025e-01 1.16277194e+00 -5.50603926e-01 -9.14448321e-01 1.09830618e+00 7.95011699e-01 -2.28487421e-02 1.22178304e+00 7.09025383e-01 -3.55545402e-01 1.97334588e-01 -9.89582300e-01 4.81760114e-01 7.12300241e-01 -6.78528249e-01 -8.45545471e-01 5.08741736e-01 8.97089005e-01 5.79541847e-02 -1.41626501e+00 8.29508305e-02 4.63427156e-01 -5.40737391e-01 6.46106958e-01 -1.03202152e+00 4.16731775e-01 -3.92132550e-02 -2.33180106e-01 -1.77761614e+00 -4.64055538e-01 -2.37906978e-01 8.07686090e-01 1.66251040e+00 8.77487659e-01 -9.28500116e-01 -1.91001192e-01 -1.24510616e-01 -2.26788476e-01 -6.89555258e-02 -9.09491658e-01 -7.98283100e-01 7.42206335e-01 -1.13272451e-01 3.08435559e-01 1.34368789e+00 2.11376902e-02 7.89769053e-01 -4.36344445e-01 2.82531172e-01 2.02963993e-01 -1.96303710e-01 3.10541689e-01 -1.12106097e+00 -3.49073708e-01 1.35044158e-01 -3.59580815e-01 4.42071073e-02 8.83100510e-01 -1.21719825e+00 2.41720118e-02 -1.30081785e+00 1.57491788e-01 -5.83924830e-01 -4.42457497e-01 6.67134523e-01 -3.62089932e-01 3.30692641e-02 1.54969797e-01 -3.29957843e-01 5.83869219e-02 2.03024983e-01 7.46731758e-01 3.02211672e-01 -1.25162318e-01 -5.33606231e-01 -6.84263825e-01 9.51727331e-01 6.68399572e-01 -6.51918888e-01 5.96079752e-02 -8.87923479e-01 2.73195505e-01 -3.45120989e-02 -4.77423966e-01 -5.63624263e-01 -1.12854756e-01 -2.42198795e-01 4.70148951e-01 -1.39904127e-01 -1.55083220e-02 -5.35214305e-01 1.55134909e-02 5.76276243e-01 -7.78303891e-02 4.87232059e-01 4.72558409e-01 9.68666971e-02 -2.33458176e-01 -4.62166280e-01 8.14656675e-01 -4.14430052e-01 -1.03595972e+00 1.08182188e-02 -4.37652916e-01 1.85225427e-01 1.15585291e+00 3.40341240e-01 -2.91291595e-01 2.85464436e-01 -7.94295788e-01 1.51050448e-01 3.95543128e-01 7.85218596e-01 1.30043209e-01 -1.13444877e+00 -1.39236915e+00 4.32442278e-01 1.90809399e-01 -6.63585365e-01 -1.66076198e-01 3.85020852e-01 -8.27118039e-01 4.96196240e-01 -7.73929834e-01 6.62189862e-03 -1.23791838e+00 6.31008327e-01 3.66798453e-02 -4.80126113e-01 -6.45235851e-02 9.45368469e-01 -1.46845197e-02 -1.15981305e+00 -2.97271103e-01 -1.74145177e-01 -4.67361718e-01 3.17010790e-01 1.69253889e-02 1.07485004e-01 -9.56452414e-02 -1.34711576e+00 -7.14239061e-01 6.35429978e-01 -2.45335639e-01 -2.69271255e-01 1.50696623e+00 3.60618271e-02 -4.72581148e-01 8.34796846e-01 1.09515536e+00 7.39343107e-01 -3.12005460e-01 -9.29961726e-02 4.87070441e-01 -1.29152551e-01 -1.98703885e-01 -8.93537343e-01 -7.53159463e-01 7.32412815e-01 4.07363415e-01 -4.17047232e-01 7.82013178e-01 1.38399646e-01 4.25563753e-01 1.49321884e-01 5.46496749e-01 -1.29244018e+00 -3.54346067e-01 1.07907081e+00 7.26052284e-01 -1.11097634e+00 -5.03981896e-02 -1.93190217e-01 -8.29824388e-01 9.74584162e-01 5.94617665e-01 -9.73086134e-02 7.96032310e-01 4.21196610e-01 2.89929599e-01 -6.88154995e-02 -7.62221217e-01 -6.33909822e-01 1.35127217e-01 9.24508452e-01 1.26206315e+00 2.86662877e-01 -7.76381850e-01 5.51429331e-01 -1.98628798e-01 -4.35847312e-01 5.16785562e-01 7.96825767e-01 -2.06642732e-01 -1.64392793e+00 -3.54491323e-01 2.69265682e-01 -1.02568519e+00 -4.94414598e-01 -4.98595536e-01 1.45225775e+00 6.64797544e-01 7.64916778e-01 2.68232614e-01 -1.15700983e-01 2.18675628e-01 3.11701834e-01 4.83706385e-01 -1.02071762e+00 -9.27937388e-01 2.43695691e-01 7.72375524e-01 3.35628688e-02 -6.21487677e-01 -8.20575058e-01 -9.97420013e-01 -1.05049416e-01 -3.38491470e-01 2.09933549e-01 5.26685894e-01 9.10244584e-01 -2.69084483e-01 5.03788292e-01 2.35534132e-01 -6.10766411e-01 8.16528946e-02 -1.30106008e+00 -6.76510632e-01 2.04871759e-01 -2.55405083e-02 -4.06524390e-01 -4.73415330e-02 -7.12621212e-02]
[10.530600547790527, 9.953643798828125]
69348a1a-71be-4049-ac7d-00c2dee6b4ef
heat-hyperedge-attention-networks
2201.12113
null
https://arxiv.org/abs/2201.12113v2
https://arxiv.org/pdf/2201.12113v2.pdf
HEAT: Hyperedge Attention Networks
Learning from structured data is a core machine learning task. Commonly, such data is represented as graphs, which normally only consider (typed) binary relationships between pairs of nodes. This is a substantial limitation for many domains with highly-structured data. One important such domain is source code, where hypergraph-based representations can better capture the semantically rich and structured nature of code. In this work, we present HEAT, a neural model capable of representing typed and qualified hypergraphs, where each hyperedge explicitly qualifies how participating nodes contribute. It can be viewed as a generalization of both message passing neural networks and Transformers. We evaluate HEAT on knowledge base completion and on bug detection and repair using a novel hypergraph representation of programs. In both settings, it outperforms strong baselines, indicating its power and generality.
['Miltiadis Allamanis', 'Marc Brockschmidt', 'Dobrik Georgiev']
2022-01-28
null
null
null
null
['knowledge-base-completion', 'knowledge-base-completion']
['graphs', 'knowledge-base']
[ 2.14132648e-02 5.18064260e-01 -7.19837546e-01 -3.36094648e-01 -3.82533491e-01 -5.03481805e-01 3.98043394e-01 6.60963714e-01 5.96233644e-02 4.98682380e-01 5.02577960e-01 -6.30191028e-01 -1.67174846e-01 -1.34477663e+00 -1.02458978e+00 -1.61706552e-01 -4.84355211e-01 3.39224428e-01 3.33086312e-01 -3.27308238e-01 -1.22677907e-01 1.14331497e-02 -1.35869169e+00 5.99507928e-01 6.84691191e-01 6.19945228e-01 -1.96170975e-02 3.93673033e-01 -2.63190329e-01 1.34832215e+00 -5.68726897e-01 -8.37569416e-01 -1.47393048e-01 9.03758034e-02 -1.12943482e+00 -4.57213402e-01 5.51891804e-01 -1.23732485e-01 -5.54813921e-01 1.19553697e+00 -4.66526113e-02 -2.01384738e-01 4.01180983e-01 -1.49133277e+00 -9.25813079e-01 1.22793245e+00 -4.43552375e-01 1.80538446e-01 5.22221148e-01 6.65971860e-02 1.54739869e+00 -5.51550329e-01 6.64601624e-01 1.33000255e+00 1.00249660e+00 3.32435638e-01 -1.46394825e+00 -2.99552470e-01 8.94587338e-02 1.59836128e-01 -1.02800965e+00 -3.87070924e-02 5.10856450e-01 -6.23978436e-01 1.30845523e+00 1.76247150e-01 4.89774734e-01 9.47186112e-01 1.26146466e-01 6.40616953e-01 5.34502208e-01 -2.91955531e-01 9.30882767e-02 -1.85958609e-01 8.83836508e-01 1.20477498e+00 5.96574843e-01 2.39112377e-02 -3.59449327e-01 -6.44130349e-01 4.04425532e-01 2.30753914e-01 -3.94572169e-01 -6.72124803e-01 -1.02679014e+00 9.54499960e-01 9.02544379e-01 1.29809871e-01 -6.38250262e-02 7.80912876e-01 7.59994984e-01 5.14829874e-01 2.59414911e-01 6.63732409e-01 -4.50505108e-01 -5.05662374e-02 -3.97784442e-01 1.25899032e-01 1.14461219e+00 1.12552738e+00 1.01605690e+00 1.99086443e-01 -3.12350303e-01 7.16380954e-01 2.83897728e-01 2.33697817e-01 2.89178729e-01 -5.33455729e-01 6.84339821e-01 1.36320841e+00 -4.87221152e-01 -1.01243901e+00 -3.49312901e-01 -4.66107368e-01 -7.86703050e-01 7.41047610e-04 1.81225985e-01 1.50678888e-01 -9.91843522e-01 1.78656518e+00 -1.76963508e-02 2.49038830e-01 -1.57113135e-01 5.36531448e-01 1.34627283e+00 6.04272127e-01 1.00634776e-01 3.24222505e-01 1.28188944e+00 -8.24565113e-01 -4.50918138e-01 -3.89974236e-01 9.12913561e-01 1.12973405e-02 1.14370298e+00 2.17658475e-01 -1.06995523e+00 -2.07951367e-01 -9.29919660e-01 -2.25217298e-01 -5.85497439e-01 -2.58199900e-01 1.21213484e+00 7.01367378e-01 -1.36263311e+00 6.30357504e-01 -8.74204278e-01 -1.00729600e-01 4.41328406e-01 3.54450166e-01 -5.76727509e-01 -4.49597210e-01 -1.23912954e+00 5.60782075e-01 4.72402066e-01 -1.16939098e-01 -1.20658803e+00 -9.32274163e-01 -1.41236854e+00 5.91514170e-01 5.37044764e-01 -7.24812806e-01 1.24978304e+00 -4.49019074e-01 -5.71016371e-01 8.72779727e-01 -9.82671324e-03 -5.24010241e-01 -3.29671323e-01 1.16277330e-01 -2.61418879e-01 -2.47557133e-01 -1.41711920e-01 1.21314839e-01 5.62381983e-01 -1.23413384e+00 -2.70788193e-01 -4.96037573e-01 7.41433620e-01 -4.26206470e-01 -5.59675157e-01 3.91716957e-02 -4.32246327e-01 -4.06443030e-01 -2.01497674e-01 -6.82936370e-01 -2.10253164e-01 -3.14437866e-01 -6.51309848e-01 -4.00080085e-01 5.63010514e-01 -6.46539450e-01 1.63396859e+00 -2.06789637e+00 3.92085493e-01 2.11924776e-01 9.14407969e-01 5.95512986e-02 -3.12840253e-01 6.85152829e-01 -2.20097810e-01 4.13314253e-01 -3.79856169e-01 -3.57930176e-02 4.15327907e-01 3.46486926e-01 -2.87787527e-01 2.71369845e-01 3.02100867e-01 1.25605083e+00 -1.03771949e+00 -1.02147758e-01 -4.51147705e-01 4.18219477e-01 -7.37718880e-01 2.57116586e-01 -7.69927621e-01 -5.02292871e-01 -4.04797018e-01 7.43858516e-01 3.89743537e-01 -7.38912046e-01 4.13997024e-01 -3.74529399e-02 5.12884438e-01 4.77442414e-01 -8.03304434e-01 1.70708668e+00 -6.22949362e-01 6.33305609e-01 5.47002517e-02 -1.01380897e+00 8.74530852e-01 1.94393381e-01 6.60835877e-02 -5.23037970e-01 -2.87122339e-01 -2.06336081e-01 1.67660601e-02 -6.52641594e-01 6.94312572e-01 2.62937725e-01 -3.64217162e-01 5.80264509e-01 1.38808548e-01 4.02998850e-02 3.38796586e-01 7.73659766e-01 1.79548895e+00 -9.40133408e-02 2.88672566e-01 -1.68109179e-01 8.92255381e-02 -1.34669347e-02 6.66598737e-01 7.29249179e-01 3.10916543e-01 2.70636588e-01 1.31781316e+00 -5.72494924e-01 -7.87382901e-01 -8.31409991e-01 1.75124526e-01 1.30875731e+00 -1.24878496e-01 -1.10429871e+00 -5.11803925e-01 -9.77609694e-01 3.20698738e-01 3.30114990e-01 -7.78008759e-01 -5.66065252e-01 -5.70732892e-01 -6.83339179e-01 6.84110224e-01 7.10565388e-01 -1.55291915e-01 -1.01937008e+00 -3.32787812e-01 1.68532878e-02 -9.39318016e-02 -7.61375308e-01 -7.43574277e-02 4.72154647e-01 -8.82563651e-01 -1.59621942e+00 -1.33905560e-01 -9.62318718e-01 6.86100781e-01 1.00175031e-01 1.99284565e+00 8.34563434e-01 -2.40041986e-01 2.58863807e-01 -4.78073746e-01 -6.71244636e-02 -6.70183301e-01 2.68652588e-01 -4.66314048e-01 -4.40378815e-01 2.36139953e-01 -6.61891580e-01 -1.08293377e-01 9.48173702e-02 -1.17311943e+00 -1.56372935e-01 4.13862437e-01 1.00064564e+00 1.99118719e-01 2.66604394e-01 3.74946773e-01 -1.59904540e+00 7.52854526e-01 -9.60854828e-01 -6.75129712e-01 4.56320882e-01 -4.43188459e-01 4.11610693e-01 8.04295599e-01 -6.18310608e-02 -7.94381678e-01 -1.95154920e-01 2.05403566e-03 -1.58151492e-01 -8.46134797e-02 1.29393363e+00 -3.24417681e-01 -1.28037140e-01 9.81108308e-01 -1.22385897e-01 -2.08361283e-01 -4.91357297e-01 4.25655603e-01 3.34083378e-01 5.69071352e-01 -1.05705893e+00 7.81354010e-01 1.41033530e-01 1.08892702e-01 -3.19699973e-01 -5.01794934e-01 -2.09562421e-01 -3.03580523e-01 2.20396638e-01 3.92101645e-01 -7.38260150e-01 -7.91962028e-01 2.02076897e-01 -1.23910594e+00 -6.39093876e-01 -7.40534961e-02 5.40463887e-02 -9.83953699e-02 4.02546406e-01 -9.63152409e-01 -3.01083684e-01 -1.67134762e-01 -1.20046067e+00 1.02571917e+00 -1.23908170e-01 -6.28827838e-03 -1.30116272e+00 2.14676663e-01 4.34996597e-02 4.94661212e-01 3.86800587e-01 1.82194042e+00 -6.90528572e-01 -7.54600286e-01 -3.09574366e-01 -3.63086969e-01 1.21467777e-01 -2.00092405e-01 -3.61019224e-02 -7.50424862e-01 -4.72869217e-01 -5.72659135e-01 -7.35954821e-01 1.07723510e+00 3.85382515e-03 1.29063892e+00 -4.99422014e-01 -5.13378322e-01 6.78898633e-01 1.55625582e+00 -2.05401242e-01 6.39000297e-01 1.81366086e-01 9.74143505e-01 4.46085006e-01 -1.16010010e-01 4.05502081e-01 7.05690444e-01 3.17646235e-01 1.08210230e+00 -9.28847119e-02 -1.67808697e-01 -4.40774858e-01 1.93470091e-01 7.38103509e-01 2.03312248e-01 -4.14966524e-01 -1.35224605e+00 6.57029569e-01 -1.87241924e+00 -7.80465543e-01 -4.73298520e-01 2.13010955e+00 9.43234205e-01 -1.54960947e-02 7.72537058e-03 5.40480204e-02 7.06478477e-01 2.45349795e-01 -4.12576646e-01 -3.20438325e-01 3.49436626e-02 3.63933057e-01 2.66319662e-01 3.21003675e-01 -8.16044867e-01 6.06151700e-01 6.41765022e+00 4.09274817e-01 -6.14217997e-01 9.50028151e-02 2.84386426e-01 3.25792789e-01 -7.41108418e-01 9.90474150e-02 -5.41775346e-01 2.85495609e-01 9.70394552e-01 -3.23295653e-01 5.55210650e-01 1.00704265e+00 -6.39468133e-01 2.05476269e-01 -1.63211513e+00 5.45950413e-01 1.42233251e-02 -1.46054816e+00 -4.57145348e-02 1.79891549e-02 7.26586401e-01 1.01176694e-01 8.43091756e-02 8.98475766e-01 9.83676791e-01 -1.43988073e+00 3.50149333e-01 2.67793685e-01 7.91392267e-01 -4.90517229e-01 6.15057051e-01 1.82895780e-01 -1.28296065e+00 -2.83402771e-01 -5.50537467e-01 -8.57473835e-02 -3.16447049e-01 6.81567788e-01 -8.17476630e-01 6.56188428e-01 6.58645272e-01 9.64478314e-01 -9.10077393e-01 1.29889035e+00 -4.70406860e-01 8.40986490e-01 3.12804505e-02 6.54452443e-02 1.12975873e-01 2.37662226e-01 1.98633552e-01 1.37282479e+00 1.21077470e-01 -2.30791941e-01 3.17055315e-01 1.16089356e+00 -6.27962232e-01 -4.20800716e-01 -9.37911510e-01 -2.13594511e-01 5.28155267e-01 1.29307103e+00 -3.54485333e-01 -3.50772649e-01 -7.93922544e-01 2.28463918e-01 8.48792672e-01 5.45283794e-01 -7.48547196e-01 -5.63536942e-01 5.97608387e-01 8.16311985e-02 6.29651472e-02 -7.15392223e-03 -1.20978676e-01 -1.15772903e+00 1.34417310e-01 -1.03985596e+00 8.18803549e-01 -8.53886902e-01 -1.26720142e+00 6.77822649e-01 -4.99868803e-02 -7.45427847e-01 -2.11387038e-01 -7.83440351e-01 -6.72208130e-01 7.76811600e-01 -1.42267692e+00 -1.18628824e+00 -4.56947088e-01 5.92579246e-01 -3.51167358e-02 -1.00753196e-01 8.78415406e-01 2.42557570e-01 -5.13082981e-01 7.31535375e-01 -1.44205287e-01 3.56953740e-01 2.62029439e-01 -1.63322973e+00 7.62001872e-01 8.75811756e-01 2.15157822e-01 1.00933611e+00 3.84480596e-01 -7.44243085e-01 -1.97365487e+00 -1.43205476e+00 7.20356882e-01 -5.87800562e-01 9.02590990e-01 -5.15894651e-01 -1.43765211e+00 1.04340494e+00 2.19704717e-01 3.66234809e-01 4.44604188e-01 6.35697246e-01 -1.03322530e+00 1.75350070e-01 -8.46671104e-01 2.20013797e-01 1.22603130e+00 -8.87221277e-01 -5.88584363e-01 4.97778803e-01 1.07773209e+00 -4.34338242e-01 -1.16719115e+00 3.55028033e-01 8.04620534e-02 -9.35878336e-01 9.90509570e-01 -1.10686719e+00 1.02711332e+00 -1.07160874e-01 -1.93331838e-02 -1.66251659e+00 -5.83028436e-01 -6.06513321e-01 -6.13398552e-01 1.08388591e+00 6.40506685e-01 -3.95681828e-01 9.58160102e-01 6.71123445e-01 -3.79893243e-01 -5.72089612e-01 -4.99570787e-01 -7.92331636e-01 3.45118940e-02 -3.67117971e-01 1.06273472e+00 1.23852873e+00 6.31395757e-01 3.67704779e-01 -1.32839784e-01 2.26974681e-01 5.50893605e-01 2.74738580e-01 5.90634167e-01 -1.56944323e+00 -5.57394922e-01 -5.34281373e-01 -6.17650986e-01 -6.26880467e-01 7.11824119e-01 -1.62498116e+00 -1.39538631e-01 -1.82742155e+00 5.15827656e-01 -5.15093744e-01 -1.36346772e-01 9.69060719e-01 -2.78673708e-01 -1.16984792e-01 -2.03703135e-01 -7.53014386e-02 -6.24018848e-01 2.93429375e-01 7.98382998e-01 -6.09242499e-01 1.79501459e-01 -1.80369228e-01 -6.59809351e-01 4.51278865e-01 6.59363627e-01 -5.31404257e-01 -5.17360151e-01 -9.17339206e-01 1.02197576e+00 3.96196246e-01 6.65891409e-01 -5.41935742e-01 2.82179505e-01 -1.30003719e-02 -2.64781326e-01 -1.99327804e-03 4.27294383e-03 -7.04077244e-01 2.41167888e-01 4.39880759e-01 -5.82411110e-01 3.65302294e-01 1.47644773e-01 7.80255854e-01 -4.34143335e-01 -3.62613171e-01 2.93168098e-01 -3.79633546e-01 -9.45015430e-01 4.98217255e-01 6.64366260e-02 3.76146287e-01 6.03881717e-01 4.22248617e-02 -1.09121048e+00 -1.87727913e-01 -4.15378630e-01 3.07065427e-01 6.05189085e-01 4.60657477e-01 6.84017062e-01 -1.30182564e+00 -4.89514679e-01 2.25736991e-01 5.58049977e-01 1.27318710e-01 9.07735806e-03 5.47162771e-01 -3.04151267e-01 3.05492580e-01 -3.47297527e-02 -2.61960149e-01 -1.19565642e+00 8.72794569e-01 2.54548758e-01 -4.87379700e-01 -7.15585053e-01 8.17003608e-01 3.97392362e-01 -6.62912607e-01 5.67147553e-01 -5.83184123e-01 -2.87220031e-01 -3.74744356e-01 4.20270473e-01 3.29849958e-01 3.18512946e-01 -2.13731885e-01 -2.83692539e-01 -3.74017768e-02 -4.01839204e-02 6.06914997e-01 1.48787344e+00 5.57749093e-01 -7.53218293e-01 1.49399117e-01 1.26787043e+00 -2.01362133e-01 -8.30935717e-01 -4.98831302e-01 3.88855040e-01 -4.65508014e-01 -1.16358651e-02 -8.86531115e-01 -1.38504052e+00 1.09829879e+00 -1.71740174e-01 5.78766167e-01 7.53900707e-01 2.19361559e-01 5.20774722e-01 6.52496636e-01 5.23725927e-01 -3.48467797e-01 7.68568963e-02 7.31171012e-01 7.08144128e-01 -1.08656716e+00 -2.18805268e-01 -5.06014168e-01 -2.33024225e-01 1.13371861e+00 8.92723560e-01 7.67261684e-02 3.93128514e-01 6.17789567e-01 -5.78549445e-01 -7.27964699e-01 -1.12496364e+00 -5.48763163e-02 3.01414937e-01 9.13127780e-01 6.57442808e-01 7.45505840e-02 1.44931838e-01 7.72680640e-01 5.50880432e-02 -1.00278370e-01 9.09309864e-01 9.40761447e-01 -2.94568002e-01 -1.06759477e+00 -1.81748122e-01 8.12805176e-01 -4.39345092e-01 -4.87864971e-01 -3.95796597e-01 7.04071879e-01 -2.50735670e-01 7.63196886e-01 -2.39177138e-01 -4.77485597e-01 2.99559265e-01 -1.25677615e-01 3.20458978e-01 -1.13768291e+00 -9.65474606e-01 -8.54386389e-01 2.91886419e-01 -6.20366812e-01 5.95925897e-02 6.05682917e-02 -1.12631428e+00 -3.87202501e-01 -2.00599715e-01 2.63726354e-01 4.39715505e-01 3.19421142e-01 3.46348077e-01 8.44504654e-01 3.06037068e-01 -3.49803805e-01 -7.33562529e-01 -6.60445035e-01 -5.95832705e-01 6.73989177e-01 4.28803533e-01 -5.69607913e-01 -1.77660272e-01 -6.86138719e-02]
[7.485530853271484, 7.83091402053833]
c4ac3d7b-e809-442f-9a85-1131d53d6dce
a-demographic-attribute-guided-approach-to
2205.10254
null
https://arxiv.org/abs/2205.10254v1
https://arxiv.org/pdf/2205.10254v1.pdf
A Demographic Attribute Guided Approach to Age Estimation
Face-based age estimation has attracted enormous attention due to wide applications to public security surveillance, human-computer interaction, etc. With vigorous development of deep learning, age estimation based on deep neural network has become the mainstream practice. However, seeking a more suitable problem paradigm for age change characteristics, designing the corresponding loss function and designing a more effective feature extraction module still needs to be studied. What is more, change of face age is also related to demographic attributes such as ethnicity and gender, and the dynamics of different age groups is also quite different. This problem has so far not been paid enough attention to. How to use demographic attribute information to improve the performance of age estimation remains to be further explored. In light of these issues, this research makes full use of auxiliary information of face attributes and proposes a new age estimation approach with an attribute guidance module. We first design a multi-scale attention residual convolution unit (MARCU) to extract robust facial features other than simply using other standard feature modules such as VGG and ResNet. Then, after being especially treated through full connection (FC) layers, the facial demographic attributes are weight-summed by 1*1 convolutional layer and eventually merged with the age features by a global FC layer. Lastly, we propose a new error compression ranking (ECR) loss to better converge the age regression value. Experimental results on three public datasets of UTKFace, LAP2016 and Morph show that our proposed approach achieves superior performance compared to other state-of-the-art methods.
['Heng Zhao', 'Liaojun Pang', 'Kaituo Zhang', 'Zhicheng Cao']
2022-05-20
null
null
null
null
['age-estimation', 'age-estimation']
['computer-vision', 'miscellaneous']
[-7.43045360e-02 -5.37850931e-02 -7.30310893e-03 -7.86773503e-01 -1.26444608e-01 2.40601182e-01 5.25767684e-01 3.69965397e-02 -5.63800454e-01 6.73626661e-01 2.91752905e-01 1.31910369e-01 -1.67950556e-01 -1.00797236e+00 -4.41154957e-01 -8.58190656e-01 -1.82786599e-01 1.24518268e-01 -2.38553420e-01 -1.19580038e-01 1.68528944e-01 2.92723656e-01 -1.81155741e+00 -3.22542161e-01 1.18613720e+00 1.48905623e+00 -2.43462533e-01 1.57408580e-01 -2.11457595e-01 4.67335194e-01 -3.64924639e-01 -6.52885139e-01 1.17244028e-01 -3.67632002e-01 -6.40508235e-01 -1.21899307e-01 4.34091330e-01 -5.62044799e-01 -3.87590945e-01 1.00519669e+00 8.76465023e-01 7.27431104e-02 6.89989090e-01 -1.31457639e+00 -8.53195548e-01 5.77399790e-01 -9.47110176e-01 1.44929618e-01 1.91203013e-01 -8.25031251e-02 5.51010787e-01 -7.66523898e-01 1.47494286e-01 1.57157600e+00 7.34935164e-01 8.24426591e-01 -5.76611042e-01 -1.21657193e+00 3.14058930e-01 4.69210744e-01 -1.44047451e+00 -3.43479842e-01 8.00261140e-01 -1.31507337e-01 2.61724055e-01 9.01209190e-03 6.12806857e-01 1.25360024e+00 -4.93872724e-02 6.78118765e-01 1.04144132e+00 -5.91398589e-03 -1.42549902e-01 2.32640300e-02 9.88366231e-02 9.72708344e-01 2.14863762e-01 -2.05789357e-01 -3.14357251e-01 5.48181124e-02 5.49196362e-01 2.43161112e-01 -2.32257053e-01 1.10914961e-01 -7.46778667e-01 7.14283109e-01 7.15456307e-01 2.67112225e-01 -4.39129651e-01 2.18424320e-01 3.34212422e-01 2.63149351e-01 9.26600754e-01 1.31150503e-02 -6.10427320e-01 -1.19483523e-01 -7.58549511e-01 2.08845809e-01 3.19252789e-01 6.05397880e-01 7.58103013e-01 -2.28916481e-03 -1.21880747e-01 9.86908734e-01 2.49223113e-01 4.67265725e-01 5.22997320e-01 -6.71639979e-01 3.53571504e-01 9.70321119e-01 -3.93867075e-01 -1.04100752e+00 -5.51724792e-01 -4.97444481e-01 -1.22742140e+00 1.58318598e-02 5.37587225e-01 -3.48586291e-01 -9.47116733e-01 1.94997883e+00 4.61888611e-01 4.24831539e-01 -2.19409779e-01 7.89629340e-01 9.45864022e-01 5.27212620e-01 4.93543535e-01 -4.71096337e-01 1.66349077e+00 -3.42913419e-01 -6.47135735e-01 -4.30631973e-02 4.18711156e-01 -4.41210866e-01 6.47615194e-01 6.84520826e-02 -9.77064908e-01 -6.15470052e-01 -7.93656409e-01 -1.58923268e-02 -4.78456140e-01 1.98821023e-01 1.07055759e+00 6.80022776e-01 -9.84946311e-01 6.52142763e-01 -7.17744410e-01 -6.84155285e-01 9.80440915e-01 7.47840822e-01 -3.97433490e-01 -7.36725330e-02 -1.44587815e+00 6.51427627e-01 2.73183167e-01 3.40250611e-01 -3.11569214e-01 -9.08510864e-01 -9.49012160e-01 6.08378984e-02 2.16871724e-01 -9.62920070e-01 8.87012899e-01 -1.23362625e+00 -1.37535727e+00 7.12581635e-01 -7.54532665e-02 -1.88405260e-01 3.24433446e-01 -3.45035106e-01 -5.54790497e-01 -7.95157626e-02 -1.41253933e-01 7.06961334e-01 9.98066962e-01 -8.13077569e-01 -7.72295594e-01 -1.06491363e+00 -6.14173636e-02 2.87678212e-01 -1.12127721e+00 3.69964600e-01 -4.73504394e-01 -5.92748702e-01 -2.99357041e-03 -6.94589794e-01 -1.42925456e-01 1.44396186e-01 1.27225474e-01 -7.19803989e-01 8.18629980e-01 -1.04631042e+00 1.68137801e+00 -1.98195171e+00 1.81271031e-01 1.17367595e-01 3.79190654e-01 2.58308321e-01 1.76049825e-02 -1.65803164e-01 -1.23746514e-01 5.04080243e-02 -2.75964558e-01 -3.89742464e-01 -3.21962029e-01 -2.67222941e-01 2.83873528e-01 4.03823167e-01 3.62711251e-01 7.49819517e-01 -7.15260506e-01 -7.52207279e-01 -5.50598241e-02 7.68204093e-01 -6.18946850e-01 1.99906185e-01 2.56977737e-01 3.70558113e-01 -8.22189152e-01 9.13842797e-01 1.02459002e+00 1.73489347e-01 -4.28839356e-01 -2.47025877e-01 -1.65556150e-03 -3.91358912e-01 -9.78275657e-01 1.51389742e+00 -4.06859517e-01 8.72423276e-02 7.37462714e-02 -9.98258591e-01 1.23936641e+00 1.12347826e-01 7.11081386e-01 -5.63051581e-01 5.04021943e-01 1.75123587e-01 4.05900739e-02 -4.71887261e-01 2.15857044e-01 1.19666807e-01 -1.96119677e-02 1.45860225e-01 6.03491738e-02 4.40801561e-01 -1.38380174e-02 3.90046090e-02 8.62348616e-01 1.09328501e-01 8.00016299e-02 -1.92097023e-01 1.03089190e+00 -7.50751913e-01 8.14677954e-01 5.58273047e-02 -4.32263702e-01 4.25134480e-01 3.38073254e-01 -6.59078181e-01 -8.48024130e-01 -6.98381186e-01 -3.00819755e-01 1.12844241e+00 4.00302596e-02 -3.46683353e-01 -1.02283132e+00 -9.31043267e-01 1.68567047e-01 6.31722361e-02 -1.05729151e+00 -4.82432425e-01 -7.19051719e-01 -1.07920337e+00 4.26808357e-01 6.38751864e-01 1.09557498e+00 -1.33549082e+00 -1.12597637e-01 7.60315731e-02 1.01155154e-01 -8.08218002e-01 -4.70257103e-01 -3.71668577e-01 -9.33569193e-01 -9.81886387e-01 -1.13297868e+00 -7.64774799e-01 8.63615453e-01 -1.85021982e-01 7.77152717e-01 3.47117037e-01 -1.63391978e-01 1.83437556e-01 -3.45505059e-01 -5.80581427e-01 5.93642816e-02 5.45728028e-01 3.00197780e-01 4.19230521e-01 6.65607989e-01 -8.28552246e-01 -9.59379017e-01 -2.79723620e-03 -6.30836010e-01 -2.81704396e-01 7.47006536e-01 5.79641581e-01 -9.03590769e-02 1.67194959e-02 9.39534247e-01 -7.81600118e-01 5.49603224e-01 -5.20049334e-01 -2.67494500e-01 1.90618366e-01 -7.50801504e-01 1.99302174e-02 6.70721412e-01 -4.13649082e-01 -1.37637103e+00 -1.14396781e-01 -3.51068556e-01 -2.23011777e-01 -6.41281828e-02 3.03283006e-01 -5.05673587e-01 1.31157666e-01 1.75224036e-01 1.06249787e-02 3.15562636e-01 -4.88300979e-01 7.64069483e-02 9.57247376e-01 4.55616593e-01 -5.50237715e-01 9.70124304e-01 2.27733195e-01 7.12499842e-02 -7.46699750e-01 -7.55889177e-01 -1.14081606e-01 -4.33066010e-01 -2.93529302e-01 8.47486079e-01 -9.23491836e-01 -1.06964672e+00 1.10066652e+00 -8.98768306e-01 3.15919757e-01 3.16351712e-01 4.90897954e-01 -9.30024311e-02 4.32255954e-01 -6.58112526e-01 -8.64870846e-01 -9.31760907e-01 -9.33914244e-01 9.74228084e-01 8.61313403e-01 8.82611200e-02 -7.57396996e-01 -3.48711491e-01 3.30522716e-01 6.40111983e-01 2.38781512e-01 7.40896225e-01 -3.38133484e-01 -2.25200340e-01 -1.78740934e-01 -5.10410428e-01 3.14479381e-01 2.14256167e-01 6.63393736e-02 -1.01501620e+00 -3.21523160e-01 -2.80702949e-01 -2.32059345e-01 1.18790364e+00 5.02115726e-01 1.82260001e+00 -2.89211124e-01 -3.04353535e-01 9.50755239e-01 1.21406972e+00 1.47940159e-01 8.15867960e-01 2.75402218e-01 9.06835914e-01 7.64119625e-01 5.50672531e-01 6.88682735e-01 5.96019566e-01 2.91830838e-01 4.70792830e-01 -6.68103248e-02 1.43331662e-01 -2.46246438e-02 1.99921072e-01 6.96209908e-01 -6.61846519e-01 2.80788302e-01 -5.71832240e-01 3.06951612e-01 -1.60113513e+00 -9.47275937e-01 1.90077588e-01 2.11018467e+00 8.96075726e-01 -1.16982736e-01 1.93622842e-01 1.44771233e-01 8.88419569e-01 2.80981064e-01 -6.55824602e-01 -3.69366586e-01 -7.21033721e-05 4.23866153e-01 4.06975150e-01 -1.59919664e-01 -1.29983556e+00 7.25363791e-01 4.26544428e+00 9.44858015e-01 -1.10264742e+00 -1.32720396e-01 1.14229214e+00 2.52630711e-01 2.05651876e-02 -3.01652640e-01 -9.83148754e-01 6.97609246e-01 6.96078420e-01 -4.67500016e-02 5.13787806e-01 8.50927174e-01 3.69585976e-02 1.06502853e-01 -7.62057900e-01 1.28744149e+00 1.75272182e-01 -6.33828998e-01 -1.38552770e-01 1.36308596e-01 4.97566044e-01 -6.39093816e-01 2.83767760e-01 7.03783751e-01 -1.96443439e-01 -1.07018483e+00 3.43853384e-01 6.61815882e-01 8.82251680e-01 -1.14063609e+00 1.00824416e+00 -3.94036546e-02 -1.43829954e+00 -4.87791628e-01 -5.02422392e-01 -2.73434054e-02 -2.53593862e-01 5.80880523e-01 -2.30247915e-01 6.95683897e-01 1.04477954e+00 8.42579722e-01 -9.85028982e-01 9.01677251e-01 1.12600572e-01 4.81071651e-01 -1.04952440e-01 2.82014720e-02 -2.98776533e-02 -2.37459764e-01 2.80403569e-02 7.13900268e-01 6.67734265e-01 3.51771355e-01 -2.31077641e-01 4.65742677e-01 -4.32133079e-01 4.16594803e-01 -3.61445665e-01 4.20065708e-02 4.21276391e-01 1.65063977e+00 -4.10832226e-01 -1.21924445e-01 -4.99399543e-01 8.61067772e-01 4.54461753e-01 1.02401771e-01 -7.42780089e-01 -4.81567979e-01 9.74961281e-01 2.40467429e-01 7.46245757e-02 1.66893452e-01 6.59712479e-02 -1.03391325e+00 -4.43992801e-02 -7.19223440e-01 5.79372466e-01 -3.00262809e-01 -1.41421294e+00 4.77456957e-01 -1.08295128e-01 -9.18223441e-01 3.29554565e-02 -4.27137733e-01 -8.22201192e-01 9.13666487e-01 -1.44655240e+00 -1.46085107e+00 -7.19484210e-01 6.51629746e-01 3.53911519e-01 -3.21644127e-01 5.38949668e-01 5.64250886e-01 -1.01459897e+00 1.02886188e+00 -2.61316031e-01 3.34654242e-01 9.27641213e-01 -1.07126260e+00 2.07596317e-01 5.19582868e-01 -7.07921505e-01 7.08286643e-01 3.27545553e-01 -5.98273337e-01 -1.17966020e+00 -1.24518287e+00 9.21438217e-01 -2.45768398e-01 3.11743379e-01 -2.47721791e-01 -1.02350712e+00 2.74027526e-01 1.78535581e-01 1.15650415e-01 2.53830642e-01 3.00515801e-01 -1.98438242e-01 -6.91530526e-01 -1.19144666e+00 4.94996697e-01 1.31257927e+00 -1.65051743e-01 -2.60409921e-01 -1.79712579e-01 6.15257442e-01 2.57251319e-04 -1.00465333e+00 9.33058321e-01 7.65203416e-01 -8.53791356e-01 9.16914046e-01 -5.06800354e-01 5.71065605e-01 -4.43445519e-02 4.42793369e-01 -1.14878583e+00 -2.78643638e-01 -4.28799510e-01 -1.68769047e-01 1.89522076e+00 -2.53917486e-03 -6.99357748e-01 1.01544273e+00 7.32787669e-01 1.32550187e-02 -1.32256973e+00 -6.80220783e-01 -3.53333265e-01 5.79875745e-02 4.71255817e-02 1.10622466e+00 9.47987318e-01 -4.35564220e-01 3.25756103e-01 -4.06664878e-01 -2.14744098e-02 6.49856627e-01 -2.51970023e-01 6.76866591e-01 -1.62628078e+00 4.16022897e-01 -6.40660167e-01 -5.66608429e-01 -3.74938458e-01 3.72038692e-01 -5.99753976e-01 -3.78546178e-01 -1.42806971e+00 2.53241867e-01 -5.55495620e-01 -4.58620101e-01 3.48354936e-01 -7.55520761e-01 2.43610248e-01 -1.24176420e-01 -2.84009159e-01 -4.36547339e-01 1.01910472e+00 1.29617751e+00 -1.27322018e-01 -3.76645625e-02 2.22228602e-01 -9.05740261e-01 7.57978618e-01 9.37077045e-01 -8.54703337e-02 -2.84927696e-01 -1.22816227e-01 5.04795238e-02 -1.75061211e-01 -2.98635103e-02 -1.17838383e+00 1.33260369e-01 -4.23064567e-02 9.68648195e-01 -4.37469870e-01 1.38882846e-01 -7.53580987e-01 -1.66474774e-01 4.37558949e-01 2.11052317e-02 2.50468165e-01 -6.37704656e-02 3.73280883e-01 -9.01530534e-02 -1.37553387e-03 7.47109115e-01 -9.24578831e-02 -8.30037177e-01 1.15430701e+00 1.32549226e-01 -1.03293307e-01 1.09763575e+00 -3.73998672e-01 -2.19402596e-01 -1.81435242e-01 -3.96714866e-01 5.07282972e-01 3.08751076e-01 6.56813383e-01 6.12141252e-01 -1.51730502e+00 -9.40554738e-01 2.40487814e-01 1.19726002e-01 -1.18929066e-01 4.99430060e-01 7.78343976e-01 -2.05910146e-01 -3.01632173e-02 -3.78845423e-01 -2.40494862e-01 -1.34362638e+00 5.19302189e-01 1.93615600e-01 -2.25648269e-01 -1.89489067e-01 9.79661465e-01 2.19233364e-01 -3.94632846e-01 3.16085279e-01 8.56715739e-02 -7.90681005e-01 4.75337774e-01 6.10284090e-01 4.22180772e-01 -1.08922988e-01 -7.34936714e-01 -4.22379196e-01 9.01869893e-01 -1.57754526e-01 5.13210893e-01 1.44070899e+00 -1.46554783e-01 -2.61262923e-01 -9.71724615e-02 1.25744891e+00 -2.54292488e-01 -1.07670963e+00 -2.45689467e-01 -1.20217860e-01 -4.15666699e-01 -5.48839010e-02 -3.75143737e-01 -1.63708854e+00 7.78754294e-01 9.12325263e-01 -1.02063455e-01 1.60175931e+00 -9.43469927e-02 1.02863646e+00 2.55614193e-03 2.00291455e-01 -1.11318040e+00 6.35851473e-02 3.26261252e-01 6.84555888e-01 -1.45875776e+00 7.87155107e-02 -2.80380994e-01 -3.19548190e-01 1.03665352e+00 1.17738473e+00 6.86100647e-02 8.29866111e-01 -1.83078334e-01 -1.51626095e-01 3.31715681e-02 -2.99929142e-01 -3.60898197e-01 3.88349533e-01 4.63639349e-01 5.78078568e-01 -1.86344936e-01 -6.81248605e-01 9.48639631e-01 -2.73985803e-01 2.05013119e-02 -1.30640015e-01 5.52330792e-01 -4.05657053e-01 -1.23491573e+00 -2.58295417e-01 9.79250729e-01 -7.83055067e-01 -3.25511396e-02 1.44110126e-02 6.84695184e-01 5.00094116e-01 5.75313807e-01 2.29791522e-01 -5.12408018e-01 1.10132247e-01 4.97294702e-02 4.93918449e-01 -2.59976417e-01 -6.56694949e-01 -4.29640979e-01 -1.42808080e-01 -3.82369787e-01 -5.04097879e-01 -7.46040642e-01 -1.02913141e+00 -5.98689198e-01 -2.83791512e-01 1.45722385e-02 6.80575907e-01 7.63760746e-01 1.94366336e-01 4.35066134e-01 9.74433899e-01 -6.35598004e-01 -4.67301548e-01 -1.26862872e+00 -5.56695700e-01 5.66040516e-01 1.90612897e-02 -8.55905712e-01 -2.27533072e-01 -2.53827065e-01]
[13.56002426147461, 0.8558171987533569]
0633767a-2c00-49f4-8ee7-01607450300d
twitter-sentiment-analysis-via-bi-sense-emoji
1807.07961
null
http://arxiv.org/abs/1807.07961v2
http://arxiv.org/pdf/1807.07961v2.pdf
Twitter Sentiment Analysis via Bi-sense Emoji Embedding and Attention-based LSTM
Sentiment analysis on large-scale social media data is important to bridge the gaps between social media contents and real world activities including political election prediction, individual and public emotional status monitoring and analysis, and so on. Although textual sentiment analysis has been well studied based on platforms such as Twitter and Instagram, analysis of the role of extensive emoji uses in sentiment analysis remains light. In this paper, we propose a novel scheme for Twitter sentiment analysis with extra attention on emojis. We first learn bi-sense emoji embeddings under positive and negative sentimental tweets individually, and then train a sentiment classifier by attending on these bi-sense emoji embeddings with an attention-based long short-term memory network (LSTM). Our experiments show that the bi-sense embedding is effective for extracting sentiment-aware embeddings of emojis and outperforms the state-of-the-art models. We also visualize the attentions to show that the bi-sense emoji embedding provides better guidance on the attention mechanism to obtain a more robust understanding of the semantics and sentiments.
['Jianbo Yuan', 'Jiebo Luo', 'Yuxiao Chen', 'Quanzeng You']
2018-07-20
null
null
null
null
['twitter-sentiment-analysis']
['natural-language-processing']
[-3.29884619e-01 -2.25878581e-02 -2.21871197e-01 -6.65569901e-01 -2.00060800e-01 -2.06382826e-01 4.82485175e-01 4.25073653e-01 -6.37154698e-01 4.15764093e-01 6.17042065e-01 -3.36768150e-01 3.94144654e-01 -8.89874995e-01 -7.05651194e-02 -3.87547493e-01 1.75888062e-01 -3.56668979e-01 -1.29798442e-01 -7.87269235e-01 1.95569932e-01 4.56499532e-02 -1.62910259e+00 2.76897848e-01 6.36723280e-01 1.33701634e+00 -1.92452773e-01 8.26901317e-01 -5.67012191e-01 1.17243361e+00 -7.71982908e-01 -8.08342695e-01 -4.34498668e-01 -4.01610248e-02 -7.02632725e-01 -3.46359670e-01 -2.37238705e-02 1.66129380e-01 -1.60661444e-01 1.15201747e+00 7.24857509e-01 2.81795293e-01 4.64399070e-01 -8.90841246e-01 -1.17277086e+00 4.06537712e-01 -6.02050245e-01 4.66934174e-01 2.75222301e-01 -3.14307183e-01 1.05013967e+00 -9.53916490e-01 5.48989058e-01 1.06047797e+00 7.07324922e-01 3.72268647e-01 -5.73679090e-01 -5.62762201e-01 4.55606759e-01 2.49337479e-01 -8.89299691e-01 -1.61913782e-01 1.06643999e+00 -3.99543732e-01 7.98384786e-01 4.18071091e-01 7.21774459e-01 1.15790653e+00 3.39640051e-01 9.16782796e-01 1.14374900e+00 -4.45632756e-01 1.47093907e-01 7.53671110e-01 6.78477049e-01 6.62494421e-01 -2.66643375e-01 -6.41603887e-01 -5.37395000e-01 -2.70441979e-01 -1.49855524e-01 3.23418051e-01 1.67944044e-01 2.54840970e-01 -6.45005584e-01 1.14774275e+00 5.02794445e-01 7.09132612e-01 -4.98782963e-01 -2.33388189e-02 9.22658384e-01 5.21157563e-01 1.25128090e+00 4.99575704e-01 -8.01353931e-01 -4.29425031e-01 -5.31797230e-01 -3.55690271e-01 8.66726458e-01 2.45686084e-01 8.64222884e-01 1.56113654e-01 -8.46192054e-03 1.17883849e+00 1.99419290e-01 5.58682501e-01 7.38912284e-01 -1.91039339e-01 1.79085195e-01 9.44592357e-01 2.63412818e-02 -1.67889631e+00 -6.43983483e-01 -3.13286036e-01 -5.61744392e-01 -2.80090630e-01 -1.22184470e-01 -7.32923210e-01 -4.24701273e-01 1.69748759e+00 4.75961208e-01 3.23454775e-02 6.58983663e-02 6.86078906e-01 1.39380324e+00 5.03294528e-01 4.32218134e-01 1.78390294e-01 1.84451377e+00 -9.29484129e-01 -1.11018121e+00 -4.84769046e-01 1.03911030e+00 -6.09231651e-01 1.19918668e+00 -3.69192809e-01 -6.14270270e-01 -3.95874530e-01 -8.73590708e-01 -3.41257811e-01 -1.33250523e+00 2.18019634e-01 9.49879885e-01 9.38471317e-01 -7.14401186e-01 4.17113990e-01 -5.61988533e-01 -4.85639572e-01 3.94025177e-01 2.48877674e-01 -3.22463781e-01 4.24469203e-01 -1.81322622e+00 1.13891268e+00 -3.91366363e-01 3.23661268e-01 1.86843514e-01 -2.81570822e-01 -1.38258183e+00 2.68489867e-03 6.72580004e-02 -2.64089108e-01 9.54533994e-01 -1.33700168e+00 -1.73826468e+00 1.09683883e+00 -4.58223909e-01 -5.38783781e-02 -1.13092534e-01 -9.45057496e-02 -9.65669155e-01 5.09032421e-02 3.75342891e-02 2.08614931e-01 6.09089077e-01 -7.66482770e-01 -3.29569042e-01 -4.90571052e-01 2.98292369e-01 5.80803230e-02 -1.32642412e+00 6.11594915e-01 -2.21681520e-01 -3.90466481e-01 -4.81335521e-01 -8.91891718e-01 -3.44613314e-01 -5.04438400e-01 -2.78726995e-01 -3.90216798e-01 1.02190292e+00 -7.63229430e-01 1.54960859e+00 -2.17796659e+00 -1.37879893e-01 4.18911614e-02 1.75313115e-01 2.13822603e-01 4.81933840e-02 4.95031476e-01 -5.78208603e-02 3.95495653e-01 3.37546170e-01 -7.15650976e-01 1.09160833e-01 -9.55608264e-02 -3.19819450e-01 3.66209716e-01 1.43392593e-01 1.08997810e+00 -9.78587925e-01 -4.63782459e-01 2.18600124e-01 6.47820950e-01 -4.17213202e-01 -2.13569645e-02 3.27905685e-01 1.91972002e-01 -8.61574888e-01 6.10361338e-01 2.56148726e-01 -4.07258958e-01 2.68926919e-01 -3.13138843e-01 -1.64768875e-01 1.98362574e-01 -4.16639030e-01 1.12556612e+00 -1.12215137e+00 1.09658957e+00 1.08489081e-01 -8.53020608e-01 1.07376158e+00 2.31696725e-01 2.92119086e-01 -7.27495313e-01 8.85925293e-01 -1.50561392e-01 -1.42272770e-01 -6.88790441e-01 1.04154062e+00 -3.35088015e-01 -7.25271225e-01 6.12674713e-01 -1.98365226e-02 1.73109487e-01 -1.47779547e-02 3.91798556e-01 7.65052319e-01 -4.60643589e-01 1.96891427e-01 -2.90456653e-01 5.77121496e-01 -2.77068287e-01 1.76484808e-01 4.49180663e-01 -4.89862829e-01 1.18113257e-01 5.16204953e-01 -4.99326050e-01 -4.87370282e-01 -1.32250413e-01 -5.55748120e-02 1.82014239e+00 2.16263339e-01 -7.62418032e-01 -5.18326938e-01 -8.05413187e-01 -2.05766886e-01 5.13345897e-01 -1.30319118e+00 -2.31942981e-01 -1.29602954e-01 -1.00084853e+00 1.70484871e-01 5.85422993e-01 2.39304960e-01 -1.30523837e+00 -2.70261943e-01 -3.59545834e-02 -1.72796413e-01 -1.10295057e+00 -2.20954314e-01 2.57133156e-01 -3.61146837e-01 -8.93041372e-01 -3.00179660e-01 -8.70371997e-01 5.98715723e-01 1.22869767e-01 1.09019530e+00 9.02985856e-02 -9.57408324e-02 4.77594614e-01 -5.35436094e-01 -7.05875039e-01 4.35161404e-02 3.11425596e-01 1.40334219e-01 3.04936379e-01 1.15184903e+00 -3.72799605e-01 -4.79321152e-01 -7.29086436e-03 -7.26938963e-01 -2.76352286e-01 -8.21369886e-02 7.03292608e-01 -6.43855333e-02 -3.38205695e-01 7.91768134e-01 -1.08907056e+00 1.11859906e+00 -9.26973820e-01 1.07315928e-01 6.06324151e-02 -1.48241088e-01 -3.74406666e-01 8.64026248e-01 -4.30839479e-01 -1.12419569e+00 -5.70386410e-01 -3.82089466e-01 -3.55113926e-03 9.50748771e-02 8.74616683e-01 2.47839198e-01 -5.02780825e-02 3.68598074e-01 -1.36109546e-01 -2.90940762e-01 -3.06009978e-01 4.87303525e-01 1.20498276e+00 -2.86826361e-02 -1.92351624e-01 1.96819007e-01 6.74225390e-01 -6.42815530e-01 -1.12953067e+00 -1.44457304e+00 -5.37100077e-01 -2.53859252e-01 -3.15042943e-01 1.14222074e+00 -8.43991876e-01 -1.09357500e+00 3.51230204e-01 -8.48226666e-01 -2.67476905e-02 -4.41324525e-02 3.31171244e-01 -3.42015550e-02 2.77959317e-01 -1.05367303e+00 -9.87581015e-01 -4.76038188e-01 -1.07530832e+00 8.91046464e-01 4.37940121e-01 -6.19940758e-01 -1.64978969e+00 2.19654664e-01 4.90374476e-01 6.57004237e-01 1.03966609e-01 4.74587142e-01 -7.63362646e-01 6.11048937e-01 -7.76680827e-01 -3.48874927e-01 2.96092451e-01 2.05415655e-02 3.12631615e-02 -1.13600361e+00 2.02069402e-01 1.36675127e-02 -7.24216998e-01 9.56608355e-01 2.57746786e-01 1.28110266e+00 -2.81774458e-02 -1.16638996e-01 2.68234313e-01 1.06793976e+00 -1.40544057e-01 5.56111217e-01 6.14836335e-01 6.55554056e-01 7.25967467e-01 4.88511741e-01 5.25827825e-01 9.13354874e-01 2.90003002e-01 4.15576071e-01 -5.20124257e-01 5.31618655e-01 -1.39796048e-01 4.70423043e-01 1.26147354e+00 1.52368620e-01 -8.40422809e-02 -5.29250622e-01 7.00756252e-01 -1.86709833e+00 -9.77510631e-01 -8.96662697e-02 1.54449630e+00 7.45263040e-01 1.04212388e-01 -1.44244745e-01 -1.05634376e-01 8.78122866e-01 7.92503417e-01 -2.98645198e-01 -1.24207377e+00 -2.61226147e-01 2.31929451e-01 2.48556226e-01 4.98668849e-01 -1.37073278e+00 1.13826859e+00 5.87770700e+00 6.93354428e-01 -1.52347887e+00 3.98661047e-01 9.16129708e-01 -4.36794804e-03 -3.51542681e-01 -2.85998434e-01 -6.08725250e-01 6.61399662e-01 1.04713476e+00 9.14890021e-02 -7.38518685e-02 1.23805428e+00 1.16492242e-01 -4.02196571e-02 -4.46410894e-01 9.07997608e-01 2.89036155e-01 -1.24262023e+00 -5.91734290e-01 -7.38704503e-02 8.01294267e-01 3.05423558e-01 1.76401317e-01 8.44379604e-01 3.89991909e-01 -7.68036723e-01 4.28582430e-01 3.59818995e-01 5.65677404e-01 -6.05219007e-01 1.08207691e+00 -3.21836807e-02 -1.08165896e+00 -9.49622393e-02 -3.78082991e-01 -6.68325424e-01 2.64368564e-01 6.83736503e-01 -2.45825738e-01 2.14725912e-01 6.17232203e-01 1.04996955e+00 -6.57263279e-01 6.66463301e-02 -3.40148956e-01 6.48186028e-01 -9.70181152e-02 -7.83269227e-01 4.35019910e-01 -2.28997737e-01 4.86377776e-02 1.56990147e+00 -1.04810774e-01 6.13195635e-02 7.49126151e-02 2.67706305e-01 -3.70588064e-01 5.25487244e-01 -5.48912644e-01 -4.45848465e-01 1.68516800e-01 1.86875427e+00 -7.27126300e-01 -6.45249069e-01 -6.88142300e-01 9.06416118e-01 6.63995981e-01 3.15869033e-01 -8.65455151e-01 -6.93402171e-01 9.76817369e-01 -4.45512086e-01 3.03628266e-01 -2.38268413e-02 -3.56151164e-01 -1.49691534e+00 -2.39283577e-01 -4.18735743e-01 3.83061141e-01 -6.72096372e-01 -1.53333020e+00 8.16443324e-01 -8.47570896e-01 -7.85963476e-01 1.43120557e-01 -7.04219699e-01 -1.09782052e+00 6.83830142e-01 -1.75634456e+00 -1.08324468e+00 -1.02659442e-01 3.34309638e-01 3.07074189e-01 2.68225055e-02 1.09664202e+00 3.67687762e-01 -8.72393370e-01 6.36623979e-01 7.63811097e-02 3.95138800e-01 6.35504782e-01 -1.11446798e+00 9.16210115e-02 3.42102170e-01 -2.41027728e-01 7.26256788e-01 7.25188196e-01 -3.69047940e-01 -1.32988191e+00 -8.20766389e-01 1.20164382e+00 -5.69906294e-01 1.25899088e+00 -3.33702117e-01 -4.02022332e-01 5.77784181e-01 5.85042834e-01 1.89041242e-01 1.50698054e+00 8.08216870e-01 -2.86358923e-01 1.02548353e-01 -9.50645149e-01 7.86257148e-01 5.20361006e-01 -1.00314641e+00 -4.94635940e-01 2.39278346e-01 5.26122510e-01 4.20094728e-02 -9.56477702e-01 -5.75304180e-02 6.19473159e-01 -8.90990913e-01 6.65938854e-01 -1.19854045e+00 1.01969504e+00 7.44770095e-02 -1.71616212e-01 -1.52341580e+00 -1.75131813e-01 -1.94558486e-01 -6.35181889e-02 1.16175151e+00 4.87889737e-01 -7.13622510e-01 6.96391463e-01 7.98238754e-01 4.71826419e-02 -1.00277102e+00 -4.91841555e-01 -6.31141812e-02 -1.00627862e-01 -7.01970100e-01 3.58934492e-01 1.49392545e+00 7.46674657e-01 8.22765112e-01 -7.51401305e-01 -1.59672186e-01 -1.28625423e-01 3.91843021e-01 7.52245009e-01 -9.84322548e-01 1.38360813e-01 -5.07999063e-01 -4.59078103e-01 -7.62822390e-01 6.00962698e-01 -7.02663362e-01 -4.72065389e-01 -1.48972130e+00 7.38897547e-02 -2.04507083e-01 -5.38012028e-01 3.14212352e-01 -5.50397277e-01 8.07033837e-01 1.22502446e-01 -4.36586559e-01 -1.11821473e+00 8.91689837e-01 1.02118289e+00 -1.74397230e-01 -1.34805247e-01 -3.47465724e-01 -1.22572064e+00 1.07310998e+00 7.31613278e-01 -3.86137277e-01 2.18836039e-01 -1.93407953e-01 9.38688099e-01 -2.67097741e-01 -1.81805059e-01 -2.25457385e-01 1.72124878e-01 -1.72852688e-02 3.03869635e-01 -3.59411418e-01 7.11170197e-01 -6.67263746e-01 -6.52609944e-01 8.96686763e-02 -3.93078208e-01 -2.72774454e-02 1.69716433e-01 4.63288963e-01 -6.04954183e-01 -2.68034339e-01 3.63410175e-01 -1.60253942e-01 -7.02520669e-01 1.30378529e-01 -6.45595968e-01 1.66034400e-01 6.68393314e-01 -6.86834827e-02 -3.64603817e-01 -7.56654859e-01 -1.02765179e+00 4.41655636e-01 3.24525177e-01 7.74800658e-01 3.38153034e-01 -1.35161400e+00 -3.09444517e-01 6.47862330e-02 3.22813690e-01 -8.62972260e-01 6.07065916e-01 9.47401345e-01 1.38887107e-01 1.44915693e-02 2.18695506e-01 -7.85217434e-02 -1.42227793e+00 3.63060385e-01 7.60491043e-02 -3.58970404e-01 2.76843522e-04 9.88796115e-01 1.42288908e-01 -7.85344422e-01 -7.46884868e-02 -1.71239898e-02 -9.70030904e-01 7.62162328e-01 7.60099649e-01 1.16992265e-01 -1.21464148e-01 -1.02914381e+00 -4.44335490e-01 5.79732656e-01 -3.79628167e-02 3.40558618e-01 1.58614647e+00 -4.26656812e-01 -3.01335126e-01 8.40862513e-01 1.53814840e+00 5.60295403e-01 -4.74995345e-01 -1.32065639e-01 -1.07490748e-01 -3.86258990e-01 3.36526573e-01 -5.34609795e-01 -1.24897754e+00 1.04799926e+00 1.94319129e-01 7.46740103e-01 7.09299982e-01 -5.76506443e-02 1.09448314e+00 4.60162461e-01 -1.95083201e-01 -1.62078726e+00 2.60567695e-01 8.47725749e-01 5.19628346e-01 -1.55820239e+00 -1.93727583e-01 -7.94387311e-02 -1.10619652e+00 1.10657394e+00 6.01021349e-01 -1.22350156e-01 1.25411069e+00 9.28020552e-02 4.84296829e-01 -6.16259217e-01 -6.13737345e-01 -3.15038234e-01 1.47099748e-01 1.44713288e-02 7.07390189e-01 1.41687527e-01 -3.35008353e-01 1.14460742e+00 -2.62657970e-01 -3.16274256e-01 4.91563588e-01 8.50482523e-01 -3.10704678e-01 -8.19467127e-01 -9.51419920e-02 6.03196621e-01 -1.02210975e+00 -2.58502692e-01 -5.78784645e-01 2.30388641e-01 -1.55620068e-01 1.25740182e+00 9.62796658e-02 -6.73944235e-01 1.76831752e-01 2.79732704e-01 -1.61857277e-01 -6.36491776e-01 -1.01634121e+00 -4.18590695e-01 5.58740079e-01 -2.75680989e-01 -5.91919601e-01 -2.54193991e-01 -8.98170412e-01 -3.97480190e-01 -6.94446146e-01 2.99781799e-01 1.09760666e+00 1.08847392e+00 6.76847339e-01 6.25617087e-01 9.63638961e-01 -8.97782445e-01 3.88715640e-02 -1.18182397e+00 -5.54030538e-01 7.30811834e-01 2.71000475e-01 -6.33320093e-01 -5.51717937e-01 -4.08201516e-01]
[11.3532075881958, 6.848485946655273]
6e10369c-8874-4b8c-8998-61412531bcab
armanemo-a-persian-dataset-for-text-based
2207.11808
null
https://arxiv.org/abs/2207.11808v1
https://arxiv.org/pdf/2207.11808v1.pdf
ArmanEmo: A Persian Dataset for Text-based Emotion Detection
With the recent proliferation of open textual data on social media platforms, Emotion Detection (ED) from Text has received more attention over the past years. It has many applications, especially for businesses and online service providers, where emotion detection techniques can help them make informed commercial decisions by analyzing customers/users' feelings towards their products and services. In this study, we introduce ArmanEmo, a human-labeled emotion dataset of more than 7000 Persian sentences labeled for seven categories. The dataset has been collected from different resources, including Twitter, Instagram, and Digikala (an Iranian e-commerce company) comments. Labels are based on Ekman's six basic emotions (Anger, Fear, Happiness, Hatred, Sadness, Wonder) and another category (Other) to consider any other emotion not included in Ekman's model. Along with the dataset, we have provided several baseline models for emotion classification focusing on the state-of-the-art transformer-based language models. Our best model achieves a macro-averaged F1 score of 75.39 percent across our test dataset. Moreover, we also conduct transfer learning experiments to compare our proposed dataset's generalization against other Persian emotion datasets. Results of these experiments suggest that our dataset has superior generalizability among the existing Persian emotion datasets. ArmanEmo is publicly available for non-commercial use at https://github.com/Arman-Rayan-Sharif/arman-text-emotion.
['Hossein Zeinali', 'Hamid Habibzadeh Moshtaghin', 'Javad Peymanfard', 'Hossein Mirzaee']
2022-07-24
null
null
null
null
['emotion-classification', 'emotion-classification']
['computer-vision', 'natural-language-processing']
[-5.28145671e-01 -7.09607899e-02 -2.06953615e-01 -6.38195634e-01 -5.09574234e-01 -5.44230342e-01 3.52828443e-01 2.50827163e-01 -4.75490808e-01 6.58327460e-01 2.58566678e-01 -5.31903543e-02 3.01425815e-01 -4.82101738e-01 1.41609028e-01 -4.58674610e-01 5.77377863e-02 3.78426105e-01 -3.56339604e-01 -6.57693326e-01 3.04454327e-01 3.31761479e-01 -1.26273656e+00 3.83005172e-01 9.41786051e-01 1.27606678e+00 -3.05107892e-01 3.82360727e-01 -2.75879651e-01 1.12485659e+00 -6.22494638e-01 -1.17289996e+00 -1.50625557e-01 -2.22488031e-01 -9.72691357e-01 -1.59884527e-01 -3.31968367e-01 8.19196776e-02 1.26044348e-01 1.04886687e+00 5.74811101e-01 1.12600565e-01 8.56171668e-01 -1.66214967e+00 -8.35404336e-01 6.85488403e-01 -6.91819251e-01 1.59763861e-02 4.00558382e-01 -3.10849577e-01 1.00292206e+00 -9.32839990e-01 6.16333902e-01 1.35640943e+00 5.62753737e-01 4.43885148e-01 -5.55454910e-01 -1.12084877e+00 -5.30510619e-02 4.64517176e-01 -1.20687628e+00 -1.53654307e-01 1.12106860e+00 -2.94190586e-01 1.00572133e+00 1.10079408e-01 5.45340002e-01 1.41014647e+00 4.32885259e-01 9.88779902e-01 1.38507891e+00 -2.93627501e-01 6.28699288e-02 7.72231638e-01 2.97897577e-01 3.70303929e-01 -3.34743261e-01 -5.51837802e-01 -4.34363216e-01 -1.52083278e-01 -3.98456156e-02 -5.46807051e-02 1.21420443e-01 2.77424634e-01 -7.63796687e-01 1.05553019e+00 2.94594616e-01 4.41903561e-01 -5.62329650e-01 -5.50901115e-01 8.50619674e-01 4.46013719e-01 1.04598606e+00 1.41970590e-01 -8.75377774e-01 -6.33925617e-01 -3.56629223e-01 -9.63848531e-02 8.93541336e-01 5.99016130e-01 6.60627782e-01 -6.68226415e-03 4.02948976e-01 1.49176764e+00 2.72838712e-01 6.18229687e-01 8.20696473e-01 -5.90228379e-01 3.48449349e-01 7.92061925e-01 -3.34548652e-02 -1.52377999e+00 -5.23980916e-01 -1.01378322e-01 -9.55273807e-01 -4.45189387e-01 -1.65082499e-01 -5.51579118e-01 -3.36979717e-01 1.48112094e+00 3.22970629e-01 -2.11930394e-01 4.85516459e-01 7.06936836e-01 1.07876706e+00 9.08010244e-01 4.00377721e-01 -2.46831700e-01 1.50396335e+00 -8.17906916e-01 -8.74902964e-01 -2.73680657e-01 8.26073050e-01 -1.08161592e+00 1.14557254e+00 8.74971867e-01 -5.48647642e-01 -2.00059131e-01 -6.87394679e-01 -1.98476622e-03 -8.82356524e-01 2.97008485e-01 8.37776542e-01 7.52441227e-01 -5.98718703e-01 1.96678355e-01 -3.40592742e-01 -7.56751120e-01 2.88333267e-01 1.41725361e-01 -5.44449031e-01 3.05556476e-01 -1.50149572e+00 9.43880975e-01 1.61141515e-01 1.05263963e-01 -1.01701334e-01 -2.71929391e-02 -8.65535736e-01 -2.37521708e-01 1.75708381e-03 6.98765665e-02 1.20671868e+00 -1.56942475e+00 -1.59400296e+00 1.14983928e+00 -2.44524390e-01 -7.95763358e-02 5.09078838e-02 -4.16659772e-01 -1.13028634e+00 8.03336352e-02 1.04261771e-01 6.82032645e-01 5.14285743e-01 -9.99760628e-01 -5.11956513e-01 -5.94494760e-01 -1.60542816e-01 7.95683786e-02 -7.54015744e-01 8.50867152e-01 -5.47922626e-02 -5.93511045e-01 -1.52238756e-01 -1.10836995e+00 1.10615447e-01 -6.90143406e-01 -5.14622033e-01 -5.79158485e-01 1.06760120e+00 -8.16900015e-01 1.18842959e+00 -2.34257030e+00 -2.12313429e-01 7.51219876e-03 -5.34969792e-02 1.59467176e-01 -1.21510737e-01 6.63341284e-01 -2.61951774e-01 3.49020660e-01 8.78383126e-03 -2.07648680e-01 1.45562559e-01 2.32778583e-02 -2.20740646e-01 2.44694576e-01 2.31023446e-01 7.54578352e-01 -6.18611932e-01 -7.15691924e-01 6.33486286e-02 5.58781981e-01 -2.59661973e-01 2.00687304e-01 2.92798042e-01 3.00397247e-01 -4.74021971e-01 9.05693233e-01 6.19619906e-01 6.42746165e-02 5.81675246e-02 -1.71877533e-01 -4.72881272e-02 1.87462747e-01 -8.55195701e-01 9.78443801e-01 -4.78101939e-01 5.60674727e-01 5.46319969e-03 -1.03468955e+00 1.38169646e+00 3.32224071e-01 6.47801995e-01 -6.80141091e-01 8.64676476e-01 7.45281056e-02 1.05899749e-02 -6.49416447e-01 8.14863741e-01 -3.70151550e-01 -7.13600099e-01 4.59977329e-01 7.74307922e-02 6.71958923e-03 2.08896905e-01 3.34111542e-01 5.49023092e-01 -2.51397669e-01 3.79584223e-01 -1.15326226e-01 6.84503913e-01 -1.56607017e-01 7.65313745e-01 -6.60300553e-02 -6.08272195e-01 1.57527223e-01 6.03717804e-01 -2.43893966e-01 -5.20332158e-01 -5.66764534e-01 -2.06948400e-01 1.26902211e+00 -1.66979909e-01 -6.44858241e-01 -4.95204151e-01 -7.42083430e-01 -2.64112145e-01 7.81347036e-01 -5.72318852e-01 -7.40287304e-02 2.36274041e-02 -6.05958939e-01 7.56372750e-01 2.92828113e-01 8.32782984e-01 -1.51223624e+00 -4.16726589e-01 1.12586524e-02 -6.57549262e-01 -1.10599136e+00 -1.93564788e-01 1.92471504e-01 -3.61806601e-01 -8.90560269e-01 -3.90422165e-01 -7.93674767e-01 1.68487817e-01 -1.43200293e-01 9.94614542e-01 -3.95989567e-01 -1.07331671e-01 3.83138061e-01 -8.46210778e-01 -9.19578373e-01 -9.60118771e-02 -9.45049617e-03 1.61408573e-01 1.42100617e-01 1.03215873e+00 -2.89700985e-01 -1.78835750e-01 3.10678333e-01 -6.18688405e-01 -2.65627831e-01 2.37416118e-01 5.55163980e-01 6.32488579e-02 2.61564761e-01 9.59820092e-01 -8.89482677e-01 9.42294240e-01 -8.32855523e-01 5.89858219e-02 -5.85145392e-02 -4.56937760e-01 -5.69771409e-01 6.23468518e-01 -4.94672179e-01 -1.39245844e+00 -3.77001882e-01 -3.35502505e-01 -6.41579106e-02 -2.98523843e-01 7.49265313e-01 -2.09523812e-01 3.24512631e-01 2.87273079e-01 -2.11734101e-01 -1.09603196e-01 -2.21004516e-01 1.84046447e-01 1.47627556e+00 1.63243562e-01 -4.90010530e-01 1.54284820e-01 1.04623884e-01 -6.10731244e-01 -9.42948699e-01 -8.45135272e-01 -6.19110227e-01 -3.57552677e-01 -5.41992664e-01 8.05418611e-01 -8.77142429e-01 -1.06007111e+00 8.01314056e-01 -1.02793014e+00 -2.51172986e-02 2.27021292e-01 4.67902005e-01 -2.11640179e-01 1.79910243e-01 -1.22760963e+00 -1.11041176e+00 -7.10030138e-01 -7.87925899e-01 7.63895750e-01 3.39991570e-01 -8.39164257e-01 -1.04906476e+00 7.44366320e-03 5.03784597e-01 3.29159856e-01 7.51388222e-02 8.41469646e-01 -1.06937432e+00 6.96703911e-01 -2.12446272e-01 -1.29268944e-01 6.76213503e-01 1.64774418e-01 1.80658832e-01 -9.12012219e-01 1.61953390e-01 1.87024862e-01 -1.05961442e+00 4.11586612e-01 -1.73219747e-03 1.05429602e+00 -4.45913672e-01 3.15886736e-02 -3.36807631e-02 1.08152425e+00 5.25334895e-01 7.90922761e-01 4.76604789e-01 4.00351942e-01 8.68434906e-01 1.07565403e+00 8.70645642e-01 7.01689184e-01 1.15357928e-01 2.15719581e-01 7.35949073e-03 6.79819524e-01 -1.09929480e-01 6.74677312e-01 1.14589214e+00 7.45232403e-02 -3.67844939e-01 -9.13359582e-01 3.89179975e-01 -1.70011270e+00 -1.00016212e+00 -1.72127411e-01 1.60375619e+00 7.56297231e-01 -9.78743285e-02 1.97678849e-01 2.95668602e-01 6.70540094e-01 2.03643620e-01 -3.46528620e-01 -1.24084330e+00 -1.73431993e-01 1.92048728e-01 3.79868574e-03 1.90400764e-01 -1.12135267e+00 1.24135494e+00 5.02806330e+00 6.57654881e-01 -1.47649455e+00 1.41734704e-01 1.00085604e+00 1.25940263e-01 -4.76897731e-02 -3.81839126e-01 -7.00019538e-01 4.34281796e-01 1.21452415e+00 -1.70161098e-01 2.57745057e-01 1.17260826e+00 2.81693935e-01 -8.26758072e-02 -5.00699162e-01 1.29681325e+00 5.12788832e-01 -4.72919583e-01 -3.23951483e-01 -1.67517588e-01 4.86067414e-01 1.70658067e-01 6.47090524e-02 9.26060677e-01 1.88048974e-01 -8.64638984e-01 5.08123636e-01 3.16466801e-02 3.98517728e-01 -1.20076609e+00 1.04561460e+00 3.31925333e-01 -8.96502316e-01 -1.19640902e-01 -2.70942152e-01 -3.51288259e-01 1.34204760e-01 6.97783589e-01 -4.97289568e-01 3.26833457e-01 9.83112335e-01 1.06673241e+00 -4.15465355e-01 1.44967139e-01 -1.30163401e-01 8.22165668e-01 -1.17700852e-01 -4.48344201e-01 2.41363704e-01 -4.87180829e-01 1.67529210e-01 1.44688225e+00 3.21883798e-01 3.26378137e-01 2.48414502e-02 4.51225638e-01 -1.10346697e-01 8.63823712e-01 -6.56480372e-01 -5.08799613e-01 3.78487945e-01 1.82572150e+00 -7.70834506e-01 -2.83226818e-01 -4.34718668e-01 1.06588376e+00 3.47034007e-01 1.30168349e-01 -1.04449522e+00 -6.34365618e-01 5.88002324e-01 -3.89264524e-01 -1.35358244e-01 1.65424615e-01 -1.27563357e-01 -1.17976785e+00 -2.35772729e-01 -1.10843837e+00 6.29380167e-01 -1.05347514e+00 -1.72240734e+00 7.48439193e-01 -2.36186385e-01 -1.12067878e+00 -1.66160181e-01 -8.27614903e-01 -5.53764224e-01 5.23857594e-01 -1.12063301e+00 -1.18029404e+00 3.04119801e-03 7.82206178e-01 5.08257687e-01 -2.46137604e-01 1.01667297e+00 5.32811105e-01 -7.49681413e-01 5.37795305e-01 -2.28110269e-01 4.01871830e-01 1.22277868e+00 -9.02647555e-01 -2.86824644e-01 3.32886130e-01 -2.05177054e-01 5.06706178e-01 5.92785537e-01 -6.82925165e-01 -1.07728541e+00 -8.66716862e-01 1.37466788e+00 -2.72862047e-01 9.26141858e-01 -1.26417145e-01 -6.49376988e-01 9.54790592e-01 6.63709700e-01 -3.68457913e-01 1.22459435e+00 4.97272968e-01 -3.86059523e-01 -9.23171788e-02 -1.47273993e+00 5.50949395e-01 3.58055085e-01 -3.32562119e-01 -5.56268752e-01 1.47125974e-01 2.49713525e-01 2.10302882e-02 -9.41062152e-01 3.39145571e-01 5.31600535e-01 -1.06194174e+00 4.15494531e-01 -6.77338898e-01 9.13526773e-01 2.08934098e-01 -3.46552581e-01 -1.52907777e+00 -1.88772187e-01 -2.30045438e-01 4.61999327e-01 1.80648661e+00 3.86400968e-01 -9.77943957e-01 3.77098143e-01 6.03726029e-01 3.83179635e-03 -7.49112487e-01 -5.71264207e-01 -3.59473288e-01 1.49123892e-01 -8.72851491e-01 3.44815373e-01 1.48158431e+00 7.32515514e-01 7.88038194e-01 -4.45015103e-01 -3.57463926e-01 -1.26851993e-02 2.54564453e-02 6.57055557e-01 -1.22095525e+00 1.69117942e-01 -3.90189588e-01 -3.62349480e-01 -4.45378214e-01 6.92937076e-01 -8.69832575e-01 -2.37209350e-01 -1.29605389e+00 3.37446868e-01 -3.22795182e-01 -3.14502865e-01 7.25475490e-01 2.47462958e-01 4.37033057e-01 1.84079722e-01 -1.14032745e-01 -6.92853153e-01 6.53168440e-01 9.15277719e-01 -1.39106736e-01 -1.72323212e-01 -3.17446023e-01 -1.06571329e+00 1.13827670e+00 1.27215362e+00 -3.50596279e-01 -2.30569452e-01 1.71423957e-01 5.61639249e-01 -8.14163238e-02 -9.51396003e-02 -5.08370101e-01 -2.05736428e-01 -2.39469737e-01 2.95240223e-01 -6.41746402e-01 4.83978361e-01 -7.58010089e-01 -6.95935637e-02 1.08777635e-01 -3.15818995e-01 3.84303272e-01 2.78228104e-01 -3.45044956e-02 -5.06043851e-01 -4.03295159e-02 7.20571995e-01 6.93242773e-02 -7.78432012e-01 3.00858878e-02 -7.26991296e-01 2.11461306e-01 1.03084791e+00 1.70908093e-01 -4.68182594e-01 -6.81207001e-01 -6.16959333e-01 1.55499607e-01 8.35267752e-02 8.20143759e-01 5.52466691e-01 -1.30726409e+00 -7.21036434e-01 -2.53241174e-02 3.34200501e-01 -7.43200004e-01 4.10594761e-01 9.36354816e-01 -1.23801388e-01 2.35785112e-01 -3.16576272e-01 -2.24131271e-01 -1.52063012e+00 3.37313890e-01 -7.40659386e-02 -1.87115565e-01 -1.47951320e-01 6.52198911e-01 -2.16954350e-01 -7.87446260e-01 -2.70201620e-02 3.68367434e-02 -5.30193031e-01 4.62311089e-01 4.44143265e-01 4.01519299e-01 -1.17298298e-01 -1.24475980e+00 -5.29104471e-01 2.79909998e-01 -1.27382323e-01 -1.39817283e-01 1.22858655e+00 -1.87799782e-01 -5.75059772e-01 9.53527510e-01 1.33393681e+00 2.28005975e-01 -5.00022657e-02 1.70642123e-01 -6.04369603e-02 -1.71700642e-01 -2.69421041e-02 -1.13900638e+00 -1.05980110e+00 8.26757669e-01 4.49371248e-01 2.46264786e-01 1.20339108e+00 8.72953981e-02 9.73193228e-01 3.53637636e-01 3.02001297e-01 -1.50658262e+00 1.40819848e-01 1.00155568e+00 9.51817751e-01 -1.45064306e+00 -4.38767165e-01 -4.48145062e-01 -1.47365880e+00 9.80283439e-01 7.59107113e-01 1.36765152e-01 8.90723348e-01 1.99933365e-01 6.58657014e-01 -9.81370509e-02 -8.30944180e-01 -1.29277678e-02 -1.99050903e-02 3.59820485e-01 9.99914110e-01 3.97124738e-01 -7.20704436e-01 1.32495975e+00 -6.36204541e-01 -7.08514974e-02 4.88147318e-01 7.34355271e-01 -2.26710275e-01 -1.08823717e+00 -2.77479321e-01 6.09509706e-01 -8.20876300e-01 -1.81464870e-02 -8.29119921e-01 7.56257594e-01 3.70092876e-02 1.38180268e+00 -3.16461734e-02 -6.78328991e-01 8.96984637e-02 4.26835597e-01 9.18339714e-02 -3.42246771e-01 -6.67173922e-01 1.05540633e-01 5.62087297e-01 -2.85903841e-01 -5.97040832e-01 -6.91901863e-01 -1.46840549e+00 -4.92762983e-01 -9.95293781e-02 4.76685703e-01 7.53556013e-01 8.73152971e-01 3.38348806e-01 3.18989679e-02 8.10612619e-01 -6.24322712e-01 -6.14678487e-02 -1.29400969e+00 -9.13831353e-01 5.92881858e-01 -3.57373416e-01 -5.80663204e-01 -4.86012220e-01 -6.67579472e-02]
[12.668844223022461, 6.175174713134766]
ab3d8bd6-a725-4b28-bc8a-99f27d29b095
bottom-up-constituency-parsing-and-nested
2110.05419
null
https://arxiv.org/abs/2110.05419v2
https://arxiv.org/pdf/2110.05419v2.pdf
Bottom-Up Constituency Parsing and Nested Named Entity Recognition with Pointer Networks
Constituency parsing and nested named entity recognition (NER) are similar tasks since they both aim to predict a collection of nested and non-crossing spans. In this work, we cast nested NER to constituency parsing and propose a novel pointing mechanism for bottom-up parsing to tackle both tasks. The key idea is based on the observation that if we traverse a constituency tree in post-order, i.e., visiting a parent after its children, then two consecutively visited spans would share a boundary. Our model tracks the shared boundaries and predicts the next boundary at each step by leveraging a pointer network. As a result, it needs only linear steps to parse and thus is efficient. It also maintains a parsing configuration for structural consistency, i.e., always outputting valid trees. Experimentally, our model achieves the state-of-the-art performance on PTB among all BERT-based models (96.01 F1 score) and competitive performance on CTB7 in constituency parsing; and it also achieves strong performance on three benchmark datasets of nested NER: ACE2004, ACE2005, and GENIA. Our code is publicly available at \url{https://github.com/sustcsonglin/pointer-net-for-nested}.
['Kewei Tu', 'Songlin Yang']
2021-10-11
null
https://aclanthology.org/2022.acl-long.171
https://aclanthology.org/2022.acl-long.171.pdf
acl-2022-5
['constituency-parsing', 'nested-named-entity-recognition']
['natural-language-processing', 'natural-language-processing']
[-7.44216219e-02 4.01713639e-01 -4.11537588e-01 -5.69003761e-01 -1.15388870e+00 -1.01047802e+00 7.92911053e-02 4.94036049e-01 -4.05080825e-01 4.89667028e-01 4.19312268e-01 -7.76459336e-01 3.05340022e-01 -1.04827428e+00 -8.13703358e-01 -1.91016257e-01 -1.15612140e-02 4.03896123e-01 5.52487373e-01 -1.29172847e-01 -1.07432175e-02 2.24389106e-01 -9.16470408e-01 4.97307897e-01 6.79779887e-01 6.99223340e-01 2.72713810e-01 8.01927984e-01 -5.04453421e-01 6.82908893e-01 -3.58193994e-01 -9.01274621e-01 1.19291916e-01 -1.04172394e-01 -1.31762397e+00 -6.18259847e-01 5.36220908e-01 -1.02565132e-01 -2.04220548e-01 9.34147120e-01 3.31625819e-01 -5.23865484e-02 7.84854069e-02 -7.09969342e-01 -4.53067124e-01 1.28436506e+00 -4.31416690e-01 4.28211898e-01 2.07154050e-01 -3.32134455e-01 1.73703933e+00 -6.87564671e-01 6.11086011e-01 9.33328092e-01 8.17192674e-01 7.65606046e-01 -1.17941451e+00 -6.26713216e-01 4.77718353e-01 -2.81030834e-01 -8.35210621e-01 -4.50683057e-01 3.49236995e-01 -1.20224245e-01 9.44931924e-01 4.06294703e-01 2.45814636e-01 7.05620170e-01 5.41412178e-03 7.35647619e-01 8.55040967e-01 -4.56572682e-01 7.05848262e-02 -5.06288111e-01 8.48184705e-01 8.66994858e-01 1.97268888e-01 1.40381260e-02 -4.14998770e-01 -1.70754135e-01 4.28789616e-01 -3.24981987e-01 1.32214418e-02 1.57144889e-01 -1.12301040e+00 7.71695435e-01 5.13431191e-01 5.48940659e-01 -1.52976349e-01 1.38606757e-01 5.69875777e-01 -9.41527486e-02 1.82959527e-01 1.76101565e-01 -8.34449410e-01 -2.06194520e-01 -7.08683670e-01 -3.34424749e-02 1.11155486e+00 1.08769000e+00 7.00819135e-01 -5.92597425e-01 -1.30022168e-01 7.85473347e-01 1.39184207e-01 1.86309740e-01 1.25954434e-01 -8.32654059e-01 1.10093653e+00 6.21218860e-01 -1.02289841e-01 -6.42640054e-01 -7.33462691e-01 -9.99751166e-02 -7.27393746e-01 -4.30260718e-01 6.29902661e-01 -2.83579826e-01 -8.13152671e-01 1.96605849e+00 6.08882189e-01 1.22676373e-01 1.41878426e-01 5.80592752e-01 1.12683916e+00 9.37761903e-01 5.28715491e-01 3.62375453e-02 1.91412997e+00 -9.54453111e-01 -4.12448585e-01 -4.25424784e-01 8.24995399e-01 -6.75722063e-01 8.79291773e-01 -9.04126279e-03 -1.26889002e+00 -5.12721717e-01 -7.74532735e-01 -4.43792760e-01 -2.38406196e-01 1.20735154e-01 6.09362006e-01 6.25400841e-01 -8.70448411e-01 7.29505420e-01 -1.16877711e+00 -2.14671448e-01 -2.51780432e-02 1.77676126e-01 -3.98945868e-01 1.28119051e-01 -1.17332256e+00 5.24784982e-01 7.37345695e-01 2.49437094e-01 -3.08616012e-01 -5.89537203e-01 -9.90513861e-01 2.12378949e-01 2.30875298e-01 -4.78673667e-01 1.59587061e+00 -2.18451500e-01 -1.22483110e+00 9.39317226e-01 -4.62494910e-01 -3.77038121e-01 1.95640966e-01 -3.52812737e-01 -5.42187333e-01 -7.95128047e-02 4.49860454e-01 8.14168572e-01 -1.73457041e-01 -9.46770787e-01 -1.10015023e+00 -4.16001976e-01 2.64962018e-01 -1.20238468e-01 -2.55163103e-01 2.64808923e-01 -7.24987686e-01 -3.45616907e-01 6.24376655e-01 -8.96918774e-01 -3.65091741e-01 -6.19752586e-01 -8.63949299e-01 -5.34147918e-01 1.85485080e-01 -9.29066896e-01 1.74248409e+00 -2.07185292e+00 -1.93226144e-01 -4.01702913e-04 2.20496476e-01 2.54369110e-01 -1.43432558e-01 5.71751535e-01 -1.90075319e-02 7.03665197e-01 -3.40691626e-01 -3.84979695e-01 6.65251911e-02 2.46272832e-01 -3.10777396e-01 1.24668278e-01 1.05521873e-01 9.12796855e-01 -8.91859233e-01 -7.39671826e-01 -1.12158895e-01 3.83893140e-02 -4.88698125e-01 1.41981855e-01 -1.04436897e-01 3.45910251e-01 -3.93385053e-01 5.74904501e-01 7.99373090e-01 -1.31108999e-01 7.68900394e-01 -1.99006736e-01 -6.44108951e-01 1.04455936e+00 -1.17490137e+00 1.52191746e+00 -6.43596470e-01 2.20789477e-01 8.61110315e-02 -6.29027784e-01 8.53971243e-01 3.37338328e-01 -1.32245838e-03 -6.80717945e-01 2.88912188e-02 3.73179078e-01 -1.72662109e-01 -8.63713995e-02 6.44982576e-01 1.11372732e-01 -7.61835873e-01 1.09643646e-01 -1.69633001e-01 4.19913173e-01 4.83530074e-01 1.09226212e-01 1.34736872e+00 6.28894493e-02 4.84940439e-01 -1.95307642e-01 5.24205744e-01 9.40031484e-02 1.36773765e+00 8.04739118e-01 -2.43231550e-01 2.83957183e-01 7.30078697e-01 -6.47968113e-01 -9.29662168e-01 -1.31212449e+00 -2.01822490e-01 1.43208075e+00 1.19094245e-01 -7.21355498e-01 -9.09131646e-01 -9.60973620e-01 -4.91147488e-01 7.97623456e-01 -4.42795932e-01 5.18174648e-01 -1.47522688e+00 -3.75167757e-01 8.62579942e-01 8.08233857e-01 5.10613680e-01 -1.34243345e+00 -5.40916562e-01 5.25111854e-01 -5.66649258e-01 -1.36638415e+00 -7.11507380e-01 5.36185682e-01 -1.02184916e+00 -1.10399330e+00 -2.28477955e-01 -1.42963946e+00 4.08508003e-01 -1.07452504e-01 1.43726742e+00 2.79068798e-01 2.08055660e-01 -3.42149109e-01 -4.13008302e-01 5.05776815e-02 -5.20443320e-01 6.65855885e-01 -5.76727629e-01 -6.49630904e-01 2.62978911e-01 -4.19985384e-01 -5.61061144e-01 3.56887221e-01 -5.78560233e-01 2.50642896e-01 4.70907390e-01 8.67982090e-01 8.93458486e-01 -2.17115998e-01 3.26499790e-01 -1.16693044e+00 1.86308593e-01 -4.60772336e-01 -7.76725113e-01 5.80861568e-01 -7.18481541e-02 6.97380081e-02 9.11943376e-01 9.10423249e-02 -1.15736437e+00 2.57614404e-01 -7.15345144e-01 3.53657603e-01 -4.43363994e-01 2.99343705e-01 -5.13416409e-01 6.61458731e-01 2.13465974e-01 2.67502833e-02 -8.90778303e-01 -8.24958682e-01 4.87147897e-01 4.93395180e-01 7.99365342e-01 -8.77907932e-01 5.46491861e-01 2.45697394e-01 -1.87752262e-01 -4.66734558e-01 -1.14261329e+00 -5.22128642e-01 -7.66794860e-01 3.25228363e-01 1.03244507e+00 -7.36551046e-01 -8.49019170e-01 2.35191658e-01 -1.59615684e+00 -3.38898450e-01 6.66943267e-02 1.00306116e-01 -1.12742603e-01 4.38796461e-01 -1.22404957e+00 -5.05175889e-01 -7.45026529e-01 -8.31279695e-01 9.48612511e-01 4.76709664e-01 -1.11744232e-01 -8.19877982e-01 2.23603591e-01 3.35532427e-01 -1.94666475e-01 1.39626533e-01 1.10814345e+00 -8.97161424e-01 -4.00696546e-01 6.54163435e-02 -1.99227586e-01 -6.64423453e-03 -9.02924761e-02 -1.38595581e-01 -6.64299726e-01 8.18511546e-02 -5.08989036e-01 1.29667344e-02 8.26681018e-01 1.52920246e-01 9.20170724e-01 -3.25078726e-01 -4.91578877e-01 6.53600395e-01 1.30653775e+00 2.37921804e-01 5.19484460e-01 5.47906935e-01 5.89024484e-01 7.40930796e-01 5.25044203e-01 1.00616723e-01 8.58916402e-01 4.34664935e-01 3.51613343e-01 -7.36553296e-02 -1.11250412e-02 -6.78686321e-01 1.81242079e-01 9.97576535e-01 2.84753352e-01 -3.90093356e-01 -1.14252377e+00 7.26521432e-01 -1.69465911e+00 -7.89973974e-01 -5.30084133e-01 2.02507687e+00 8.41385782e-01 3.86947125e-01 4.10711169e-02 -8.86296108e-02 1.17166746e+00 2.23401174e-01 -3.70372474e-01 -7.98770666e-01 -3.33206244e-02 5.04673123e-01 6.66460931e-01 4.05862421e-01 -1.32357919e+00 1.40790474e+00 5.03247404e+00 5.12850761e-01 -6.58008277e-01 2.64105886e-01 8.99685919e-01 4.16026682e-01 -2.29270473e-01 2.63275445e-01 -1.50350046e+00 3.80386442e-01 1.08545077e+00 1.89835578e-01 5.04420996e-02 7.62022555e-01 -8.11117366e-02 2.13812307e-01 -1.01157272e+00 3.35507333e-01 -6.73426211e-01 -1.39312148e+00 -2.99535990e-01 -9.55323875e-02 3.30412090e-01 2.91813910e-01 -3.54358315e-01 4.47233438e-01 6.76655889e-01 -5.84605753e-01 9.76958215e-01 -1.91409394e-01 6.12790227e-01 -7.34033287e-01 6.88171864e-01 5.52947640e-01 -1.90034914e+00 -3.79725322e-02 -3.02896917e-01 1.59621298e-01 6.46364033e-01 5.37076712e-01 -4.77071345e-01 5.88579237e-01 8.73235464e-01 9.16812271e-02 -9.09997299e-02 8.38610530e-01 -7.38894165e-01 9.81777191e-01 -4.52893764e-01 -6.12361617e-02 3.77801001e-01 -3.05978414e-02 3.80700111e-01 1.59038222e+00 2.03451946e-01 4.14062768e-01 3.14925164e-01 4.87661213e-01 -3.48807573e-01 3.16295803e-01 -8.09579268e-02 2.50604838e-01 1.07025099e+00 1.38518143e+00 -8.71658385e-01 -2.53257483e-01 -5.41380227e-01 7.62666047e-01 9.27151680e-01 -2.69016325e-01 -1.05670249e+00 -4.07605737e-01 5.85629702e-01 -1.98462233e-01 7.64644384e-01 -2.39486784e-01 -4.37565207e-01 -1.00858057e+00 2.63135076e-01 -5.42868793e-01 9.85019565e-01 -1.68396518e-01 -1.13171089e+00 9.92268801e-01 -2.91076869e-01 -7.59370148e-01 1.10082828e-01 -4.81258035e-01 -9.26068664e-01 7.45128930e-01 -1.53824103e+00 -1.23439181e+00 -7.86812603e-03 2.21311778e-01 4.92812991e-01 6.24486387e-01 9.66119945e-01 3.08592767e-01 -8.64448071e-01 7.88615108e-01 -1.84471428e-01 8.23194027e-01 2.48087674e-01 -1.38686800e+00 1.33543169e+00 1.17809892e+00 4.07527953e-01 7.02944994e-01 1.91717178e-01 -6.49191856e-01 -1.03128684e+00 -1.14142251e+00 1.64587128e+00 -2.37355679e-01 6.98060930e-01 -6.26748919e-01 -8.01908374e-01 8.67554188e-01 1.25677302e-01 7.33919144e-02 6.59732401e-01 4.85101998e-01 -5.74577808e-01 4.54180054e-02 -8.76501143e-01 5.43138862e-01 1.39486277e+00 -1.14310302e-01 -8.66082191e-01 -3.18397544e-02 1.04893267e+00 -7.55425513e-01 -1.13297117e+00 3.51196587e-01 5.14789045e-01 -1.08176410e+00 7.31612206e-01 -5.93582213e-01 3.65523994e-01 -1.52251184e-01 -2.84973145e-01 -8.76785100e-01 -5.07464528e-01 -5.50836861e-01 2.47906428e-03 1.82838202e+00 9.68464196e-01 -6.30481660e-01 9.06691849e-01 5.42909980e-01 -4.34506327e-01 -9.41785157e-01 -1.03248656e+00 -8.07091534e-01 3.30086380e-01 -4.88756746e-01 8.57029080e-01 5.80051541e-01 -9.93583202e-02 4.46073294e-01 -7.14661479e-02 6.66553438e-01 5.09596705e-01 4.51163024e-01 4.07156259e-01 -1.08432555e+00 -2.07159489e-01 -2.00230852e-01 8.61138627e-02 -1.50114608e+00 2.75211722e-01 -1.03195715e+00 2.65211284e-01 -1.61264634e+00 9.01539028e-02 -1.00274217e+00 -2.87404716e-01 9.08641875e-01 -3.18220198e-01 4.87204175e-03 4.22200561e-01 1.57576039e-01 -6.04584515e-01 -3.52516174e-02 8.16368043e-01 2.42558166e-01 -3.47904235e-01 9.78764370e-02 -7.80906022e-01 8.49041283e-01 1.19964767e+00 -7.96467423e-01 3.29987228e-01 -8.36952507e-01 2.03429446e-01 4.55504328e-01 5.16702496e-02 -8.52660537e-01 3.30585301e-01 6.51407838e-02 5.56334108e-02 -9.57822144e-01 4.03168947e-02 -5.10027647e-01 -4.59762514e-02 5.45542836e-01 -3.28074455e-01 5.07793486e-01 2.03654036e-01 3.73483926e-01 -1.42941505e-01 -4.63481635e-01 5.32378733e-01 -2.45999202e-01 -9.39446211e-01 3.31704617e-01 3.97823984e-03 5.63301027e-01 6.44590735e-01 -3.47270854e-02 -5.91345906e-01 3.69681746e-01 -7.26917446e-01 4.14615512e-01 1.05431624e-01 3.50380659e-01 1.72797516e-01 -8.62260342e-01 -7.56087065e-01 -1.60167575e-01 6.20040158e-03 2.50421673e-01 2.20091358e-01 4.07817662e-01 -5.94408989e-01 4.60733294e-01 2.70033568e-01 -3.77236784e-01 -1.39554644e+00 2.66189128e-01 1.35929808e-01 -8.89639497e-01 -7.97741294e-01 9.77443039e-01 2.78133363e-01 -9.06337380e-01 1.03537321e-01 -5.22393465e-01 -1.78219453e-01 -2.50992149e-01 4.89756107e-01 2.78727055e-01 1.36323467e-01 -6.95260584e-01 -5.55691242e-01 5.84857345e-01 -2.17629656e-01 6.31243661e-02 1.31645429e+00 8.04216340e-02 -3.37038547e-01 1.09315693e-01 1.14221418e+00 4.32535976e-01 -9.48110402e-01 -1.91082090e-01 7.50109017e-01 -4.23342921e-02 -2.87700146e-01 -7.95675933e-01 -9.76874828e-01 6.07830942e-01 1.27151772e-01 3.36587995e-01 1.00897861e+00 4.80928987e-01 1.36603081e+00 4.01462883e-01 3.86898965e-01 -8.10911417e-01 -5.17646492e-01 9.72577691e-01 2.03707576e-01 -9.32577014e-01 -4.72987056e-01 -9.28761840e-01 -4.11861122e-01 1.14712656e+00 7.27090836e-01 1.64150119e-01 4.10846591e-01 4.64964449e-01 1.42804394e-02 -5.93295582e-02 -7.42524922e-01 -1.41168535e-01 1.05571756e-02 1.22632526e-01 7.67531574e-01 4.26073760e-01 -6.31881297e-01 9.94827330e-01 -5.38574576e-01 -4.82803315e-01 1.45358309e-01 1.03173292e+00 -4.60933805e-01 -1.50348949e+00 -1.77905098e-01 1.21074803e-01 -8.51024926e-01 -5.25456190e-01 -1.59213230e-01 8.51051271e-01 2.79638380e-01 1.15342975e+00 1.20687768e-01 -1.64147481e-01 6.04851484e-01 1.44732418e-02 1.10657901e-01 -7.09308863e-01 -1.04000652e+00 -1.69252515e-01 5.89636147e-01 -4.95578170e-01 -4.15799487e-03 -6.83703840e-01 -1.73957586e+00 -4.40553904e-01 -4.85556841e-01 4.74213213e-01 5.94016790e-01 5.50319135e-01 3.61611694e-01 3.12549293e-01 5.78177631e-01 -2.58706152e-01 -4.25069779e-01 -6.46340549e-01 -2.52006620e-01 1.70125857e-01 -1.40646309e-01 -1.34056091e-01 -1.71320096e-01 -1.57726720e-01]
[10.048771858215332, 9.581765174865723]
dc0f5d11-93d7-4be9-b5eb-4b286a45e85a
measuring-and-improving-compositional
null
null
https://openreview.net/forum?id=-B3vVVeVyTr
https://openreview.net/pdf?id=-B3vVVeVyTr
Measuring and Improving Compositional Generalization in Text-to-SQL via Component Alignment
Recently, the challenge of compositional generalization in NLP has attracted more and more attention. Specifically, many prior works show that neural networks struggle with compositional generalization where training and testing distributions differ. However, most of these works are based on word-level synthetic data or a specific data split method to generate compositional biases. In this work, we propose a clause-level compositional example generation method, and we focus on text-to-SQL tasks. We start by splitting the sentences in the Spider text-to-SQL dataset into several sub-sentences, annotating each sub-sentence with its corresponding SQL clause, resulting in a new dataset, Spider-SS. Building upon Spider-SS, we further construct a new dataset named Spider-CG, by substituting and appending Spider-SS sub-sentences to test the ability of models to generalize compositionally. Experiments show that previous models suffer significant performance degradation when evaluated on Spider-CG, even though every sub-sentence has been seen during training. To deal with this problem, we modify the RATSQL+GAP model to fit the segmented data of Spider-SS, and results show that this method can improve generalization performance.
['Anonymous']
2021-10-16
null
null
null
acl-arr-october-2021-10
['text-to-sql']
['computer-code']
[ 6.14654422e-01 1.61847249e-01 -1.37379453e-01 -7.85206914e-01 -7.06688166e-01 -7.53895402e-01 4.31400508e-01 1.91204716e-02 -2.86977381e-01 8.65606129e-01 1.24787934e-01 -5.94315946e-01 4.20618027e-01 -1.03321779e+00 -1.00755715e+00 -3.28028500e-01 1.54375777e-01 6.05039954e-01 4.03514266e-01 -4.31160629e-01 8.91679674e-02 1.91433921e-01 -1.44940460e+00 1.00366747e+00 1.43206692e+00 8.32827091e-01 3.76414180e-01 3.10939968e-01 -4.91904110e-01 4.92193252e-01 -9.14508402e-01 -6.40482664e-01 2.52547443e-01 -9.20736194e-01 -8.76063585e-01 -1.90473542e-01 7.60364532e-01 5.15631633e-03 1.36249557e-01 1.03509831e+00 4.37245458e-01 4.18692939e-02 5.19758463e-01 -1.17900121e+00 -8.51239383e-01 1.30787647e+00 -1.29779741e-01 5.93212526e-03 3.22792023e-01 2.20623184e-02 1.20572376e+00 -9.45658088e-01 7.40847945e-01 1.52037656e+00 8.61986399e-01 9.92259324e-01 -1.74586022e+00 -8.34169924e-01 3.79877388e-01 9.68884304e-02 -1.12262774e+00 -9.15552154e-02 8.02257776e-01 -3.87327485e-02 9.19915974e-01 3.83763283e-01 6.06768429e-01 1.59870207e+00 -1.60000414e-01 1.19432664e+00 1.31702304e+00 -4.13815200e-01 4.48881626e-01 9.58592668e-02 9.38492790e-02 2.87358165e-01 2.30117679e-01 -7.68636614e-02 -4.63219047e-01 5.17512932e-02 1.77372441e-01 -3.12573701e-01 -3.92054886e-01 -2.54723787e-01 -1.09518659e+00 9.38115299e-01 6.41901016e-01 2.60561913e-01 5.00024110e-02 -2.54002422e-01 6.63443327e-01 3.77843231e-01 4.27500308e-01 6.35213792e-01 -5.34856677e-01 2.22640023e-01 -1.05560899e+00 9.30837035e-01 8.95078063e-01 1.27763820e+00 6.36507571e-01 -4.27568443e-02 -2.72937387e-01 1.09924746e+00 -1.78908095e-01 3.43891531e-01 1.02564144e+00 -5.88590503e-01 9.06965137e-01 7.18464017e-01 -4.52577740e-01 -5.89856088e-01 -2.40961790e-01 -6.45908415e-01 -7.55251110e-01 -2.88339376e-01 2.21470863e-01 -8.19037259e-02 -8.64188790e-01 2.11319923e+00 8.86303782e-02 -2.42483586e-01 4.25134212e-01 6.23318553e-01 7.90140629e-01 7.16934502e-01 -1.34365018e-02 -2.93246675e-02 9.72412884e-01 -1.14062893e+00 -4.36212689e-01 -5.96710742e-01 8.05372000e-01 -2.80918658e-01 1.94317591e+00 3.39894205e-01 -1.06769609e+00 -7.12446213e-01 -1.27237809e+00 6.21245280e-02 -5.51055133e-01 6.67525129e-03 1.05703808e-01 6.47460818e-01 -7.43289053e-01 6.09495461e-01 -5.49427271e-01 -4.39998358e-01 4.08021390e-01 -3.26478528e-03 -3.89144458e-02 -2.99163610e-01 -1.33947110e+00 7.62898147e-01 8.68419647e-01 -3.72061491e-01 -6.49629712e-01 -9.31989610e-01 -9.53519404e-01 1.74404725e-01 5.11784375e-01 -8.38164389e-01 1.45607758e+00 -1.02809250e+00 -1.23244524e+00 6.47588193e-01 -2.67326593e-01 -6.74162626e-01 3.80466610e-01 -1.30033717e-02 -3.15375805e-01 -2.11177453e-01 2.99581975e-01 9.70289528e-01 5.96167266e-01 -1.59379411e+00 -5.68409503e-01 -4.27186906e-01 1.45457769e-02 4.14322168e-02 -4.31529015e-01 -2.28011206e-01 -2.95672230e-02 -1.12856555e+00 1.49212867e-01 -9.51363921e-01 2.51776762e-02 -7.18225598e-01 -6.87845945e-01 -2.85967082e-01 6.94057226e-01 -3.81112754e-01 1.42403507e+00 -2.05560851e+00 1.83640853e-01 -5.02581000e-02 -3.59301597e-01 2.99562663e-01 -4.07618165e-01 5.81077218e-01 -1.93661362e-01 4.72145140e-01 -8.07076573e-01 -4.40756559e-01 1.23205177e-01 4.28978771e-01 -7.03801334e-01 -4.79970336e-01 5.13088584e-01 1.00103724e+00 -7.69218981e-01 -4.00085568e-01 -4.51672435e-01 -1.86736345e-01 -1.00455439e+00 1.27520859e-01 -8.11154842e-01 1.89634442e-01 -4.16010395e-02 4.88521129e-01 6.72346413e-01 3.28146815e-02 2.36070126e-01 -7.96334222e-02 3.05343956e-01 6.97049916e-01 -8.16950500e-01 1.81304228e+00 -5.72207689e-01 3.30123276e-01 -4.13751274e-01 -1.10420597e+00 9.93798077e-01 -1.77935854e-01 -2.32630298e-01 -8.94766212e-01 -3.28102201e-01 5.54636180e-01 2.58820564e-01 -5.17247260e-01 7.63866484e-01 -5.30427992e-01 -2.25975871e-01 4.25206989e-01 -5.20516466e-03 -5.26959836e-01 7.87501037e-01 2.32987985e-01 9.83242273e-01 3.46464843e-01 -1.17673866e-01 -9.20285136e-02 5.95800281e-01 9.61714163e-02 4.87116396e-01 7.16696918e-01 3.00576359e-01 8.32931817e-01 6.08906627e-01 -2.25254402e-01 -1.11497247e+00 -1.41253972e+00 -8.18438679e-02 1.25621879e+00 -1.29201367e-01 -5.43249786e-01 -1.14094603e+00 -1.14294493e+00 7.93568566e-02 1.51966262e+00 -6.68211997e-01 -3.28034550e-01 -8.14584315e-01 -9.05981421e-01 6.42678916e-01 6.33502722e-01 5.47777593e-01 -1.46866381e+00 -4.51614767e-01 2.29139760e-01 -2.93477863e-01 -1.06532061e+00 -4.62916404e-01 3.26116949e-01 -9.38002765e-01 -6.63677275e-01 -5.53655982e-01 -1.03757024e+00 6.41068161e-01 -2.13995367e-01 1.33156455e+00 -2.39796177e-01 -7.92249739e-02 -2.51338124e-01 -6.22742891e-01 -4.27428037e-01 -8.04420769e-01 5.53699255e-01 -3.21976095e-01 -2.76439756e-01 2.76037216e-01 -5.17185211e-01 -2.57027358e-01 3.43851238e-01 -1.49590385e+00 1.22597203e-01 7.58656979e-01 1.01109850e+00 4.34511572e-01 -1.51279464e-01 1.15405023e+00 -1.24950850e+00 1.22441638e+00 -3.34011376e-01 -1.18034273e-01 2.73526907e-01 -6.34662986e-01 2.43647546e-01 1.03260350e+00 -7.54617631e-01 -9.81371522e-01 -1.22332536e-01 -1.79859146e-01 -1.98621914e-01 -1.43295288e-01 9.50726151e-01 -4.73968506e-01 5.10629475e-01 1.04236889e+00 5.70914924e-01 1.01646371e-01 -3.80594254e-01 5.08164406e-01 7.10296392e-01 5.32706678e-01 -8.69730711e-01 6.07606173e-01 -1.53641794e-02 -3.62588465e-01 -5.82038462e-01 -1.13026929e+00 2.31865391e-01 -3.52090269e-01 3.38660777e-01 5.11753201e-01 -4.62865293e-01 -1.45948723e-01 2.50050277e-01 -1.06752944e+00 -7.13156343e-01 -5.07243097e-01 1.02770008e-01 -4.69538838e-01 2.89007723e-01 -5.26385188e-01 -4.66315359e-01 -2.92827576e-01 -1.16075003e+00 9.41073954e-01 -1.24552935e-01 -5.39654970e-01 -7.81952620e-01 9.04452577e-02 2.00854063e-01 4.90966499e-01 5.78739904e-02 1.62150013e+00 -1.08925045e+00 -2.52285391e-01 -9.96449739e-02 -1.53564841e-01 8.71535599e-01 3.45276371e-02 -3.08503300e-01 -7.37027586e-01 -2.36096874e-01 6.72510862e-02 -6.19834065e-01 8.25885415e-01 -8.10161531e-02 1.49923611e+00 -5.24816990e-01 -1.77537858e-01 4.82529670e-01 1.13600135e+00 1.84398070e-01 4.26719189e-01 4.72807795e-01 4.03898180e-01 7.77362049e-01 6.04872286e-01 -9.68745276e-02 3.51862490e-01 5.55129707e-01 3.87042820e-01 2.43213102e-01 -8.24826956e-02 -6.06427252e-01 4.09118712e-01 6.59669518e-01 5.66267550e-01 -4.75008905e-01 -8.24571729e-01 6.37731791e-01 -1.65605974e+00 -8.60187411e-01 1.70320541e-01 2.16093588e+00 1.28104067e+00 3.41803372e-01 1.39647901e-01 2.36925155e-01 3.61122072e-01 3.92678380e-02 -6.25708342e-01 -4.31345373e-01 -3.16731334e-01 4.42052931e-01 2.04826356e-03 2.95640260e-01 -7.59995282e-01 1.14577234e+00 5.83490372e+00 9.61394608e-01 -1.14356053e+00 -2.37673268e-01 6.06680274e-01 -2.14032874e-01 -7.23389983e-01 -1.67959884e-01 -9.07137215e-01 7.19575524e-01 8.79476964e-01 -2.05617234e-01 4.65640992e-01 6.82163775e-01 -1.29115045e-01 1.84754550e-01 -1.58336675e+00 5.31322837e-01 4.94784892e-01 -1.16146457e+00 8.24843347e-01 -4.02228683e-01 8.36709738e-01 -1.91071495e-01 1.31063879e-01 8.16356361e-01 3.64734195e-02 -1.10073173e+00 7.94119060e-01 -3.95297594e-02 5.42676747e-01 -7.75108039e-01 7.70398319e-01 6.85626924e-01 -7.74024725e-01 -1.12759516e-01 -3.70644242e-01 2.52977312e-01 6.02180660e-02 3.86231512e-01 -9.95642841e-01 6.07495248e-01 7.46802270e-01 3.54019523e-01 -1.01288128e+00 7.98595548e-01 -3.54406297e-01 7.24769473e-01 -1.14921451e-01 -4.01581764e-01 2.29656637e-01 -1.96514368e-01 3.50564301e-01 1.32624304e+00 5.42160988e-01 -3.46957535e-01 5.08310422e-02 1.40523827e+00 -3.10390592e-01 2.21451491e-01 -6.19695902e-01 -1.60358176e-01 5.74579895e-01 8.50111544e-01 -3.68634284e-01 -6.03648484e-01 -3.98516387e-01 9.64339972e-01 5.61662316e-01 5.08504152e-01 -7.39584327e-01 -6.45622373e-01 2.60695755e-01 2.49100894e-01 3.96144539e-01 9.47560817e-02 -5.00091791e-01 -9.79857266e-01 4.38606560e-01 -1.39828610e+00 2.25755021e-01 -8.30964684e-01 -1.56160998e+00 8.60123456e-01 3.34054917e-01 -9.13813829e-01 -4.78049606e-01 -5.63188493e-01 -5.99660516e-01 9.52846229e-01 -1.10061169e+00 -1.12990665e+00 -2.14955688e-01 2.16221616e-01 8.33960474e-01 -3.01324606e-01 7.73842812e-01 7.21206516e-02 -5.03124177e-01 8.38240981e-01 -2.08285600e-01 8.90510827e-02 7.14567244e-01 -1.46421456e+00 9.26666915e-01 6.28509283e-01 1.20466530e-01 9.46355581e-01 7.54751384e-01 -6.54824257e-01 -9.27475870e-01 -1.33641124e+00 9.62988913e-01 -3.61815274e-01 6.01269901e-01 -8.30836475e-01 -1.27010643e+00 8.75767171e-01 4.00620282e-01 -2.75893778e-01 6.43692255e-01 -1.32468836e-02 -4.56641138e-01 -1.94111899e-01 -1.12619507e+00 1.01922774e+00 1.27296507e+00 -3.88976336e-01 -1.08492494e+00 2.50861973e-01 1.18097782e+00 -4.85044628e-01 -4.47440982e-01 7.80007184e-01 1.83711529e-01 -1.02194619e+00 6.84290767e-01 -8.05593669e-01 8.35093379e-01 -1.90834656e-01 -2.32142195e-01 -1.73014569e+00 1.93946421e-01 -1.73554674e-01 2.53764957e-01 1.28178453e+00 9.47185636e-01 -7.33104587e-01 1.12628436e+00 6.25194758e-02 -5.75362980e-01 -1.02062833e+00 -7.21214771e-01 -1.19543445e+00 4.89957780e-01 -3.75534177e-01 8.65007401e-01 9.37406600e-01 8.33346397e-02 7.34698057e-01 2.74598151e-01 -2.47877613e-01 2.99549848e-01 3.94003093e-01 8.49249244e-01 -8.02160084e-01 -3.94908041e-01 -5.89513123e-01 1.03319064e-01 -1.30258286e+00 3.01228493e-01 -1.32182217e+00 1.87326387e-01 -1.36505878e+00 1.69212013e-01 -3.22406143e-01 2.75951952e-01 3.74382377e-01 -4.63467151e-01 1.23353168e-01 1.71392858e-01 -1.55714646e-01 -1.71147063e-01 7.30459332e-01 1.24746978e+00 -3.80197734e-01 -2.47146562e-01 -2.84052710e-03 -6.58696055e-01 3.79973918e-01 9.50609267e-01 -3.13863575e-01 -7.70118117e-01 -3.88698816e-01 3.52602154e-01 -4.11639512e-01 1.85933903e-01 -1.03625536e+00 -1.45512253e-01 4.87522036e-02 1.01728283e-01 -8.75557125e-01 8.33463073e-02 -6.64149404e-01 -7.39313588e-02 4.55695331e-01 -8.08930397e-01 3.02964598e-01 3.98140341e-01 3.85479897e-01 -5.64061880e-01 -6.94494843e-01 5.29913366e-01 -1.28511414e-01 -2.09598228e-01 -4.28098962e-02 -1.31532520e-01 5.60192347e-01 9.64400113e-01 -1.24276683e-01 -4.13886011e-01 -1.08116552e-01 -7.48493075e-01 2.93024033e-01 5.32990396e-01 5.01713216e-01 5.31714082e-01 -1.43683791e+00 -7.72759259e-01 4.63907838e-01 4.15507525e-01 5.03700495e-01 -3.67356651e-02 4.34884727e-01 -5.83113432e-01 3.55448693e-01 1.40948640e-02 -6.31176770e-01 -1.04804587e+00 8.14064443e-01 2.36328915e-01 -1.51733086e-01 -3.49898458e-01 8.48541677e-01 3.55121374e-01 -1.07827199e+00 2.69317061e-01 -9.17657316e-01 1.23370275e-01 2.24980395e-02 1.70438737e-01 1.38410836e-01 1.19092412e-01 -7.19617009e-02 5.82956187e-02 2.57382661e-01 -3.09705079e-01 -2.66353518e-01 1.19106233e+00 2.51733124e-01 -1.25701323e-01 5.93972087e-01 1.15005445e+00 1.53003976e-01 -7.21520066e-01 -2.24952355e-01 3.64768893e-01 -1.83520839e-01 -7.99700022e-01 -1.00180447e+00 -7.42157578e-01 8.45014155e-01 1.32009834e-01 5.09105146e-01 9.56087232e-01 8.97108484e-03 9.61039066e-01 4.94381070e-01 1.62786186e-01 -9.27553236e-01 2.81013131e-01 6.95209503e-01 1.28439760e+00 -1.02372611e+00 -3.35397005e-01 -4.88847136e-01 -7.33123004e-01 9.11917031e-01 9.35818553e-01 -7.50523573e-03 3.53458291e-03 8.26184079e-02 3.63563970e-02 1.90559492e-01 -8.25609326e-01 1.30384922e-01 1.43077567e-01 4.86915022e-01 5.67869008e-01 -4.88929339e-02 -3.35302353e-01 7.84650087e-01 -8.79101455e-01 -1.89780280e-01 2.09202871e-01 9.18438971e-01 -2.55442321e-01 -1.51219344e+00 -9.36615169e-02 5.49932003e-01 -1.48511380e-01 -4.65528905e-01 -8.03046703e-01 1.00249839e+00 1.07001401e-01 5.64802349e-01 7.80223310e-02 -4.30967420e-01 6.19709313e-01 5.88553369e-01 5.07135868e-01 -1.02810705e+00 -7.23445535e-01 -4.33903724e-01 2.29285493e-01 -2.96420485e-01 -1.47576518e-02 -7.08534181e-01 -1.35412550e+00 2.09809896e-02 1.49881011e-02 2.44644448e-01 5.58610380e-01 7.02700496e-01 1.87855139e-01 7.70267963e-01 3.82152051e-01 -6.55425251e-01 -1.03420639e+00 -1.37479079e+00 -2.81578690e-01 7.96976805e-01 1.50273427e-01 -1.71908453e-01 -4.80133384e-01 -5.90906329e-02]
[11.252366065979004, 9.009405136108398]
f03d4101-0439-49fc-88b5-1142d7d3afb7
iiitt-lt-edi-eacl2021-hope-speech-detection
2104.09066
null
https://arxiv.org/abs/2104.09066v1
https://arxiv.org/pdf/2104.09066v1.pdf
IIITT@LT-EDI-EACL2021-Hope Speech Detection: There is always Hope in Transformers
In a world filled with serious challenges like climate change, religious and political conflicts, global pandemics, terrorism, and racial discrimination, an internet full of hate speech, abusive and offensive content is the last thing we desire for. In this paper, we work to identify and promote positive and supportive content on these platforms. We work with several transformer-based models to classify social media comments as hope speech or not-hope speech in English, Malayalam and Tamil languages. This paper portrays our work for the Shared Task on Hope Speech Detection for Equality, Diversity, and Inclusion at LT-EDI 2021- EACL 2021.
['Bharathi Raja Chakravarthi', 'Sajeetha Thavareesan', 'Ruba Priyadharshini', 'Adeep Hande', 'Karthik Puranik']
2021-04-19
iiitt-lt-edi-eacl2021-hope-speech-detection-1
https://aclanthology.org/2021.ltedi-1.13
https://aclanthology.org/2021.ltedi-1.13.pdf
null
['hope-speech-detection']
['natural-language-processing']
[-3.83401334e-01 3.80449623e-01 -6.34102762e-01 9.40471292e-02 -4.86318409e-01 -6.69138551e-01 1.24267375e+00 5.19807696e-01 -2.49443561e-01 9.38283741e-01 1.27200878e+00 -6.24796867e-01 2.15124905e-01 -5.51914513e-01 3.39501172e-01 -9.50839892e-02 1.36733890e-01 1.27658576e-01 -3.17944407e-01 -9.63733256e-01 6.06255114e-01 2.28598759e-01 -7.67036855e-01 4.42398608e-01 1.03919268e+00 3.89324397e-01 -7.49763668e-01 7.98183203e-01 -3.40917170e-01 1.58420324e+00 -8.17026019e-01 -9.06785488e-01 -1.82851002e-01 -4.27541018e-01 -1.04185593e+00 -3.98398340e-01 3.59646082e-01 -2.85182357e-01 -2.59371430e-01 1.41999924e+00 9.32083845e-01 -2.42451698e-01 5.92747211e-01 -1.23488545e+00 -8.86234403e-01 9.30212796e-01 -8.11824679e-01 3.75062704e-01 6.62090361e-01 1.61606316e-02 5.91743946e-01 -1.01570714e+00 9.34206188e-01 1.50030971e+00 6.78694069e-01 5.52291572e-01 -8.15022767e-01 -8.89360547e-01 -5.03606379e-01 -3.24802279e-01 -1.16161287e+00 -7.95379996e-01 7.39475608e-01 -8.60802531e-01 6.85181022e-01 7.96673894e-01 8.15082312e-01 1.57789969e+00 1.65828303e-01 4.06307250e-01 1.22134387e+00 -1.23890430e-01 -1.86003912e-02 6.02032244e-01 -3.14018756e-01 5.93271911e-01 -1.58620343e-01 -2.91109085e-01 -6.97148144e-01 -7.70863175e-01 -1.87732279e-01 -7.46645555e-02 -1.32235378e-01 7.65151501e-01 -9.30205166e-01 1.49717987e+00 2.65077859e-01 4.52295154e-01 -1.97308213e-01 -5.85050464e-01 7.49343753e-01 2.84248769e-01 1.28404629e+00 3.26212674e-01 4.56980973e-01 -7.53313065e-01 -1.13338470e+00 2.10307259e-02 8.89352977e-01 3.38422805e-01 2.03525141e-01 2.91561056e-02 3.52808684e-02 1.16627729e+00 3.86561006e-01 7.65825152e-01 3.03343773e-01 -7.67948270e-01 4.96271759e-01 2.94066072e-01 7.32827485e-02 -1.73108220e+00 -5.30325413e-01 -4.19727385e-01 -1.07036829e+00 -1.55325234e-01 -1.05301887e-01 -6.41788602e-01 -4.27088529e-01 1.33449435e+00 3.07475150e-01 -3.75423729e-01 -5.78212962e-02 6.54972494e-01 9.67773795e-01 9.82672453e-01 2.64059871e-01 -5.12804747e-01 1.01265728e+00 -7.31860816e-01 -1.19147503e+00 -2.19439924e-01 6.88498616e-01 -1.39362526e+00 7.19231308e-01 1.31689087e-01 -1.09521866e+00 1.89038306e-01 -5.69616795e-01 -1.76092938e-01 -6.82974160e-01 -4.99866784e-01 2.16158509e-01 1.18402779e+00 -1.02519035e+00 3.40194762e-01 3.27058062e-02 -7.38927782e-01 6.92699850e-01 -3.71725500e-01 -3.61511618e-01 3.36405605e-01 -1.35986006e+00 1.05500758e+00 -4.50444734e-03 -2.00667724e-01 -4.01569963e-01 -3.96362424e-01 -5.80529451e-01 -5.22951126e-01 -1.32243857e-01 5.81617691e-02 6.76271260e-01 -8.51977944e-01 -1.03360975e+00 1.14493668e+00 3.11833262e-01 -2.05974683e-01 4.45576578e-01 -2.75040984e-01 -9.52452004e-01 -7.20847920e-02 1.55260488e-01 5.02805471e-01 8.61242890e-01 -9.41583931e-01 -4.04311299e-01 -5.28685510e-01 -1.27222672e-01 -4.96400632e-02 -9.06373680e-01 1.01191819e+00 8.65895212e-01 -6.64241970e-01 -3.53380769e-01 -6.97007477e-01 2.89679579e-02 -4.82773632e-01 -8.29184234e-01 -2.15585832e-03 1.53131080e+00 -1.29656565e+00 1.60677242e+00 -2.34494972e+00 -1.82467461e-01 -8.90665129e-02 6.82409763e-01 5.27479351e-01 3.17429960e-01 1.08637512e+00 1.85874745e-01 1.06782150e+00 3.63774210e-01 5.80241494e-02 9.00068954e-02 1.19641311e-01 -4.63623106e-01 7.67540574e-01 -7.50802606e-02 4.49190944e-01 -1.21334338e+00 -5.87837040e-01 -4.96290363e-02 5.43042421e-01 -4.69339222e-01 -1.15100764e-01 4.23046380e-01 4.34913218e-01 -3.03597361e-01 6.23235643e-01 6.32158995e-01 -8.46067518e-02 1.21424809e-01 1.99948862e-01 -8.56712878e-01 3.91682297e-01 -1.61226302e-01 8.54016006e-01 -1.24029487e-01 9.62306499e-01 4.38501924e-01 -2.37462312e-01 1.09658158e+00 2.59457290e-01 3.44117701e-01 -7.16542363e-01 4.26124394e-01 3.55624497e-01 1.89207681e-02 -5.35825908e-01 1.07798457e+00 -3.28367412e-01 -2.52906412e-01 3.36315423e-01 -3.31583768e-01 -1.95944265e-01 -4.29744363e-01 7.89865196e-01 6.77718580e-01 -7.52677321e-01 4.92982656e-01 -5.78007460e-01 4.61363852e-01 -2.76182264e-01 1.69269189e-01 3.66441399e-01 -8.82353902e-01 2.35705614e-01 6.76925123e-01 -4.43827003e-01 -1.25689268e+00 -7.35961616e-01 1.46295121e-02 1.17618632e+00 -2.62153387e-01 -5.18566847e-01 -3.98390651e-01 -4.87922847e-01 -4.04149681e-01 1.12654066e+00 -2.32809290e-01 -1.38512701e-01 -2.61804193e-01 -4.42821413e-01 8.84383082e-01 -5.94532013e-01 7.87714779e-01 -6.60602927e-01 -2.66420335e-01 1.03305262e-02 -6.92352772e-01 -8.65359068e-01 -5.48119545e-01 -3.02731782e-01 -2.39589855e-01 -5.89958727e-01 -8.57155263e-01 -4.96273845e-01 9.61698741e-02 1.67453647e-01 7.64762282e-01 -1.51926950e-01 -1.95466533e-01 9.64457318e-02 -4.59425002e-01 -6.39856935e-01 -9.32577312e-01 -3.60977314e-02 6.43829927e-02 -1.48077667e-01 6.06327057e-02 -3.90808493e-01 -4.82899517e-01 -1.27183631e-01 -6.66936040e-01 1.62189499e-01 -3.50805432e-01 5.60392678e-01 -6.55268133e-01 -1.56226814e-01 5.44898450e-01 -8.85882437e-01 9.98253465e-01 -1.12544107e+00 3.26079428e-01 -1.55035421e-01 -2.78992236e-01 -8.05109680e-01 3.11448216e-01 -1.75882071e-01 -1.08821881e+00 -7.51917958e-01 -6.37089312e-01 1.92817435e-01 8.53745341e-02 5.68836927e-01 4.71455693e-01 2.16526296e-02 9.98796046e-01 -1.40247509e-01 6.38954565e-02 -4.21708167e-01 3.97665590e-01 1.57342911e+00 2.67992496e-01 -7.82987941e-03 9.01115775e-01 5.79224050e-01 -5.56171238e-01 -1.44213390e+00 -1.04783845e+00 -4.31250721e-01 -7.71678090e-02 -9.08923507e-01 9.65584099e-01 -8.77956092e-01 -4.26770687e-01 6.41934693e-01 -1.42081022e+00 5.50739951e-02 -7.35543221e-02 1.28244743e-01 1.32352144e-01 7.12443769e-01 -8.52007508e-01 -1.50553358e+00 -7.13634789e-01 -3.53247792e-01 4.98008668e-01 2.18188673e-01 -6.74498379e-01 -1.22180128e+00 3.94496769e-01 7.94643939e-01 1.00109661e+00 6.98195934e-01 7.73148835e-01 -9.05034602e-01 5.69999397e-01 -4.66059940e-03 -2.88747907e-01 3.89184177e-01 4.52959478e-01 3.62056941e-01 -6.53836608e-01 -1.88534468e-01 2.20419616e-01 -9.07297313e-01 2.70502090e-01 -3.08486372e-01 4.17186558e-01 -1.43790829e+00 1.82371765e-01 -2.02207327e-01 1.02882254e+00 -6.49370551e-02 8.75717044e-01 3.68132383e-01 2.83698797e-01 8.25714231e-01 4.43656176e-01 1.24480271e+00 3.46582085e-01 8.74076411e-02 4.55709875e-01 -1.70426071e-01 -8.00320879e-02 -5.70767283e-01 7.49903321e-01 1.37190402e+00 1.03554957e-01 -3.70729804e-01 -1.35258305e+00 5.74205101e-01 -1.31046855e+00 -1.56296980e+00 -5.00074506e-01 1.76986015e+00 8.79187822e-01 2.11959165e-02 4.85466331e-01 1.06059097e-01 7.19450295e-01 8.33297074e-01 2.74551779e-01 -1.25065982e+00 -2.12985307e-01 -1.75395563e-01 4.53347683e-01 7.47087181e-01 -9.49009776e-01 8.00167322e-01 6.76780891e+00 9.65744376e-01 -1.38799202e+00 5.18620253e-01 1.18756855e+00 4.29987628e-03 -7.26104796e-01 -2.35893324e-01 -3.77723008e-01 4.98773873e-01 1.20560229e+00 -4.76320952e-01 2.85958737e-01 6.11108482e-01 6.04151428e-01 1.56146493e-02 3.65872458e-02 8.32086921e-01 3.92251939e-01 -1.43053699e+00 -3.21101844e-01 1.37529671e-01 8.75860333e-01 4.14811611e-01 6.78408206e-01 1.27493426e-01 4.20399457e-01 -9.67629075e-01 6.36958003e-01 9.51758586e-04 6.89161837e-01 -5.82914352e-01 5.48561037e-01 6.05504811e-01 -2.62133002e-01 -7.51407146e-02 -1.67782649e-01 -2.59402156e-01 2.88325399e-01 7.02018440e-01 -5.94838619e-01 4.07164954e-02 4.08373684e-01 7.62933552e-01 -4.49000388e-01 4.93338138e-01 -1.00715995e-01 9.61358190e-01 -2.73580551e-02 -3.29902351e-01 6.32796586e-01 1.17242029e-02 9.90319848e-01 1.56925845e+00 4.20540869e-01 -1.61953568e-01 3.10132373e-02 4.31682736e-01 -1.15273573e-01 5.38302958e-01 -1.10486054e+00 -7.36274302e-01 2.40684569e-01 1.26458514e+00 -3.40999037e-01 -1.45164460e-01 -3.27993304e-01 8.11971843e-01 1.59365535e-01 -4.16175090e-02 -8.95432115e-01 1.02184145e-02 3.47269118e-01 4.26721185e-01 -7.00064600e-01 -1.43838167e-01 -4.50075306e-02 -1.02557349e+00 -8.11838746e-01 -1.22297001e+00 3.75088274e-01 -6.12699866e-01 -1.38223231e+00 3.94274443e-01 -4.12936121e-01 -7.59368598e-01 -4.76860143e-02 -1.46749288e-01 -7.21200645e-01 6.36408150e-01 -1.25294185e+00 -1.02265108e+00 2.33449101e-01 4.44750816e-01 3.39378059e-01 -2.48330742e-01 6.91736102e-01 5.29707193e-01 -4.32471424e-01 3.09711188e-01 5.47186434e-02 1.91130489e-01 7.41882741e-01 -6.60933852e-01 2.08678797e-01 4.94043201e-01 -2.89131194e-01 2.75961339e-01 1.07065558e+00 -8.51108551e-01 -9.36751246e-01 -9.39529538e-01 1.72434843e+00 2.72712726e-02 1.50610220e+00 -3.54210585e-01 -3.64169091e-01 2.80835241e-01 9.11747038e-01 -6.22107506e-01 1.27637339e+00 2.27329299e-01 -6.73901200e-01 5.45827031e-01 -1.49002635e+00 7.21293628e-01 8.67824614e-01 -8.80781949e-01 -3.93733293e-01 1.06767941e+00 5.80901980e-01 3.49145606e-02 -9.47589099e-01 -2.76454687e-01 4.50755328e-01 -1.12355661e+00 5.85140228e-01 -1.00727367e+00 9.15596247e-01 4.63845730e-01 -2.57158577e-01 -1.23460305e+00 -3.53300035e-01 -1.33193016e+00 3.31165522e-01 1.59910905e+00 3.94500464e-01 -7.15767145e-01 5.85323870e-01 5.98764598e-01 7.80267417e-02 -9.15721878e-02 -1.12398303e+00 -4.78903532e-01 5.07621586e-01 2.50349134e-01 -5.03284624e-03 1.79688179e+00 5.43163955e-01 6.13794625e-01 -1.23417294e+00 -2.89480269e-01 3.69193822e-01 -5.94062746e-01 6.91846609e-01 -9.11984801e-01 2.73698777e-01 -6.44834757e-01 -1.88667864e-01 -2.47450560e-01 -2.59956032e-01 -6.75100029e-01 -6.14865243e-01 -1.37965775e+00 6.46641433e-01 -1.34367123e-01 2.77550310e-01 3.20708036e-01 2.08642721e-01 2.90726781e-01 5.39658129e-01 3.23970109e-01 -5.70757747e-01 4.94876593e-01 1.09500575e+00 -4.10843819e-01 3.05338223e-02 -4.50260133e-01 -9.38755751e-01 7.68405616e-01 9.94372129e-01 -5.44971466e-01 2.24304367e-02 1.85814619e-01 9.54890072e-01 1.58240020e-01 1.81387216e-01 -6.46629810e-01 -6.89249784e-02 -5.58483839e-01 -2.97631562e-01 -5.85901082e-01 3.51831585e-01 -4.81599927e-01 2.23526701e-01 9.66344059e-01 -5.20326316e-01 7.97591209e-02 -1.46335185e-01 1.28965648e-02 -2.14382559e-01 -1.69106513e-01 9.97167647e-01 3.95747721e-02 1.77462026e-02 3.48960683e-02 -8.54521036e-01 6.65878415e-01 6.26353443e-01 1.82993457e-01 -1.25382602e+00 -1.19671524e+00 -6.15903497e-01 2.78610121e-02 2.75502294e-01 5.31316340e-01 6.52601480e-01 -1.28696895e+00 -1.54744041e+00 -3.72126609e-01 -1.74141601e-02 -1.19486928e+00 2.04602361e-01 1.00374329e+00 -3.51805836e-01 1.85614428e-03 -3.19672197e-01 1.43678263e-01 -1.38732708e+00 3.74552697e-01 1.83255449e-02 -1.55607179e-01 -7.77254999e-02 6.52417243e-01 -3.90514314e-01 -5.59433937e-01 -1.44993588e-01 7.71797478e-01 -4.16193396e-01 3.95994633e-01 8.26186001e-01 7.96770930e-01 -5.65947473e-01 -1.38868022e+00 -3.70405257e-01 -3.40819180e-01 3.84137407e-02 1.85855273e-02 1.16095650e+00 -3.42498332e-01 -6.89561605e-01 5.61809719e-01 1.47242832e+00 8.65705013e-01 1.86198562e-01 2.68356144e-01 -4.02335338e-02 -8.06115806e-01 2.70922631e-01 -9.33477700e-01 -6.79331779e-01 7.64302135e-01 3.08385432e-01 1.04233944e+00 5.16527176e-01 -1.78263185e-03 8.64549398e-01 -2.94905186e-01 -5.55362999e-02 -1.28067195e+00 2.59857923e-01 5.42863131e-01 1.15520918e+00 -1.27109849e+00 -2.32576253e-03 -4.58389461e-01 -6.30446315e-01 1.06217039e+00 9.32102874e-02 5.42947769e-01 9.19190824e-01 5.66658452e-02 1.97883129e-01 -2.21804157e-01 -4.23622370e-01 1.84779525e-01 -9.81357247e-02 6.58121049e-01 7.85124898e-01 3.41188431e-01 -9.04597402e-01 -3.98993753e-02 -3.16368550e-01 -3.96706134e-01 8.92607152e-01 4.89931464e-01 -9.54184175e-01 -6.77956045e-01 -3.31113815e-01 4.20279741e-01 -1.07417083e+00 -2.61903733e-01 -1.24823451e+00 7.01143384e-01 2.03472953e-02 1.24194014e+00 -1.24836579e-01 -7.53128767e-01 -3.10054809e-01 -1.74781326e-02 -5.07994831e-01 -3.20717126e-01 -1.04583883e+00 6.98628500e-02 1.04120362e+00 -6.76722676e-02 -5.67760527e-01 -6.35117173e-01 -8.36661994e-01 -1.27054226e+00 -1.43339202e-01 2.08099261e-01 8.14315438e-01 4.29651946e-01 3.31453502e-01 -2.01890111e-01 1.08387434e+00 -1.29967779e-01 -4.66573715e-01 -8.54104400e-01 -4.89011705e-01 3.92980188e-01 3.50002289e-01 1.16582550e-02 -5.70343256e-01 -6.52729809e-01]
[8.876163482666016, 10.619806289672852]
b8286124-ef59-466d-87d8-6b3a50d7e824
improving-empathetic-response-generation-by
null
null
https://aclanthology.org/2021.findings-emnlp.70
https://aclanthology.org/2021.findings-emnlp.70.pdf
Improving Empathetic Response Generation by Recognizing Emotion Cause in Conversations
Current approaches to empathetic response generation focus on learning a model to predict an emotion label and generate a response based on this label and have achieved promising results. However, the emotion cause, an essential factor for empathetic responding, is ignored. The emotion cause is a stimulus for human emotions. Recognizing the emotion cause is helpful to better understand human emotions so as to generate more empathetic responses. To this end, we propose a novel framework that improves empathetic response generation by recognizing emotion cause in conversations. Specifically, an emotion reasoner is designed to predict a context emotion label and a sequence of emotion cause-oriented labels, which indicate whether the word is related to the emotion cause. Then we devise both hard and soft gated attention mechanisms to incorporate the emotion cause into response generation. Experiments show that incorporating emotion cause information improves the performance of the model on both emotion recognition and response generation.
['Ruifeng Xu', 'Jiachen Du', 'Yu Cao', 'Wei Wang', 'Haolin Deng', 'YuHan Liu', 'Jun Gao']
null
null
null
null
findings-emnlp-2021-11
['empathetic-response-generation', 'recognizing-emotion-cause-in-conversations']
['natural-language-processing', 'natural-language-processing']
[ 1.51440397e-01 9.65228453e-02 -2.35288486e-01 -1.01787663e+00 -5.42681575e-01 -3.46723557e-01 4.13351297e-01 -1.50627956e-01 -1.91980585e-01 6.54495776e-01 9.84283626e-01 1.51875004e-01 4.05061424e-01 -7.93171644e-01 -5.57059608e-02 -5.05779326e-01 6.47133410e-01 1.95223421e-01 -7.60642946e-01 -6.06302977e-01 3.85250926e-01 9.53003466e-02 -1.10833073e+00 9.84525383e-01 6.96108401e-01 9.16501582e-01 -8.79626349e-02 6.22227430e-01 -3.86990666e-01 1.63848984e+00 -8.38862598e-01 -4.50462013e-01 -3.46021622e-01 -1.35727930e+00 -1.26288438e+00 -2.43617028e-01 -4.68220621e-01 -1.78912625e-01 2.13022515e-01 6.82145953e-01 6.10072374e-01 4.43644345e-01 8.70738447e-01 -1.37932169e+00 -7.41919518e-01 9.82828498e-01 -1.81530342e-01 -1.12029910e-01 8.49220753e-01 -9.98400152e-02 1.07275796e+00 -7.87389994e-01 4.19292927e-01 1.41037822e+00 4.62184459e-01 1.19747806e+00 -7.02740312e-01 -8.70204210e-01 1.32718235e-01 4.66774791e-01 -7.88774729e-01 -3.43692452e-01 1.32862723e+00 -2.05785394e-01 7.79225767e-01 2.47084424e-01 6.44792736e-01 1.43856585e+00 1.18753135e-01 8.14348280e-01 9.73564863e-01 -4.54621285e-01 3.21105272e-01 2.51594335e-01 3.15859437e-01 2.87113130e-01 -9.36646104e-01 -1.36568874e-01 -7.18719900e-01 -2.81406641e-01 3.50966513e-01 -8.42944980e-02 -2.73544312e-01 5.62896967e-01 -8.03845227e-01 1.29318237e+00 7.77595222e-01 2.68676341e-01 -8.47500265e-01 1.91716582e-01 5.79096854e-01 3.53161961e-01 4.34740543e-01 7.81543255e-01 -1.68210298e-01 -5.08599401e-01 -2.31309116e-01 1.06756382e-01 1.03101945e+00 4.37957674e-01 6.20569229e-01 -4.60267961e-02 -2.99237013e-01 1.22762394e+00 1.97591051e-01 3.29789937e-01 5.61151445e-01 -9.95857775e-01 1.48680806e-01 8.14543068e-01 2.09357485e-01 -1.30468321e+00 -6.72364712e-01 1.42139457e-02 -5.93637109e-01 -2.31671423e-01 4.43405770e-02 -4.93361920e-01 -2.71720082e-01 2.11471295e+00 3.66246939e-01 -3.03652771e-02 5.12986422e-01 1.25167060e+00 1.27010226e+00 1.10917580e+00 4.64356363e-01 -2.92124957e-01 1.37095976e+00 -1.15784025e+00 -8.68318200e-01 -4.34400290e-01 9.59366441e-01 -8.35453153e-01 1.29927075e+00 3.18380088e-01 -6.76980615e-01 -4.69677716e-01 -5.49700499e-01 -1.65570229e-01 -4.12629247e-02 2.47112855e-01 7.13904500e-01 1.97257444e-01 -4.39738512e-01 1.73899636e-01 1.43079627e-02 -4.04398561e-01 -5.50417081e-02 -1.40738422e-02 -2.09450915e-01 4.14857604e-02 -1.65117884e+00 9.56890523e-01 9.23540327e-04 1.71883434e-01 -3.38995993e-01 -2.97351301e-01 -9.05834377e-01 1.23789482e-01 -2.64657319e-01 -4.92304742e-01 1.50310278e+00 -1.59872246e+00 -1.90936124e+00 7.76345730e-01 -3.75070035e-01 -6.47784173e-02 -2.42485888e-02 -1.82466641e-01 -5.03037930e-01 1.45443499e-01 -1.59137726e-01 9.44086492e-01 7.00460494e-01 -1.18434024e+00 -4.06599849e-01 -7.20065907e-02 -1.06231458e-01 5.73901951e-01 -4.34199154e-01 5.48593581e-01 3.53098452e-01 -5.32939732e-01 -2.05569386e-01 -8.03255975e-01 -2.18886092e-01 -5.83684742e-01 -1.85428172e-01 -6.36121511e-01 5.50696194e-01 -4.84589428e-01 1.10281765e+00 -1.90922236e+00 -2.98740976e-02 -4.25503217e-02 -1.44230174e-02 -1.17571592e-01 -3.36314321e-01 6.59633815e-01 -3.73382837e-01 -1.89000010e-01 4.88555692e-02 -6.24932647e-02 9.47173014e-02 1.05095826e-01 -9.53478575e-01 -5.23605570e-02 1.33015111e-01 9.43255484e-01 -8.03899527e-01 -2.35964939e-01 -5.13974838e-02 4.26203042e-01 -7.47218490e-01 8.76001298e-01 -2.34868228e-01 5.89798331e-01 -6.34925902e-01 1.43123776e-01 2.29155794e-01 -9.59043577e-02 2.29343213e-03 -1.15449913e-01 1.76212206e-01 5.62756777e-01 -5.55780411e-01 1.19700515e+00 -8.98476005e-01 2.96459764e-01 -1.57970786e-01 -6.26388967e-01 1.56330502e+00 5.57506263e-01 3.29530299e-01 -6.81558430e-01 6.35306656e-01 2.60876548e-02 7.37140998e-02 -8.63405526e-01 4.43994194e-01 -8.36136222e-01 -6.51826918e-01 1.14657998e+00 -5.08172154e-01 -2.08921552e-01 -4.29598331e-01 1.30276650e-01 9.58137035e-01 -1.21276341e-01 3.02108139e-01 2.35507563e-01 6.97631776e-01 1.17619231e-01 7.28501976e-01 2.33002707e-01 -3.33500385e-01 3.02661061e-01 6.07626975e-01 -6.58661067e-01 -4.30848926e-01 -3.53384465e-01 3.69969189e-01 1.55427289e+00 1.44107416e-01 -3.62205476e-01 -6.79434180e-01 -5.70156097e-01 -5.48976183e-01 1.13870072e+00 -8.05081904e-01 -7.51476467e-01 -6.20623589e-01 -5.73646426e-01 4.48690146e-01 6.81869507e-01 2.88022250e-01 -1.95567691e+00 -5.93084395e-01 3.97022665e-01 -8.22143376e-01 -7.79511154e-01 -4.52549815e-01 1.61909401e-01 -2.58286864e-01 -7.36581624e-01 -3.37116718e-01 -9.19358552e-01 5.68301320e-01 9.40307304e-02 9.62209344e-01 5.48553132e-02 5.04558012e-02 2.05474943e-01 -8.46259117e-01 -4.72781986e-01 -5.89593530e-01 -1.76465347e-01 -9.59636867e-02 2.41773695e-01 6.75996780e-01 -3.39950323e-01 -4.09850270e-01 2.34579757e-01 -5.03373742e-01 3.21114361e-01 2.15065449e-01 8.22870851e-01 7.69781396e-02 -4.15062040e-01 1.26340616e+00 -7.98630893e-01 1.30602360e+00 -8.02936256e-01 5.55577219e-01 4.37854789e-02 -2.80006230e-01 -1.52795330e-01 1.06049025e+00 -5.86104751e-01 -1.44282103e+00 3.98803726e-02 -6.37102604e-01 1.85790695e-02 -4.35727477e-01 5.51574051e-01 -1.53520271e-01 3.07536155e-01 7.39698470e-01 3.74835124e-03 -1.43629432e-01 -1.56039357e-01 5.07237434e-01 1.08927965e+00 4.45613831e-01 -8.34175825e-01 -1.30179331e-01 2.19660982e-01 -3.97803545e-01 -3.43474597e-01 -1.26268005e+00 -4.38342988e-01 2.58811004e-02 -5.63201904e-01 9.63300169e-01 -7.85482049e-01 -9.56326783e-01 3.23337585e-01 -1.61996269e+00 -4.71959531e-01 2.01937426e-02 5.68893671e-01 -9.30661321e-01 -7.51574934e-02 -9.71325576e-01 -7.89995134e-01 -6.80795312e-01 -7.82943726e-01 7.57042229e-01 5.02316713e-01 -1.13357949e+00 -9.27406251e-01 2.61371076e-01 5.44269681e-01 3.83608341e-01 1.27369687e-02 1.17935038e+00 -9.41974223e-01 5.33509016e-01 -2.57693917e-01 -2.04966769e-01 7.98407942e-02 5.69882318e-02 -2.80659467e-01 -9.87136185e-01 4.61529225e-01 2.73164660e-01 -8.70611787e-01 4.90049213e-01 -1.05795801e-01 9.27426994e-01 -7.43285358e-01 5.07600941e-02 3.05356979e-01 7.05498517e-01 3.73017222e-01 6.76066399e-01 -1.12859778e-01 5.83526671e-01 1.19545007e+00 8.23500633e-01 7.91870296e-01 6.34527862e-01 5.72888613e-01 2.14642063e-01 -1.45954877e-01 2.20148399e-01 -3.75674874e-01 6.15422428e-01 7.72954702e-01 3.32271695e-01 -2.61762857e-01 -6.33118093e-01 2.93608099e-01 -2.00284791e+00 -1.18138349e+00 -4.72315192e-01 1.44366860e+00 1.13176203e+00 -5.72437882e-01 8.57098773e-02 -8.65502357e-02 7.52574384e-01 1.60817772e-01 -3.31276119e-01 -1.18098211e+00 3.39872949e-02 8.24853480e-02 -6.70447171e-01 6.56080008e-01 -5.93106925e-01 1.36061561e+00 5.82717991e+00 3.13195735e-01 -1.38288939e+00 -7.27771670e-02 6.53699458e-01 1.70886870e-02 -4.31262821e-01 -4.29072119e-02 -4.71114784e-01 3.60162914e-01 6.46444678e-01 -3.40614289e-01 3.05180550e-01 1.13767993e+00 5.37407279e-01 2.56985545e-01 -1.14563501e+00 1.02116346e+00 4.33360398e-01 -7.90295124e-01 2.33163331e-02 -4.17477220e-01 4.46100235e-01 -8.12402248e-01 -2.04382658e-01 5.97868681e-01 3.76173705e-01 -1.18743038e+00 3.84499907e-01 6.61104739e-01 3.10670257e-01 -1.00898576e+00 9.02467966e-01 4.36289877e-01 -7.42793441e-01 -1.97071522e-01 -4.02717590e-01 -7.45036006e-01 4.03104484e-01 1.45210758e-01 -9.60133493e-01 -1.14487678e-01 2.92811930e-01 7.68315852e-01 1.53441494e-02 4.81671989e-01 -9.68049347e-01 7.43501604e-01 1.61724746e-01 -3.73084903e-01 3.16449434e-01 -6.94414526e-02 1.51173905e-01 1.21765065e+00 2.97472686e-01 7.07038403e-01 2.06938162e-01 9.14975703e-01 -1.07174367e-01 7.65629411e-01 -3.95983934e-01 4.24472569e-03 5.09266496e-01 1.56082880e+00 -4.05493408e-01 -1.46555766e-01 3.28758061e-02 1.14871621e+00 5.35888672e-01 2.97189087e-01 -9.75304961e-01 -3.49248737e-01 6.40609980e-01 -4.33062255e-01 -2.07159758e-01 6.26500368e-01 -4.34225112e-01 -7.51085937e-01 -3.70034933e-01 -9.16205227e-01 5.68333447e-01 -1.27282917e+00 -1.60341334e+00 8.25815737e-01 -6.39095724e-01 -9.40709352e-01 -6.46710336e-01 -2.60202140e-01 -1.33315969e+00 9.77993190e-01 -1.09517217e+00 -1.09637105e+00 -4.99876440e-01 5.71212828e-01 4.45807189e-01 6.81629777e-02 1.30987132e+00 -1.23734564e-01 -5.14459610e-01 5.45373857e-01 -8.57658505e-01 1.52888700e-01 1.19528818e+00 -8.02057862e-01 -2.78287947e-01 3.71640176e-01 -1.84588164e-01 6.70827627e-01 7.75510311e-01 -5.59279740e-01 -8.15159380e-01 -9.13330257e-01 1.44532728e+00 -2.50492424e-01 4.34448063e-01 1.18738236e-02 -8.73100221e-01 3.95289719e-01 3.57329726e-01 -3.72724742e-01 1.30663359e+00 3.36263746e-01 -5.27154386e-01 1.60495743e-01 -1.09086287e+00 6.76445544e-01 6.14649951e-01 -4.32242721e-01 -8.62678051e-01 2.42247462e-01 7.13800430e-01 -6.54794574e-02 -5.82920909e-01 5.90487532e-02 4.68366444e-01 -9.26864862e-01 5.02979040e-01 -1.03794169e+00 1.10484695e+00 7.05349594e-02 1.96766295e-02 -1.61334264e+00 -3.00018281e-01 -6.91467881e-01 1.52244866e-01 1.36231422e+00 2.81348675e-01 -3.45951706e-01 6.56368434e-01 1.03524137e+00 -2.37955526e-01 -5.83354533e-01 -5.36548376e-01 -1.02405705e-01 8.86623561e-02 -4.96673584e-01 7.45063305e-01 1.37303901e+00 8.34675550e-01 1.09381855e+00 -8.08359861e-01 -3.65469903e-01 -1.90091908e-01 6.06451094e-01 8.16628575e-01 -9.09798563e-01 -1.27910450e-01 -6.65926456e-01 1.26378253e-01 -9.33685303e-01 8.05656731e-01 -9.33368027e-01 5.15745461e-01 -1.63980150e+00 2.22959504e-01 -4.49161083e-01 -4.93475311e-02 8.17436457e-01 -6.18830264e-01 6.60877898e-02 1.38071299e-01 6.24034666e-02 -4.24385875e-01 8.10871124e-01 1.15951872e+00 1.90214068e-01 -4.07822669e-01 1.22764790e-02 -1.07559144e+00 9.06885564e-01 1.05175543e+00 -6.19930208e-01 -3.48148972e-01 -5.13931289e-02 6.17815137e-01 3.85143846e-01 3.38956565e-01 -5.56434751e-01 2.28338316e-01 -7.01077759e-01 2.12097004e-01 -2.07472846e-01 4.08027768e-01 -6.19811952e-01 -2.13192686e-01 2.74819642e-01 -1.06713831e+00 2.38850713e-02 -3.01997185e-01 1.33995160e-01 -4.31376755e-01 -3.61817598e-01 7.79749513e-01 -3.55500095e-02 -7.23840952e-01 -7.27469251e-02 -5.30870676e-01 2.65389048e-02 9.23579097e-01 1.31522819e-01 -3.61606449e-01 -1.11732054e+00 -5.85157573e-01 2.80718952e-01 1.71604052e-01 7.88403749e-01 7.82938838e-01 -1.63697112e+00 -8.64360273e-01 -1.33578122e-01 3.77247036e-01 -5.35433352e-01 4.83889520e-01 6.12759888e-01 1.08086085e-02 7.94419423e-02 -2.22677946e-01 1.66457772e-01 -1.28285348e+00 5.02847552e-01 3.29054534e-01 -1.36985540e-01 -4.48903173e-01 1.20471871e+00 3.66952300e-01 -6.42501175e-01 2.10004121e-01 2.04772651e-01 -7.69420326e-01 2.03560472e-01 8.64571691e-01 3.85091528e-02 -4.32350367e-01 -8.44597816e-01 -2.34264016e-01 3.77261877e-01 3.81870233e-02 -1.73838198e-01 1.18642461e+00 7.71887461e-03 -3.76848668e-01 4.89641160e-01 1.16324520e+00 1.18409172e-02 -7.10675061e-01 -5.43012731e-02 -2.34455943e-01 -1.32835895e-01 -2.08911046e-01 -1.16842556e+00 -7.80335009e-01 1.04845381e+00 -1.01426385e-01 -2.69210357e-02 1.17339361e+00 9.03798733e-03 1.29701304e+00 3.27379674e-01 9.41994935e-02 -1.15438914e+00 5.33296466e-01 8.75285983e-01 1.25066745e+00 -1.03286505e+00 -4.89559650e-01 -3.49214852e-01 -1.32420063e+00 1.23623800e+00 1.04694784e+00 2.12870985e-02 1.43770397e-01 2.57499427e-01 8.28311741e-01 -2.49650031e-01 -1.02584314e+00 8.11097585e-03 8.43641460e-02 4.05633092e-01 8.65130723e-01 1.98910296e-01 -5.98589003e-01 1.46294773e+00 -6.03155911e-01 -5.61693087e-02 4.36726451e-01 4.39877480e-01 -6.00722492e-01 -1.09476149e+00 -3.61349791e-01 5.39069138e-02 -3.10256839e-01 -1.15806811e-01 -1.20769322e+00 -1.87356435e-02 -1.34230480e-01 1.37150621e+00 -1.80659927e-02 -6.91611886e-01 2.39904314e-01 3.66087079e-01 7.69133046e-02 -7.15782821e-01 -1.11739862e+00 -3.14115912e-01 4.15981263e-01 -5.34411311e-01 -2.59630889e-01 -2.36542135e-01 -1.89471471e+00 -2.95510709e-01 6.11082911e-02 5.18413365e-01 4.76791650e-01 1.10455227e+00 2.76064277e-01 3.33452433e-01 1.21110344e+00 -4.43872064e-01 -4.65233564e-01 -1.06870508e+00 -1.95835426e-01 9.21347141e-01 -2.35145271e-01 -2.54120499e-01 -4.16810751e-01 -4.39460855e-03]
[13.145059585571289, 7.614245414733887]
95f5f8e3-12f9-4c02-9e2a-440fe519d25b
spatio-temporal-tubelet-feature-aggregation
2004.00451
null
https://arxiv.org/abs/2004.00451v2
https://arxiv.org/pdf/2004.00451v2.pdf
Spatio-temporal Tubelet Feature Aggregation and Object Linking in Videos
This paper addresses the problem of how to exploit spatio-temporal information available in videos to improve the object detection precision. We propose a two stage object detector called FANet based on short-term spatio-temporal feature aggregation to give a first detection set, and long-term object linking to refine these detections. Firstly, we generate a set of short tubelet proposals containing the object in $N$ consecutive frames. Then, we aggregate RoI pooled deep features through the tubelet using a temporal pooling operator that summarizes the information with a fixed size output independent of the number of input frames. On top of that, we define a double head implementation that we feed with spatio-temporal aggregated information for spatio-temporal object classification, and with spatial information extracted from the current frame for object localization and spatial classification. Furthermore, we also specialize each head branch architecture to better perform in each task taking into account the input data. Finally, a long-term linking method builds long tubes using the previously calculated short tubelets to overcome detection errors. We have evaluated our model in the widely used ImageNet VID dataset achieving a 80.9% mAP, which is the new state-of-the-art result for single models. Also, in the challenging small object detection dataset USC-GRAD-STDdb, our proposal outperforms the single frame baseline by 5.4% mAP.
['Víctor M. Brea', 'Manuel Mucientes', 'Daniel Cores']
2020-04-01
null
null
null
null
['small-object-detection']
['computer-vision']
[ 1.85630110e-03 -1.77021567e-02 4.73624319e-02 -3.28249723e-01 -8.66863728e-01 -4.15899664e-01 4.97807384e-01 2.53222972e-01 -1.11827207e+00 4.13653404e-01 -2.07345232e-01 2.48282805e-01 1.92574367e-01 -6.41837776e-01 -1.11417687e+00 -6.50477946e-01 -2.87121952e-01 1.83503389e-01 1.13986325e+00 1.01580741e-02 1.35041818e-01 5.65719545e-01 -1.58698535e+00 7.32961953e-01 4.18826848e-01 1.62984097e+00 5.92295170e-01 7.01775014e-01 5.13975359e-02 7.50777066e-01 -5.75240254e-01 -2.81220287e-01 4.15187567e-01 -6.50123656e-02 -6.83569789e-01 1.21914670e-01 7.21386075e-01 -6.37802899e-01 -2.92054385e-01 8.13298523e-01 6.38141274e-01 1.17955916e-01 5.02317011e-01 -9.53228533e-01 -1.58043131e-01 6.62672162e-01 -8.44476104e-01 8.07602406e-01 4.65444066e-02 5.08996606e-01 9.00939643e-01 -1.26347601e+00 7.77644396e-01 1.25381160e+00 6.25361800e-01 3.53550047e-01 -1.05407310e+00 -6.62091792e-01 3.48560452e-01 2.07495943e-01 -1.32885659e+00 -3.79682899e-01 2.43279204e-01 -5.62071919e-01 1.21739256e+00 2.92525478e-02 6.40354633e-01 7.94327199e-01 1.60441890e-01 8.57624948e-01 7.28458345e-01 -1.76884279e-01 5.42514026e-02 1.51559170e-02 1.93121374e-01 8.47263992e-01 1.02760963e-01 1.35530740e-01 -4.73103017e-01 2.43312106e-01 6.84012353e-01 6.26567230e-02 -4.92394753e-02 -2.86956847e-01 -1.30038059e+00 6.69962585e-01 1.08684063e+00 4.51013654e-01 -5.85827053e-01 4.54169095e-01 3.73980671e-01 -1.70962989e-01 4.42359447e-01 2.64022827e-01 -5.63300133e-01 3.22432518e-01 -1.18804133e+00 1.36983946e-01 2.60596693e-01 7.91773260e-01 8.68663013e-01 -1.95048332e-01 -6.59340918e-01 5.34470856e-01 2.70537704e-01 3.93347681e-01 3.09004515e-01 -7.25391328e-01 6.92723632e-01 6.34740770e-01 1.01967715e-01 -7.81322896e-01 -6.05670631e-01 -9.68577981e-01 -5.11408150e-01 2.72221714e-01 5.51882088e-01 -5.32733984e-02 -1.15176475e+00 1.55004621e+00 3.78175944e-01 4.96074885e-01 -2.98556477e-01 1.10753953e+00 9.81377304e-01 6.65528655e-01 2.38597140e-01 -1.06025673e-01 1.77504444e+00 -1.12469411e+00 -3.29136699e-01 -2.63469726e-01 6.83663845e-01 -7.26781785e-01 5.64861298e-01 2.29889899e-01 -1.20776522e+00 -9.57041264e-01 -1.03387105e+00 -1.63054198e-01 -4.39537138e-01 7.34833121e-01 2.76615620e-01 4.68652278e-01 -1.08065701e+00 4.20035392e-01 -9.74088490e-01 -3.53444278e-01 7.60163367e-01 5.67976832e-01 -4.13246363e-01 3.63725275e-02 -7.73230910e-01 8.62644076e-01 8.85586023e-01 1.68760329e-01 -9.72701550e-01 -7.10099339e-01 -8.20899189e-01 -1.14686158e-03 5.63373446e-01 -7.45722771e-01 9.74554539e-01 -6.40680075e-01 -9.31811988e-01 7.00307965e-01 -1.31785303e-01 -9.53839660e-01 6.69509172e-01 -1.58190191e-01 5.35505265e-02 3.16772431e-01 3.09892952e-01 1.32623756e+00 8.81454945e-01 -8.06281269e-01 -1.23811340e+00 -4.07889336e-01 9.81859192e-02 6.16926514e-02 -1.66067705e-01 2.04484627e-01 -9.65069234e-01 -6.27225578e-01 1.68653950e-01 -7.16482580e-01 -3.19750398e-01 1.92069590e-01 -2.13005185e-01 -4.46819812e-01 9.77725267e-01 -4.60632473e-01 1.15756238e+00 -2.20188332e+00 9.90501046e-02 -5.99440783e-02 2.49059722e-01 4.81868446e-01 -3.72603685e-01 -1.24694020e-01 3.88824730e-03 -9.81008485e-02 -6.56351149e-02 -7.62094855e-01 -3.66923958e-01 -1.38492793e-01 -1.98656783e-01 4.30605352e-01 7.15793550e-01 1.04240811e+00 -8.93903017e-01 -5.63064873e-01 4.53135461e-01 3.13293576e-01 -7.41825700e-01 2.91307420e-02 -1.98960051e-01 2.12584615e-01 -4.96614873e-01 4.83828068e-01 6.66892111e-01 -2.37930074e-01 -3.58999580e-01 -6.09594405e-01 -3.59400690e-01 -7.36425491e-03 -1.32692337e+00 1.79769373e+00 -5.11960872e-02 5.76450646e-01 -1.69021979e-01 -8.34600270e-01 6.93958640e-01 3.67960660e-03 5.24511933e-01 -6.93927288e-01 3.67118210e-01 3.39261830e-01 -7.55058825e-02 -4.24929470e-01 4.40290570e-01 4.23737586e-01 1.68051869e-01 2.17466448e-02 5.32268107e-01 1.89908639e-01 5.90064883e-01 2.03974366e-01 1.17839313e+00 1.90546438e-01 8.53781477e-02 -5.71383275e-02 6.86648548e-01 -9.07581151e-02 4.04826939e-01 8.78479540e-01 -3.00300032e-01 9.39105809e-01 4.04780984e-01 -6.07688904e-01 -8.09152782e-01 -8.50719750e-01 -1.03082307e-01 1.26420522e+00 1.48724049e-01 -4.72060204e-01 -7.58039951e-01 -1.05356920e+00 -9.62621272e-02 2.38164067e-01 -8.10426950e-01 1.20702945e-01 -9.58033800e-01 -8.79534662e-01 4.24378604e-01 6.82200730e-01 6.86316252e-01 -1.17604685e+00 -1.08020258e+00 5.30112863e-01 -1.92500606e-01 -1.41593480e+00 -5.28000593e-01 3.74122351e-01 -6.04645669e-01 -1.08165932e+00 -8.99867356e-01 -7.48997450e-01 4.20463324e-01 2.18614712e-01 9.14769769e-01 -3.50559354e-02 -6.91314459e-01 7.13146031e-02 -3.50855052e-01 -3.10814619e-01 8.38915855e-02 1.61996067e-01 -1.99440584e-01 2.49137327e-01 6.86128139e-02 -3.95490490e-02 -9.74068046e-01 5.44515312e-01 -8.20189118e-01 -4.40038443e-02 7.72707522e-01 6.12634301e-01 6.77779078e-01 -2.62385488e-01 3.86722147e-01 -2.89165497e-01 -8.11760128e-02 -2.56223738e-01 -6.85175896e-01 1.14040941e-01 1.15251176e-01 2.10084431e-02 1.46118402e-01 -4.41108793e-01 -6.86051548e-01 4.96202171e-01 -1.65886447e-01 -5.09448886e-01 3.70164886e-02 6.76923618e-02 1.85184881e-01 -1.94621444e-01 7.75390267e-01 1.31464645e-01 -1.71506897e-01 -4.40159559e-01 4.15670544e-01 3.54144335e-01 5.76128066e-01 -2.02913046e-01 5.16318083e-01 6.51501477e-01 -1.12739898e-01 -4.73246455e-01 -8.35189939e-01 -7.47367322e-01 -8.64133179e-01 -2.96054065e-01 1.34317446e+00 -1.00960660e+00 -6.86916590e-01 4.15798008e-01 -1.41385472e+00 -2.52737671e-01 -3.83181870e-01 6.72752261e-01 -4.63002920e-01 1.32949203e-01 -5.45597255e-01 -7.08673477e-01 -3.57332975e-01 -1.34921658e+00 1.38836777e+00 1.21552095e-01 1.59654856e-01 -2.40585029e-01 -2.61861473e-01 1.42723965e-02 2.59928882e-01 1.19904116e-01 1.61161020e-01 -5.36507368e-01 -1.05586886e+00 -1.35373712e-01 -7.61856019e-01 3.70811850e-01 -3.77741754e-01 -1.45101905e-01 -9.55309272e-01 -3.84607613e-01 -2.52337039e-01 -1.00286461e-01 1.54059887e+00 6.35243475e-01 1.01295733e+00 7.05858096e-02 -6.17545485e-01 6.93189383e-01 1.32175934e+00 1.84976961e-02 4.57987398e-01 2.74024099e-01 5.86420655e-01 4.52744961e-01 7.85576046e-01 4.38655049e-01 1.90792575e-01 9.10868764e-01 6.69549525e-01 -1.47874445e-01 -6.00207269e-01 2.51281142e-01 3.45544279e-01 -6.87113777e-02 -1.95096970e-01 -1.96378365e-01 -7.96537101e-01 6.16528034e-01 -1.95781779e+00 -9.79133964e-01 -1.62033200e-01 1.91928506e+00 2.96355069e-01 3.29585642e-01 4.57705110e-01 -3.12540717e-02 7.51892328e-01 1.21424198e-01 -3.06146562e-01 3.04968655e-01 -2.47974768e-01 2.53922790e-01 8.12217891e-01 9.21971202e-02 -1.54239047e+00 1.04447424e+00 5.32671833e+00 9.08144653e-01 -1.18323302e+00 4.10079002e-01 6.77817881e-01 -2.93738306e-01 5.80501497e-01 -3.31059784e-01 -1.25535536e+00 3.96200210e-01 6.33580565e-01 4.85840946e-01 -1.30983695e-01 7.97864020e-01 1.90327048e-01 -4.01657432e-01 -1.13081717e+00 1.01867712e+00 1.88465510e-02 -1.50771081e+00 1.95092391e-02 -1.43446326e-01 5.72910488e-01 4.90530789e-01 2.35636588e-02 3.07006240e-01 -2.39277184e-01 -6.65343344e-01 1.23414862e+00 3.14917058e-01 6.38182819e-01 -5.32869995e-01 8.15284491e-01 3.31668645e-01 -1.69703305e+00 -4.62449610e-01 -2.87752062e-01 2.01095581e-01 2.15489283e-01 4.70574349e-01 -8.97022128e-01 4.71832722e-01 1.01096165e+00 7.22357631e-01 -1.00069165e+00 1.61845219e+00 -2.57255342e-02 2.62727886e-01 -6.56285286e-01 1.01029411e-01 5.46994507e-01 2.81343907e-01 5.54770410e-01 1.57451618e+00 4.90221173e-01 1.75103158e-01 2.36746490e-01 7.82568574e-01 6.69900775e-02 -8.72855559e-02 -1.88795090e-01 5.30136824e-01 1.36403441e-01 1.40480375e+00 -1.06284463e+00 -5.92476666e-01 -2.97626287e-01 9.37495589e-01 4.41262484e-01 2.32276082e-01 -1.07761598e+00 -1.71108902e-01 2.47201979e-01 3.30129534e-01 9.31734204e-01 -1.66210473e-01 8.73953011e-03 -9.79454815e-01 2.99083024e-01 -3.02629918e-01 5.97963154e-01 -8.00327957e-01 -8.25501919e-01 8.99128973e-01 -7.33275665e-03 -1.26047671e+00 -1.28800288e-01 -7.14833140e-01 -3.66170913e-01 7.19881475e-01 -1.49696219e+00 -1.22568452e+00 -3.62404734e-01 5.80138862e-01 7.30471969e-01 2.19494067e-02 2.33336627e-01 5.70345998e-01 -7.14635432e-01 5.44732690e-01 -4.95134175e-01 3.84010464e-01 6.64304435e-01 -9.94931877e-01 4.87191498e-01 1.07954550e+00 2.45158941e-01 2.54374951e-01 3.68853450e-01 -6.13581300e-01 -1.03423345e+00 -1.66369390e+00 6.31601334e-01 -5.53949535e-01 4.58076835e-01 -4.40754473e-01 -7.92365074e-01 6.10188425e-01 -6.19880669e-03 8.27020645e-01 -1.49088532e-01 -4.87046570e-01 -3.05618405e-01 -2.26992920e-01 -1.12583280e+00 9.80068445e-02 1.19452393e+00 -1.09723762e-01 -5.44276655e-01 3.09791744e-01 7.82798886e-01 -4.93769318e-01 -3.90112370e-01 5.86439848e-01 4.43379521e-01 -1.07366300e+00 1.17092395e+00 -2.68211782e-01 1.86139658e-01 -7.51357555e-01 -3.09157986e-02 -8.34745526e-01 -4.15808111e-01 -3.84242505e-01 2.57706200e-03 9.07587409e-01 4.19971079e-01 -4.22217280e-01 8.02168727e-01 6.94930926e-02 -3.36635858e-01 -7.46092439e-01 -1.13739550e+00 -7.95965612e-01 -4.81515348e-01 -6.56105816e-01 2.51468778e-01 2.51918107e-01 -4.71927494e-01 1.96729600e-01 -1.09105133e-01 2.78685927e-01 6.20170414e-01 -1.92335755e-01 6.86763048e-01 -9.13315356e-01 -3.10484678e-01 -5.47497571e-01 -7.39945829e-01 -1.20531261e+00 -3.20365280e-01 -7.55582869e-01 7.73925260e-02 -1.39032865e+00 2.73590535e-01 -3.01566690e-01 -4.48392898e-01 4.23318386e-01 -2.85265833e-01 9.70276535e-01 6.25495136e-01 1.78502798e-02 -1.02487457e+00 2.39495605e-01 1.12771308e+00 1.16451317e-02 -2.35363826e-01 -7.82696456e-02 -3.49245816e-02 6.90183461e-01 3.50773484e-01 -6.28541589e-01 7.37609640e-02 -4.74320024e-01 -2.22254410e-01 -3.38341706e-02 8.69996011e-01 -1.43328786e+00 3.93507540e-01 2.43380800e-01 7.34345317e-01 -1.17686057e+00 5.60968697e-01 -7.48104692e-01 -2.13632733e-01 7.45058596e-01 -7.94829130e-02 -6.20995946e-02 2.68285185e-01 5.88637590e-01 -1.11766174e-01 -1.00607770e-02 1.00949359e+00 2.17079334e-02 -9.98915911e-01 5.19117415e-01 -1.15120351e-01 -1.79853544e-01 1.35658276e+00 -3.21196318e-01 -2.25594923e-01 3.39477628e-01 -9.62001085e-01 2.70305634e-01 -1.85690969e-02 4.64125991e-01 5.66477954e-01 -1.12980807e+00 -8.70878994e-01 2.52022237e-01 1.47328272e-01 5.45403026e-02 3.40186983e-01 1.05210543e+00 -3.82351309e-01 4.75913703e-01 -2.32074708e-01 -1.12571239e+00 -1.25330317e+00 6.61114633e-01 4.77291375e-01 -1.86978117e-01 -6.11539662e-01 1.33659005e+00 5.69336534e-01 2.32485190e-01 3.36214334e-01 -8.45145285e-01 -2.99261302e-01 4.64517623e-01 7.65133142e-01 2.09857762e-01 2.85509109e-01 -7.58635283e-01 -5.76967895e-01 9.03308690e-01 -7.24741146e-02 -1.30383939e-01 1.22982979e+00 -1.19067691e-01 1.41414464e-01 3.64189893e-02 1.35086191e+00 -3.63271803e-01 -1.55874431e+00 -5.01654923e-01 -6.93542957e-02 -4.32477802e-01 3.28028873e-02 -6.25868261e-01 -1.25108182e+00 7.60912836e-01 9.87106502e-01 2.70025209e-02 1.02425158e+00 2.91325510e-01 4.98277247e-01 1.86603501e-01 3.16101849e-01 -7.66617537e-01 3.75980109e-01 5.43670535e-01 9.91657078e-01 -1.35018599e+00 -1.28855973e-01 -5.59630454e-01 -3.41872007e-01 1.06593633e+00 7.32037127e-01 -3.80353093e-01 4.74444687e-01 2.15605780e-01 -2.72651583e-01 -1.50758773e-01 -5.83759010e-01 -8.09151411e-01 6.62688017e-01 3.89088482e-01 1.60359517e-01 -1.79381222e-01 -1.49884552e-01 5.30702472e-01 1.63872436e-01 -5.56828082e-02 1.21078543e-01 6.87719047e-01 -8.14398766e-01 -6.88099384e-01 -4.29605395e-01 3.86680722e-01 -5.07175624e-01 -3.05639021e-02 2.70230547e-02 8.61535311e-01 7.31177390e-01 7.73316443e-01 3.94960403e-01 -1.87100336e-01 4.46651489e-01 -2.36832052e-01 4.60791022e-01 -8.02308857e-01 -8.29015076e-01 3.53868365e-01 -1.37890249e-01 -8.84928882e-01 -3.52059335e-01 -6.13373101e-01 -1.29062116e+00 2.72031128e-01 -4.76775676e-01 -2.25742027e-01 6.28347874e-01 8.45725715e-01 3.56422126e-01 7.96697795e-01 2.70688832e-01 -1.39221239e+00 -2.91270196e-01 -9.93687630e-01 -1.15316495e-01 2.33983427e-01 5.28132915e-01 -7.65647948e-01 -9.50366035e-02 3.15295048e-02]
[8.85002326965332, -0.1820325404405594]
e04e7397-cbc1-49c5-8916-2bdaec09ff48
overprompt-enhancing-chatgpt-capabilities
2305.14973
null
https://arxiv.org/abs/2305.14973v1
https://arxiv.org/pdf/2305.14973v1.pdf
OverPrompt: Enhancing ChatGPT Capabilities through an Efficient In-Context Learning Approach
The exceptional performance of pre-trained large language models has revolutionised various applications, but their adoption in production environments is hindered by prohibitive costs and inefficiencies, particularly when utilising long prompts. This paper proposes OverPrompt, an in-context learning method aimed at improving LLM efficiency and performance by processing multiple inputs in parallel. Evaluated across diverse datasets, OverPrompt enhances task efficiency and integrates a diverse range of examples for improved performance. Particularly, it amplifies fact-checking and sentiment analysis tasks when supplemented with contextual information. Synthetic data grouping further enhances performance, suggesting a viable approach for data augmentation.
['Lin Gui', 'Yulan He', 'Runcong Zhao', 'Jiazheng Li']
2023-05-24
null
null
null
null
['sentiment-analysis']
['natural-language-processing']
[ 3.79700750e-01 -4.80921566e-02 -4.12411541e-01 -5.05904198e-01 -9.22986746e-01 -6.01963639e-01 9.89595175e-01 6.06471062e-01 -8.10943842e-01 7.02265501e-01 1.87086120e-01 -5.74395418e-01 1.83455557e-01 -4.34117496e-01 -4.32656199e-01 -4.30737853e-01 2.61010174e-02 4.33523059e-01 -8.67406279e-02 -3.04277152e-01 3.63054246e-01 4.07592773e-01 -1.52760494e+00 7.64089704e-01 7.82586575e-01 6.90350652e-01 2.23676458e-01 6.58113062e-01 -6.78500235e-01 9.21948612e-01 -9.70910847e-01 -5.74195325e-01 8.37489814e-02 -1.62678510e-02 -6.63817823e-01 4.71930690e-02 4.31303620e-01 9.55910832e-02 1.42246068e-01 4.32089895e-01 5.73364377e-01 1.58699796e-01 2.42041439e-01 -1.11005735e+00 -4.72672850e-01 5.51590919e-01 -5.89732111e-01 3.95918548e-01 3.29530150e-01 1.99397057e-01 1.03736162e+00 -1.07048273e+00 3.58452111e-01 1.02444172e+00 8.42241287e-01 2.95401931e-01 -1.22851169e+00 -4.60153013e-01 3.21900755e-01 -1.79577515e-01 -1.04310572e+00 -6.10143006e-01 6.15216792e-01 -2.48570857e-03 1.64869535e+00 4.11769897e-01 3.14435720e-01 1.17181492e+00 -4.26104963e-02 1.10445011e+00 1.25331795e+00 -8.75646234e-01 -6.54500946e-02 3.66761476e-01 7.81189725e-02 2.95848131e-01 1.86213359e-01 -1.54949456e-01 -6.73499763e-01 -8.09266120e-02 4.03113365e-01 6.56082034e-02 2.10881263e-01 3.44950967e-02 -1.32972729e+00 6.11213505e-01 -1.42365163e-02 6.33095086e-01 -3.01545292e-01 -1.22267820e-01 8.29471529e-01 4.81588751e-01 5.88960588e-01 1.02403462e+00 -8.62290502e-01 -3.97426218e-01 -1.02484930e+00 3.54008079e-01 7.30017900e-01 8.77903938e-01 5.13398409e-01 3.57783377e-01 -2.13285998e-01 1.17406070e+00 -2.26262525e-01 4.53954905e-01 7.25292325e-01 -6.73368871e-01 7.78097093e-01 9.14593756e-01 -5.89503013e-02 -7.50112832e-01 -7.18304873e-01 -5.79807162e-01 -7.76509643e-01 -9.47049409e-02 2.25711673e-01 -1.64531365e-01 -7.75578201e-01 1.65072119e+00 4.01079021e-02 -1.70118615e-01 1.31439388e-01 2.88967282e-01 6.92308962e-01 7.02249348e-01 6.70383036e-01 -2.39632487e-01 1.22331202e+00 -1.24482679e+00 -8.64342034e-01 -8.31004143e-01 1.24962068e+00 -1.01775706e+00 1.63899708e+00 6.79030776e-01 -1.00794530e+00 -6.43302977e-01 -7.75346577e-01 -1.01690158e-01 -6.67880714e-01 2.21479163e-01 1.11908567e+00 7.02990830e-01 -9.49725449e-01 3.27126533e-01 -4.90437508e-01 -2.39828154e-01 4.47158128e-01 4.82875168e-01 -3.57116818e-01 -1.27039507e-01 -1.08950150e+00 9.09146905e-01 5.38606346e-01 -1.80319473e-01 -1.43338040e-01 -7.00073957e-01 -9.94446695e-01 -7.99589604e-02 5.10239601e-01 -4.23167288e-01 1.42445803e+00 -7.23279059e-01 -1.23459470e+00 7.82839179e-01 -1.77749410e-01 -5.47138333e-01 3.23821813e-01 -4.53111887e-01 -8.15590918e-01 -2.25319698e-01 -7.10428804e-02 7.39069879e-01 6.94128573e-01 -9.52675045e-01 -6.85612857e-01 -1.45593598e-01 -9.10741240e-02 3.69585454e-01 -8.52125406e-01 3.54175329e-01 -4.70270574e-01 -8.71896982e-01 -2.40895361e-01 -6.49146736e-01 -5.68563044e-01 -6.42787874e-01 -2.31484830e-01 -2.71206468e-01 9.24749494e-01 -4.50486749e-01 1.60708129e+00 -1.94271946e+00 -5.24130285e-01 8.99801329e-02 4.51986529e-02 7.00291991e-01 -3.42455029e-01 5.41357100e-01 -9.21237320e-02 3.02390575e-01 2.44262572e-02 -5.22592425e-01 -9.86259505e-02 3.78596544e-01 -4.23610449e-01 -1.58269495e-01 5.47833502e-01 1.14166534e+00 -8.93432736e-01 -4.63731855e-01 2.81550705e-01 2.26052478e-01 -6.72940433e-01 1.12632394e-01 -3.95080268e-01 2.93019414e-01 -1.91653073e-01 5.96467912e-01 3.40029955e-01 -4.63489324e-01 5.51325157e-02 1.97518975e-01 -6.86047375e-02 5.86886048e-01 -1.15340400e+00 1.64971149e+00 -9.60931659e-01 6.49430275e-01 -2.97770202e-01 -8.15110207e-01 9.59526718e-01 2.24182367e-01 2.51812577e-01 -9.58801568e-01 -4.54774797e-02 1.82148650e-01 -1.36211857e-01 -6.14164054e-01 1.11019230e+00 -1.07099831e-01 -3.32326710e-01 5.47747672e-01 -9.20260623e-02 -3.46363038e-01 5.60195208e-01 3.98295701e-01 9.39164400e-01 8.71278718e-02 4.85412598e-01 -4.58302610e-02 4.62604970e-01 1.20044602e-02 1.43711850e-01 8.72868896e-01 1.83042303e-01 2.05400556e-01 2.86324143e-01 -4.06499803e-01 -1.05839276e+00 -5.01843333e-01 -1.35854200e-01 1.68089509e+00 -4.61916089e-01 -7.59253144e-01 -3.69945437e-01 -8.47442985e-01 1.33014135e-02 9.45123255e-01 -2.20709339e-01 -3.07344794e-02 -6.39095008e-01 -9.67449307e-01 5.88548958e-01 7.83239841e-01 1.73722222e-01 -1.26357377e+00 -3.77831489e-01 3.13092649e-01 -2.24251539e-01 -1.35172808e+00 -2.25266442e-01 5.41526914e-01 -1.00797129e+00 -6.31373048e-01 -1.77982613e-01 -7.22117782e-01 7.64476418e-01 2.27436468e-01 1.47388017e+00 2.29442433e-01 -2.26928189e-01 3.45140994e-02 -3.14102829e-01 -6.81078315e-01 -6.01025581e-01 3.44430000e-01 1.41886443e-01 -4.34205741e-01 6.95816219e-01 -3.36757749e-01 -4.93922457e-02 6.71244785e-02 -9.41698730e-01 2.52192505e-02 9.59115446e-01 9.55106497e-01 3.23051631e-01 -1.09509006e-01 9.80129480e-01 -1.40290487e+00 9.86569524e-01 -3.06355596e-01 -1.81468502e-01 1.71926603e-01 -9.66401875e-01 7.39304423e-02 9.62045550e-01 -5.90585232e-01 -1.29124749e+00 -2.23856300e-01 -3.04696798e-01 9.85560492e-02 -3.34044218e-01 5.83953023e-01 6.70543697e-04 5.42276688e-02 9.09187257e-01 1.53484195e-02 -1.96471840e-01 -5.11451542e-01 4.97136295e-01 7.51387537e-01 5.70798278e-01 -7.32509792e-01 4.65557128e-01 3.14838961e-02 -3.37295353e-01 -8.67804706e-01 -9.42475319e-01 -6.14354074e-01 -5.29177487e-01 1.43674120e-01 2.82103084e-02 -9.55500543e-01 -3.84798169e-01 3.79320562e-01 -9.27545547e-01 -3.35800111e-01 -4.25561041e-01 2.47080237e-01 -2.37703070e-01 2.40129620e-01 -5.53654194e-01 -8.42126787e-01 -2.96843439e-01 -8.33378732e-01 8.39536011e-01 6.51173070e-02 -7.73488104e-01 -1.19355226e+00 -1.54394642e-01 4.51776296e-01 4.26919103e-01 -6.54610917e-02 7.79560745e-01 -1.14340889e+00 -1.76823795e-01 -6.44149125e-01 -5.06497286e-02 5.35286963e-01 1.23515464e-01 -1.91693202e-01 -1.21903801e+00 -2.28915930e-01 -2.95323431e-01 -7.84036458e-01 5.67394853e-01 -1.21377409e-01 1.34167385e+00 -4.82172996e-01 -2.93817014e-01 3.47164422e-01 1.08768559e+00 1.89779446e-01 2.32341230e-01 4.88860548e-01 5.52701056e-01 6.44740403e-01 7.16232181e-01 3.87536556e-01 1.93961248e-01 3.43936235e-01 -5.46478927e-02 -3.98357540e-01 -1.28928155e-01 -3.01449776e-01 1.45781428e-01 1.04376853e+00 2.37233251e-01 -3.84440929e-01 -1.12507248e+00 7.53378272e-01 -1.50762463e+00 -7.31975734e-01 -1.65857315e-01 2.03485584e+00 1.18473065e+00 5.59888303e-01 -6.25037029e-02 3.71589392e-01 2.24393159e-01 2.88967907e-01 -2.73392767e-01 -7.78243303e-01 -4.57178831e-01 4.30824310e-01 2.69213319e-01 3.42536569e-01 -9.75887060e-01 1.05688977e+00 7.59530926e+00 9.59498942e-01 -1.01993799e+00 9.15114284e-02 7.26688147e-01 -2.95060366e-01 -3.66494387e-01 -2.13911921e-01 -1.02315664e+00 3.64855468e-01 1.27550924e+00 -1.39185507e-02 5.61089478e-02 9.59728658e-01 1.37961701e-01 -2.89350122e-01 -1.16139185e+00 7.76634753e-01 2.02859506e-01 -1.40540504e+00 1.93588987e-01 -1.98222399e-01 9.22692358e-01 2.78153662e-02 3.25337768e-01 7.78564215e-01 3.57582390e-01 -9.36984837e-01 3.59834790e-01 -4.12136763e-02 8.38023067e-01 -8.88918519e-01 8.35595071e-01 5.31462908e-01 -8.16717505e-01 -2.26711392e-01 -5.99724846e-03 -5.13244271e-01 1.08366400e-01 4.81634706e-01 -1.27702928e+00 2.78674066e-01 4.48097408e-01 2.98739344e-01 -8.64449859e-01 5.31180680e-01 -3.04192126e-01 6.65441573e-01 -3.80887270e-01 -1.20594107e-01 4.13933694e-01 1.19753264e-01 1.72393143e-01 1.88037038e+00 -1.28906310e-01 -1.33311376e-01 3.01227778e-01 1.51625350e-01 -1.89902097e-01 3.96154284e-01 -7.40872383e-01 -3.82631034e-01 5.32646716e-01 1.40352488e+00 -6.22621536e-01 -6.47238791e-01 -5.48676014e-01 4.85798717e-01 4.26558524e-01 2.85433263e-01 -4.92298931e-01 -4.33516860e-01 3.25206012e-01 8.93356130e-02 -5.48347197e-02 -2.42067069e-01 -8.24603140e-01 -9.90907252e-01 6.39120564e-02 -1.30795681e+00 5.68011582e-01 -6.43736184e-01 -9.79765117e-01 6.53225183e-01 -5.28609455e-02 -1.02476895e+00 -5.49172580e-01 -5.38613379e-01 -4.71888721e-01 8.26320112e-01 -1.60323799e+00 -1.37145948e+00 -5.08738309e-02 2.54888594e-01 7.69455910e-01 -2.57012814e-01 9.88513649e-01 4.85551387e-01 -6.22607410e-01 9.92468119e-01 2.81395502e-02 -1.04237728e-01 9.93043780e-01 -1.16364908e+00 6.73961341e-01 9.76635158e-01 3.50665718e-01 8.98792386e-01 6.11816525e-01 -5.18380105e-01 -1.06462812e+00 -1.20790386e+00 1.49569774e+00 -8.00125659e-01 8.66657376e-01 -4.54687834e-01 -1.06434810e+00 6.81857944e-01 1.26545161e-01 -1.50998324e-01 8.61970067e-01 5.69187880e-01 -3.14977437e-01 -1.27512038e-01 -9.41055417e-01 8.02027583e-01 8.14378917e-01 -6.24427021e-01 -7.63523757e-01 4.37800467e-01 7.61063993e-01 -4.69782829e-01 -7.89230943e-01 4.79157925e-01 1.28281459e-01 -5.30179322e-01 1.00521493e+00 -9.15150046e-01 6.15769446e-01 1.52221844e-01 2.78516617e-02 -1.20972240e+00 -8.23576525e-02 -9.17018175e-01 -3.70995075e-01 1.44471776e+00 7.32242405e-01 -5.66091061e-01 7.62865126e-01 6.92726195e-01 -1.80854917e-01 -9.69388008e-01 -3.24776053e-01 -7.26330042e-01 -6.37829080e-02 -9.29284453e-01 6.30793750e-01 1.13607991e+00 1.96819067e-01 5.85959613e-01 -3.97160858e-01 -2.00568661e-01 1.56827211e-01 1.19676869e-02 7.98332810e-01 -7.95532703e-01 -2.65892982e-01 -4.64445055e-01 7.10298121e-02 -1.01938999e+00 3.02790347e-02 -9.21698868e-01 -1.57987610e-01 -1.16859663e+00 -8.23905766e-02 -7.70559311e-01 -3.39633256e-01 7.78316915e-01 -5.39361596e-01 4.60693628e-01 2.02819943e-01 1.18745953e-01 -8.25543582e-01 1.16893679e-01 9.13782477e-01 8.78896490e-02 -3.85988951e-01 -1.36022950e-02 -9.52396095e-01 8.07092071e-01 9.80352283e-01 -3.35955441e-01 -4.41253960e-01 -4.56358880e-01 4.68997449e-01 -3.40737134e-01 -5.82210906e-02 -7.93747246e-01 2.08452314e-01 -6.35070428e-02 4.72860724e-01 -6.11218035e-01 2.92109311e-01 -5.95001280e-01 -3.90021533e-01 9.51705351e-02 -6.52562737e-01 5.24450779e-01 7.18671978e-01 3.28888625e-01 -3.55505347e-01 -2.21361786e-01 6.29892528e-01 -1.75427496e-01 -9.88234162e-01 -1.86320227e-02 -3.54301065e-01 3.30801040e-01 7.10874856e-01 -2.07368374e-01 -3.08536768e-01 -4.55376990e-02 -4.63774830e-01 2.25132257e-01 2.59003729e-01 5.57928324e-01 2.98181683e-01 -1.19712937e+00 -4.46609378e-01 6.55003548e-01 3.83147717e-01 1.34130195e-01 1.22734077e-01 7.85931587e-01 -1.88510597e-01 6.45820796e-01 1.17397904e-01 -5.56824028e-01 -1.35915852e+00 6.15430892e-01 -2.85673201e-01 -7.15490341e-01 -4.97584999e-01 9.50745344e-01 -2.07828984e-01 -5.76808453e-01 3.87641132e-01 -3.71133476e-01 -2.99710989e-01 1.93714499e-01 7.63954461e-01 1.94523349e-01 4.98118877e-01 -1.35456279e-01 -6.98964000e-02 9.22117233e-02 -4.76623684e-01 8.49331766e-02 1.29886889e+00 6.18941039e-02 -8.09032768e-02 5.24232328e-01 9.25036490e-01 2.40527511e-01 -1.04185712e+00 -5.46058476e-01 5.13168514e-01 -4.39325452e-01 5.41104227e-02 -1.23629022e+00 -5.64193964e-01 7.98594832e-01 -6.96735829e-02 2.25190982e-01 1.20300329e+00 -2.00397745e-01 7.95760274e-01 8.57031345e-01 2.74635017e-01 -1.26610434e+00 3.30924749e-01 6.79588139e-01 7.11420000e-01 -1.45625830e+00 5.54261804e-02 -9.35605764e-02 -9.36980009e-01 8.98441911e-01 7.88449705e-01 3.65574896e-01 1.93301648e-01 7.31827259e-01 2.80021071e-01 8.45467448e-02 -1.08645189e+00 8.13219622e-02 1.63209200e-01 4.49587464e-01 7.60352910e-01 -1.00345746e-01 -2.53306329e-01 5.59073746e-01 -3.03613305e-01 -1.17534228e-01 4.05614585e-01 1.20163918e+00 -2.53378242e-01 -1.46447945e+00 -3.33240449e-01 7.14400649e-01 -6.59889281e-01 -4.53553766e-01 -1.74725935e-01 1.01093459e+00 1.66701272e-01 8.70373547e-01 4.82514650e-02 -1.41780898e-01 3.63264531e-01 5.39393008e-01 1.48303956e-01 -8.53051662e-01 -1.08895826e+00 5.86568862e-02 5.00471413e-01 -3.39368582e-01 -2.16355771e-01 -6.03032112e-01 -1.19200289e+00 -2.14738742e-01 -2.50585020e-01 2.13883996e-01 8.51978958e-01 9.24076915e-01 4.99818921e-01 6.21713817e-01 4.76250440e-01 -5.01373947e-01 -8.71556997e-01 -1.13236690e+00 -1.60916567e-01 4.92968321e-01 3.15511256e-01 -2.59091765e-01 -2.05260247e-01 1.01974271e-01]
[10.803539276123047, 8.406320571899414]
743768d0-9e68-4a7d-8e7f-5f1db0e2dde8
knowledge-enriched-visual-storytelling
1912.01496
null
https://arxiv.org/abs/1912.01496v1
https://arxiv.org/pdf/1912.01496v1.pdf
Knowledge-Enriched Visual Storytelling
Stories are diverse and highly personalized, resulting in a large possible output space for story generation. Existing end-to-end approaches produce monotonous stories because they are limited to the vocabulary and knowledge in a single training dataset. This paper introduces KG-Story, a three-stage framework that allows the story generation model to take advantage of external Knowledge Graphs to produce interesting stories. KG-Story distills a set of representative words from the input prompts, enriches the word set by using external knowledge graphs, and finally generates stories based on the enriched word set. This distill-enrich-generate framework allows the use of external resources not only for the enrichment phase, but also for the distillation and generation phases. In this paper, we show the superiority of KG-Story for visual storytelling, where the input prompt is a sequence of five photos and the output is a short story. Per the human ranking evaluation, stories generated by KG-Story are on average ranked better than that of the state-of-the-art systems. Our code and output stories are available at https://github.com/zychen423/KE-VIST.
['Lun-Wei Ku', "Ting-Hao 'Kenneth' Huang", 'Tzu-Yuan Lin', 'Chih-Chia Li', 'Chi-Yang Hsu', 'Zi-Yuan Chen', 'Chao-Chun Hsu']
2019-12-03
null
null
null
null
['visual-storytelling']
['natural-language-processing']
[ 2.51745284e-02 4.28340107e-01 -1.88209172e-02 -4.31502573e-02 -8.36345255e-01 -7.93325961e-01 8.39462578e-01 1.47010073e-01 -3.94582041e-02 8.23098063e-01 8.64153981e-01 1.36937663e-01 4.83266339e-02 -1.00550270e+00 -7.36594200e-01 -2.85241246e-01 1.30653203e-01 7.80180752e-01 2.87581980e-01 -3.86568695e-01 1.00958414e-01 -2.19710737e-01 -1.77791536e+00 8.11585844e-01 6.93687081e-01 4.84761626e-01 5.83812475e-01 8.79940808e-01 -4.30315077e-01 9.21844363e-01 -6.32771730e-01 -6.11846566e-01 -5.54060563e-02 -7.03035235e-01 -8.42719495e-01 5.05179092e-02 1.82814717e-01 -4.91406798e-01 -4.28304702e-01 4.09779191e-01 8.07265520e-01 3.66100788e-01 5.81437349e-01 -1.17560279e+00 -9.10753191e-01 1.30697429e+00 -1.79631919e-01 -3.32589298e-01 7.19335496e-01 1.85918123e-01 1.34045184e+00 -1.26099634e+00 1.36668205e+00 1.10500145e+00 1.75887018e-01 7.00122237e-01 -9.82134402e-01 -4.56428349e-01 5.27595356e-02 3.84047866e-01 -1.42574692e+00 -2.14730635e-01 6.74062192e-01 -5.61116636e-01 1.01076424e+00 4.56630647e-01 1.15891588e+00 1.17945027e+00 -4.99291182e-01 1.14684558e+00 6.41396701e-01 -4.93953139e-01 1.34774774e-01 3.97840321e-01 -4.54940908e-02 4.80949551e-01 2.51646135e-02 3.77681898e-03 -1.06359923e+00 7.84434527e-02 8.24438334e-01 -2.42099345e-01 -4.23909038e-01 -3.62122089e-01 -1.40102017e+00 6.79797173e-01 4.02714133e-01 2.23365158e-01 -4.94445592e-01 1.92254454e-01 3.67503315e-01 3.36765870e-02 4.26353365e-01 9.14892912e-01 -2.97231320e-02 -3.49296421e-01 -9.89337027e-01 7.09587574e-01 7.48380899e-01 1.29942131e+00 4.53774363e-01 -2.13460550e-01 -9.50173914e-01 8.07328999e-01 6.28675967e-02 3.31863582e-01 4.17012632e-01 -6.33693993e-01 4.35647786e-01 6.57737792e-01 2.22187251e-01 -6.81885242e-01 -8.66692737e-02 -3.74871939e-01 -3.83150727e-01 9.30895209e-02 2.14809507e-01 -2.66610682e-01 -9.36229050e-01 1.64360785e+00 4.03210491e-01 9.79603305e-02 1.38535574e-01 7.63100147e-01 1.55524015e+00 9.54871416e-01 -5.39904200e-02 1.63575802e-02 1.28476143e+00 -1.31998146e+00 -6.61255240e-01 -6.68784678e-02 4.34432596e-01 -7.62686193e-01 1.60012114e+00 2.63713926e-01 -1.14033568e+00 -4.92453933e-01 -8.79097700e-01 -1.12955384e-01 -5.50132632e-01 2.25763470e-01 3.37569624e-01 3.78732085e-02 -1.10467207e+00 5.31686008e-01 -1.26399726e-01 -5.81479549e-01 4.02585477e-01 -2.55159348e-01 -1.81895360e-01 -2.99526960e-01 -1.23809540e+00 7.36254275e-01 8.17143202e-01 -5.21442473e-01 -1.03164577e+00 -9.05103087e-01 -7.67122269e-01 -1.27881184e-01 6.03285313e-01 -1.03555262e+00 1.46670437e+00 -6.68782353e-01 -1.33268821e+00 6.15393043e-01 4.93711717e-02 -1.22255795e-01 6.77161515e-01 -4.87930417e-01 -1.40977427e-01 1.49439573e-01 2.18342572e-01 1.09002912e+00 5.94013572e-01 -1.45321381e+00 -4.55832064e-01 2.76037633e-01 3.31345499e-01 6.30563319e-01 -2.90968031e-01 -2.01848283e-01 -8.91405165e-01 -8.27915311e-01 -6.06215060e-01 -6.66855037e-01 -1.97424605e-01 -4.48045582e-01 -6.65230155e-01 -1.02423713e-01 6.29330099e-01 -6.77201092e-01 1.47467542e+00 -2.09506297e+00 2.61028081e-01 2.04241499e-02 1.26665473e-01 -6.68707713e-02 -3.42693955e-01 9.55876112e-01 2.41998173e-02 2.03792617e-01 2.10310087e-01 -4.26353484e-01 7.90585577e-02 -1.74264669e-01 -3.91338676e-01 -4.56409872e-01 3.97913978e-02 1.19996834e+00 -1.33206260e+00 -5.64445138e-01 2.58999914e-01 4.11555767e-01 -4.50832635e-01 3.71258676e-01 -6.83867931e-01 2.83023089e-01 -3.21336538e-01 3.30028743e-01 -6.10968396e-02 -3.97191226e-01 -1.27186805e-01 -1.61942299e-02 7.47528821e-02 2.87457675e-01 -1.02291489e+00 1.92061198e+00 -3.29325140e-01 8.42276394e-01 -9.86159146e-01 1.77345082e-01 8.54993582e-01 4.82778400e-01 1.09116383e-01 -3.79520625e-01 6.24679402e-02 -1.19908173e-02 -5.64352512e-01 -6.30801380e-01 1.08750784e+00 -1.12544611e-01 -2.72980213e-01 5.71991086e-01 3.15378666e-01 -4.87571031e-01 7.90811896e-01 8.17412972e-01 1.03819287e+00 5.18156111e-01 3.37838531e-01 2.72867531e-01 -1.60045266e-01 4.16167408e-01 -6.91399351e-02 7.31351137e-01 6.34046078e-01 1.00708210e+00 5.33009470e-01 -2.06714258e-01 -1.29266202e+00 -1.11908388e+00 4.36322421e-01 1.08354676e+00 1.49012715e-01 -1.02496862e+00 -6.43846989e-01 -5.56071401e-01 -1.43542320e-01 1.27193666e+00 -7.06451774e-01 5.34176938e-02 -1.17620394e-01 -1.67647943e-01 3.50874126e-01 5.43149948e-01 1.37909114e-01 -1.48419952e+00 -8.55233669e-01 2.29629070e-01 -5.62011957e-01 -9.11411405e-01 -4.97637242e-01 -2.71705747e-01 -4.07533258e-01 -8.35701346e-01 -9.30580258e-01 -4.92760301e-01 6.43104792e-01 2.59885192e-01 1.49158370e+00 -4.80926558e-02 -2.06099465e-01 4.23691541e-01 -9.54126000e-01 -6.03973985e-01 -3.69602114e-01 -5.33826509e-03 -4.42967951e-01 -1.71694607e-01 -9.79035273e-02 -4.78358328e-01 -4.72979575e-01 -8.55078474e-02 -9.40845311e-01 1.17552018e+00 4.63902593e-01 6.00961149e-01 7.76354373e-01 -1.00654699e-01 4.80784327e-01 -9.28824246e-01 8.20155740e-01 -6.31904542e-01 -1.43771261e-01 3.08383703e-01 -2.17657536e-01 -5.26128672e-02 3.37882578e-01 -5.78734577e-01 -1.16419160e+00 -1.79376546e-02 2.44711369e-01 -4.67514277e-01 3.12015153e-02 8.00830543e-01 -4.34022993e-02 7.53854454e-01 9.76666152e-01 1.87453881e-01 -4.99924868e-01 -3.05602372e-01 1.11531031e+00 4.01895672e-01 6.22981429e-01 -3.89861733e-01 7.58636832e-01 8.91266093e-02 -5.16578257e-01 -6.20386899e-01 -6.74321353e-01 -4.00045067e-01 -2.14388937e-01 -9.65598762e-01 6.48179293e-01 -1.04807806e+00 -2.75936164e-03 3.03447127e-01 -1.20583916e+00 -7.42098987e-01 -1.01318920e+00 2.23107487e-01 -7.11252809e-01 -3.01766604e-01 -3.34985375e-01 -5.94625950e-01 -4.56392318e-01 -6.47899747e-01 9.97581184e-01 4.83285487e-01 -7.31010020e-01 -5.87338448e-01 1.21036135e-01 2.54924804e-01 2.10163072e-01 5.84832788e-01 7.92833209e-01 -6.28038466e-01 -7.11886644e-01 -3.04245293e-01 -7.56251514e-02 -3.35654646e-01 -8.11688751e-02 1.42283306e-01 -8.98727179e-01 1.21533178e-01 -8.63180339e-01 -4.95193154e-01 8.19873750e-01 2.65966713e-01 7.66532779e-01 -4.39547122e-01 -3.10021877e-01 1.78126007e-01 1.41728854e+00 6.67724684e-02 7.08856106e-01 3.77091408e-01 9.03426528e-01 7.03270733e-01 7.41208792e-01 8.08281720e-01 6.72510326e-01 6.96037114e-01 3.24876577e-01 -6.64206222e-02 -5.06093144e-01 -9.11179185e-01 4.08243805e-01 6.19889200e-01 -1.10786669e-01 -5.93469262e-01 -8.86655450e-01 1.12519658e+00 -2.07396555e+00 -1.23769379e+00 -1.56096816e-01 1.99878311e+00 1.02988386e+00 1.98650174e-03 3.79991382e-01 -5.55578209e-02 5.90817392e-01 2.13136598e-01 -3.30689818e-01 -1.70836776e-01 -1.80850238e-01 2.76479665e-02 3.99752185e-02 4.58336622e-01 -6.40253603e-01 1.23930800e+00 5.69111872e+00 1.10009742e+00 -6.61175609e-01 5.94366118e-02 4.72121030e-01 -7.91031182e-01 -7.84944654e-01 -1.84795167e-02 -6.26869678e-01 2.41914794e-01 4.98202354e-01 -7.60693789e-01 3.31326067e-01 9.27351058e-01 1.16129659e-01 -2.43259639e-01 -1.10008049e+00 8.51013839e-01 1.49223581e-01 -1.70832992e+00 5.22715628e-01 -4.07615542e-01 9.82209563e-01 -2.37600863e-01 -8.06100741e-02 4.58727837e-01 6.71058297e-01 -9.56514120e-01 1.20060778e+00 6.56589866e-01 1.04580140e+00 -8.74294043e-01 5.14144003e-01 2.45489702e-01 -1.23261774e+00 2.30529413e-01 -1.29282087e-01 -3.12721692e-02 5.34271002e-01 5.91270387e-01 -1.26570606e+00 6.61405504e-01 7.13745236e-01 6.17396355e-01 -5.67237973e-01 1.12338412e+00 -8.02336991e-01 3.87544245e-01 5.99691784e-03 -3.41697186e-01 1.35602161e-01 2.98149616e-01 8.66810679e-01 1.44123077e+00 6.08604670e-01 2.51113802e-01 6.40766546e-02 9.33469534e-01 -2.58006424e-01 4.77030963e-01 -8.11742365e-01 -3.53336513e-01 6.21795475e-01 1.28548336e+00 -7.19265342e-01 -6.95713162e-01 3.17645967e-02 1.01062369e+00 2.84885854e-01 3.89062136e-01 -6.75555348e-01 -5.47153056e-01 2.37413079e-01 2.71713883e-01 4.03983533e-01 1.36503041e-01 -2.42626712e-01 -9.26168382e-01 -1.82719175e-02 -7.26166546e-01 2.70985007e-01 -1.53171241e+00 -8.78878772e-01 7.71184146e-01 3.79637599e-01 -1.20770562e+00 -6.23419404e-01 6.20750263e-02 -7.70110905e-01 6.78146958e-01 -1.02383626e+00 -1.38802910e+00 -7.02066123e-01 4.72117096e-01 9.47029233e-01 -3.90557200e-02 7.54235387e-01 -1.31254420e-01 -8.87365416e-02 3.41917068e-01 -1.45041794e-01 -1.01754241e-01 6.78259254e-01 -1.40113091e+00 7.52950788e-01 9.15730596e-01 4.21723306e-01 8.33788067e-02 9.86844659e-01 -1.02468514e+00 -8.20825279e-01 -1.20214164e+00 1.03130472e+00 -5.81102073e-01 5.48459768e-01 -3.76888931e-01 -6.31042838e-01 4.62437987e-01 6.13956332e-01 -5.00531971e-01 8.77703130e-01 1.87233668e-02 -3.43682557e-01 3.57022732e-01 -7.40109265e-01 1.12143958e+00 1.18871820e+00 -2.24331692e-01 -5.32120645e-01 2.51294672e-01 1.11118042e+00 -6.42080426e-01 -5.09473741e-01 -8.09319541e-02 6.43406034e-01 -8.27623308e-01 7.59322941e-01 -4.87733275e-01 1.21546483e+00 -4.05252606e-01 7.67618716e-02 -1.63793647e+00 -4.35612142e-01 -7.37943113e-01 -3.68043721e-01 1.53995740e+00 8.71481180e-01 1.58506304e-01 4.39662874e-01 2.67878741e-01 -1.93638310e-01 -7.52297103e-01 -3.22386175e-01 -6.47037804e-01 -5.01157701e-01 -5.71894467e-01 8.93501997e-01 6.80682302e-01 4.40122604e-01 5.83889365e-01 -6.36054575e-01 -3.17075938e-01 3.67776692e-01 1.60151511e-01 1.09488082e+00 -8.13930392e-01 -6.08246624e-01 -4.26042706e-01 2.79803369e-02 -7.07092166e-01 -4.91984397e-01 -9.55826640e-01 2.84568965e-01 -2.32595849e+00 5.48379719e-01 -1.06357351e-01 -1.08155735e-01 5.62859237e-01 -4.90703166e-01 2.70084739e-01 6.85942292e-01 7.56561980e-02 -7.65200913e-01 4.72614348e-01 1.47666585e+00 -6.27636313e-02 -5.91755509e-01 -4.32537884e-01 -9.23688710e-01 4.31278855e-01 8.21549237e-01 -3.70206207e-01 -8.86009753e-01 -3.89961332e-01 6.82928324e-01 2.54721027e-02 3.18139374e-01 -9.93641615e-01 -5.11433817e-02 -1.74234793e-01 4.47814971e-01 -7.09992826e-01 4.40515429e-01 -3.10785651e-01 7.23500669e-01 1.44142574e-02 -5.17575324e-01 -7.82580599e-02 3.79258215e-01 3.68623942e-01 -6.53960556e-02 -1.73137903e-01 2.48793453e-01 -2.64254928e-01 -8.79522264e-01 1.78398862e-01 -2.59953141e-01 1.15696952e-01 1.13572204e+00 -3.96793693e-01 -4.87829745e-01 -7.80062556e-01 -7.24477172e-01 2.80540258e-01 5.07899046e-01 7.17944086e-01 7.91737020e-01 -1.86373055e+00 -1.07077968e+00 -3.38140965e-01 5.80507040e-01 1.23861611e-01 4.80012476e-01 3.17272633e-01 -3.63687813e-01 1.74594428e-02 -2.49612197e-01 -1.88256994e-01 -1.50456381e+00 5.37105083e-01 -3.14910382e-01 -5.97829044e-01 -8.25796783e-01 1.19634104e+00 2.74312496e-01 1.38872683e-01 8.51609558e-02 6.55358797e-03 -4.57337141e-01 3.76776427e-01 7.86026716e-01 2.59352446e-01 -4.01254088e-01 -4.44280922e-01 4.67690527e-02 1.99685127e-01 -7.46955872e-02 -6.51715934e-01 1.38217628e+00 6.70772269e-02 3.42667460e-01 5.67903876e-01 5.69961190e-01 4.19155061e-02 -1.11844313e+00 -2.54669249e-01 -1.30863354e-01 -4.91254538e-01 -2.87997603e-01 -1.16385174e+00 -7.36090958e-01 5.09196222e-01 1.08765550e-02 1.92886978e-01 1.19814515e+00 4.55887914e-01 6.58743322e-01 5.65266572e-02 3.65883797e-01 -9.88036811e-01 5.43696404e-01 4.15511250e-01 1.55354404e+00 -7.13427126e-01 -1.34271666e-01 -3.00876200e-01 -1.23177242e+00 8.63999486e-01 6.49357021e-01 1.32792100e-01 -5.78249292e-03 5.63620217e-02 -5.03076948e-02 -2.73411393e-01 -1.05388570e+00 -4.13830400e-01 5.30948341e-01 6.88110590e-01 3.95620197e-01 3.57273310e-01 -1.76594079e-01 8.92416537e-01 -6.23559773e-01 2.54594982e-01 6.35600567e-01 7.37193406e-01 -5.18108964e-01 -1.06917763e+00 -1.82928070e-01 5.09642184e-01 1.25836981e-02 -2.61151820e-01 -8.03126454e-01 7.44363129e-01 1.57016560e-01 8.14682186e-01 -2.37738073e-01 -6.30315840e-01 5.62358201e-01 1.05704226e-01 3.98484230e-01 -1.01574647e+00 -4.88231570e-01 -1.27828987e-02 5.18073022e-01 -4.80032653e-01 -7.92108402e-02 -5.86033821e-01 -1.27581108e+00 -3.00742894e-01 -2.45324954e-01 6.03371449e-02 4.25987095e-01 6.90728128e-01 4.18961763e-01 8.30184162e-01 2.84808040e-01 -9.65076089e-01 3.31816614e-01 -1.12496507e+00 -3.84388775e-01 5.00520468e-01 -1.27326399e-01 -5.24546623e-01 1.47344604e-01 4.46782023e-01]
[11.211369514465332, 0.784960150718689]
2ebadc83-f1fb-4298-a495-3ad93c11394f
deep-multi-metric-learning-for-text
2007.10479
null
https://arxiv.org/abs/2007.10479v1
https://arxiv.org/pdf/2007.10479v1.pdf
Deep multi-metric learning for text-independent speaker verification
Text-independent speaker verification is an important artificial intelligence problem that has a wide spectrum of applications, such as criminal investigation, payment certification, and interest-based customer services. The purpose of text-independent speaker verification is to determine whether two given uncontrolled utterances originate from the same speaker or not. Extracting speech features for each speaker using deep neural networks is a promising direction to explore and a straightforward solution is to train the discriminative feature extraction network by using a metric learning loss function. However, a single loss function often has certain limitations. Thus, we use deep multi-metric learning to address the problem and introduce three different losses for this problem, i.e., triplet loss, n-pair loss and angular loss. The three loss functions work in a cooperative way to train a feature extraction network equipped with Residual connections and squeeze-and-excitation attention. We conduct experiments on the large-scale \texttt{VoxCeleb2} dataset, which contains over a million utterances from over $6,000$ speakers, and the proposed deep neural network obtains an equal error rate of $3.48\%$, which is a very competitive result. Codes for both training and testing and pretrained models are available at \url{https://github.com/GreatJiweix/DmmlTiSV}, which is the first publicly available code repository for large-scale text-independent speaker verification with performance on par with the state-of-the-art systems.
['Wenyu Liu', 'Xinggang Wang', 'Bin Feng', 'Jiwei Xu']
2020-07-17
null
null
null
null
['text-independent-speaker-verification']
['speech']
[-7.60591552e-02 -2.22632840e-01 7.17176497e-02 -9.18032825e-01 -1.22018600e+00 -3.82216930e-01 1.87103078e-01 -3.37787002e-01 -4.76814687e-01 4.57491636e-01 4.83397357e-02 -5.43936074e-01 1.80997252e-01 -2.18014836e-01 -5.76779008e-01 -9.24446523e-01 3.00375652e-02 3.83452594e-01 -3.40272695e-01 -2.77382672e-01 -3.90456466e-04 3.63469630e-01 -1.21431136e+00 -1.08631611e-01 7.94357777e-01 1.29744887e+00 -2.21876740e-01 4.59365845e-01 2.19956674e-02 4.65945810e-01 -7.19079256e-01 -7.76674032e-01 1.99301243e-01 -3.41648072e-01 -5.77981055e-01 -2.02280179e-01 1.58043206e-01 -4.04875636e-01 -5.08751273e-01 1.06588566e+00 1.04825985e+00 9.49588567e-02 3.57299060e-01 -1.43771112e+00 -6.03170693e-01 8.93446326e-01 -6.40239239e-01 3.10822755e-01 1.32356942e-01 1.52648598e-01 1.06943107e+00 -1.00432146e+00 -9.54199210e-02 1.13048244e+00 5.98976433e-01 8.37182879e-01 -8.64703178e-01 -1.19726002e+00 7.86354244e-02 4.59376812e-01 -1.81406796e+00 -1.10177517e+00 1.01606309e+00 -1.80723682e-01 7.07091153e-01 2.33991921e-01 -2.51970235e-02 1.27798939e+00 -2.63058901e-01 1.08890188e+00 7.10847259e-01 -2.69824088e-01 -4.12101597e-02 4.56990659e-01 2.43282512e-01 7.04563737e-01 -1.49220452e-01 1.61536470e-01 -7.14834511e-01 -1.10583544e-01 1.90843180e-01 -1.75947741e-01 -5.60503364e-01 1.09718777e-01 -9.93105352e-01 1.02774954e+00 2.95333624e-01 3.30318719e-01 7.82144740e-02 4.89397198e-02 5.50751626e-01 3.64644140e-01 4.33002621e-01 -1.82544947e-01 -4.39644128e-01 -3.34814906e-01 -7.95047641e-01 6.41908497e-02 6.54001594e-01 9.66182768e-01 4.73906517e-01 4.48117852e-01 1.09380186e-01 1.18116033e+00 6.84251130e-01 7.77914286e-01 6.96114600e-01 -5.17803848e-01 7.47569382e-01 1.74252734e-01 -2.89619416e-01 -5.20529270e-01 -1.93693250e-01 -4.54881012e-01 -9.69191372e-01 5.97336665e-02 3.69543940e-01 -4.51643348e-01 -7.58479774e-01 1.96126103e+00 2.79117197e-01 1.12808794e-01 6.10706545e-02 9.66592729e-01 9.79016006e-01 5.43459356e-01 -1.70998931e-01 6.09704182e-02 1.32160687e+00 -9.30097401e-01 -6.77402437e-01 -2.00253531e-01 5.34827948e-01 -8.14720154e-01 8.91869485e-01 2.86024302e-01 -9.67755914e-01 -5.00841081e-01 -1.18859947e+00 -1.34530932e-01 -2.40258694e-01 2.34994873e-01 2.39481866e-01 1.06227243e+00 -1.07436252e+00 1.67696431e-01 -6.57793164e-01 -1.95719451e-02 4.47696000e-01 4.27620858e-01 -4.02797222e-01 -5.26831485e-02 -1.34676576e+00 7.13946402e-01 -1.73657268e-01 4.15821493e-01 -9.34291720e-01 -4.35835123e-01 -1.14573336e+00 1.29564613e-01 4.47543748e-02 -1.45713359e-01 1.24299836e+00 -8.14048827e-01 -1.74490631e+00 9.17058051e-01 -5.00142097e-01 -3.93441826e-01 4.54491794e-01 4.25503626e-02 -7.74004519e-01 -2.41922915e-01 -3.69833037e-02 3.62500340e-01 1.05414450e+00 -8.12817276e-01 -4.01208609e-01 -5.93315601e-01 -3.11613709e-01 -4.97634932e-02 -5.68092108e-01 6.34680092e-01 -2.28517294e-01 -5.53555250e-01 4.23353575e-02 -7.11605132e-01 2.69036531e-01 -1.35612875e-01 -7.66541541e-01 -4.45654333e-01 9.30425167e-01 -1.04640198e+00 1.06501377e+00 -2.54675221e+00 -4.46505025e-02 1.22203678e-01 7.60928839e-02 2.60327727e-01 -2.30472624e-01 6.28538355e-02 -2.03448936e-01 2.10556105e-01 -4.49166894e-01 -9.88969386e-01 4.14338380e-01 -4.06906217e-01 -1.65731475e-01 7.36372054e-01 8.31605494e-02 7.45139420e-01 -2.26098344e-01 -3.16108525e-01 -6.30587712e-03 8.13545763e-01 -1.64509222e-01 2.28799626e-01 3.51023644e-01 3.54024321e-01 -2.39763841e-01 8.62666249e-01 8.85023355e-01 2.69908719e-02 -1.38398066e-01 1.44813910e-01 1.06972069e-01 5.83710134e-01 -1.11302245e+00 1.76778173e+00 -3.80800158e-01 7.77327359e-01 5.33112347e-01 -1.24974978e+00 1.06613731e+00 6.54520571e-01 2.80230224e-01 -5.99759877e-01 5.68058789e-01 4.18795645e-01 1.70989111e-01 -2.92639971e-01 2.28145331e-01 -3.87154937e-01 -1.59056857e-01 5.87730050e-01 1.47037089e-01 2.66651601e-01 -7.56641701e-02 -1.02986908e-03 8.55436146e-01 -3.70137721e-01 -1.97200000e-01 -2.14908831e-02 9.09313738e-01 -7.38702059e-01 8.69221807e-01 2.36812249e-01 -7.17660904e-01 6.64895356e-01 3.05963188e-01 -1.06191024e-01 -5.99694669e-01 -8.44015181e-01 -3.57125401e-01 1.09697616e+00 -5.42921051e-02 1.11929372e-01 -6.79487824e-01 -7.18374431e-01 8.22177157e-02 6.32877350e-01 -4.00525391e-01 -2.09444776e-01 -6.87741697e-01 -6.43562794e-01 9.41555083e-01 4.74956751e-01 6.79152429e-01 -1.02133965e+00 6.95201755e-02 7.46029913e-02 -4.84259784e-01 -1.05158150e+00 -9.98314977e-01 2.12448090e-01 -3.87203634e-01 -7.21943438e-01 -8.80416036e-01 -9.16769624e-01 4.01542395e-01 1.75266787e-01 7.97805309e-01 8.38941932e-02 -1.53313875e-01 2.50438582e-02 -2.57491052e-01 -5.14950037e-01 -3.45993459e-01 2.02697828e-01 3.92625868e-01 4.09800470e-01 7.52657473e-01 -3.75861943e-01 -5.08403540e-01 5.59295118e-01 -5.31035066e-01 -6.46080017e-01 2.44245544e-01 9.93518174e-01 2.92265534e-01 -4.73255441e-02 8.20769370e-01 -2.13094100e-01 4.81912434e-01 -3.16375643e-01 -6.00119233e-01 2.06713244e-01 -1.77856237e-01 -7.62543976e-02 4.97650355e-01 -2.44973511e-01 -9.40871119e-01 3.39961536e-02 -7.14912236e-01 -4.50109303e-01 -2.44610623e-01 3.37200791e-01 -7.18382061e-01 -7.49212801e-02 3.64282608e-01 5.57136297e-01 1.18883163e-01 -5.67906618e-01 4.64287475e-02 1.18666887e+00 2.49853075e-01 -2.45524645e-01 8.91352594e-01 1.90050438e-01 -6.80231988e-01 -9.29170728e-01 -4.67271149e-01 -6.39381230e-01 -2.08638713e-01 4.95035797e-02 6.55847967e-01 -1.01717925e+00 -1.08515215e+00 8.72267902e-01 -1.20318580e+00 -1.33634210e-01 3.17020379e-02 6.33045614e-01 -2.11791918e-01 3.23397338e-01 -5.14725626e-01 -9.45022762e-01 -6.91512585e-01 -1.48386979e+00 1.01785123e+00 1.66194826e-01 1.38300523e-01 -7.19712436e-01 -1.21846035e-01 7.41360843e-01 6.87726140e-01 -2.71604598e-01 5.65259278e-01 -1.02591133e+00 -3.67007077e-01 -3.22156072e-01 -1.00442111e-01 7.03667164e-01 1.73198149e-01 -1.81310251e-01 -1.44493759e+00 -6.08705103e-01 3.93775702e-01 -4.44353670e-01 7.80270159e-01 4.41963762e-01 1.20731699e+00 -2.04143479e-01 -1.99781299e-01 7.88705766e-01 1.03485072e+00 3.56785476e-01 4.48636711e-01 -4.21550125e-02 5.83424926e-01 5.48673749e-01 8.06018189e-02 4.51182753e-01 5.02508879e-01 7.56241620e-01 2.08334476e-01 -2.25891452e-03 1.24106981e-01 6.97755739e-02 5.03503859e-01 8.61158013e-01 3.12761933e-01 -4.38182324e-01 -8.58105421e-01 5.03873110e-01 -1.43450761e+00 -1.13887179e+00 2.73371875e-01 2.25079107e+00 6.57666206e-01 -1.04608256e-02 2.27941230e-01 5.35810828e-01 9.50452268e-01 2.33946472e-01 -7.76912749e-01 -4.26146448e-01 -1.34066284e-01 1.47588968e-01 1.23009585e-01 5.82660556e-01 -1.23635566e+00 6.19769990e-01 5.02140188e+00 8.35077345e-01 -1.53216171e+00 3.70154589e-01 9.40000653e-01 -2.44492233e-01 -8.86195973e-02 -4.22658801e-01 -1.12781954e+00 6.40968978e-01 1.14616203e+00 -1.31936744e-01 4.14304137e-01 7.89950430e-01 3.10728759e-01 5.07945538e-01 -9.67387080e-01 1.27946663e+00 3.91057014e-01 -8.97884369e-01 -4.58508521e-01 1.70473993e-01 2.45883986e-01 3.32313776e-01 2.88271666e-01 5.38076103e-01 1.63540497e-01 -1.06444168e+00 9.78961587e-01 3.75726856e-02 9.61445928e-01 -8.08777273e-01 9.05585647e-01 2.28395090e-01 -1.34194815e+00 -1.61567941e-01 -1.02197222e-01 3.79784554e-01 4.10609156e-01 5.15037954e-01 -4.83621389e-01 3.19987148e-01 8.76855135e-01 4.92723316e-01 -6.14185147e-02 9.33894277e-01 -2.36471847e-01 8.06196928e-01 -3.24874133e-01 -4.11817767e-02 -1.23594470e-01 3.00367884e-02 5.84193408e-01 1.06099093e+00 4.07201648e-01 -5.45637347e-02 -2.92727649e-01 9.03792083e-01 -5.79264820e-01 9.56603736e-02 -4.17734534e-01 5.83784729e-02 4.99073505e-01 1.09757364e+00 -5.96739091e-02 -9.98494253e-02 -4.29457664e-01 9.85888600e-01 3.78406674e-01 3.76453161e-01 -1.16454077e+00 -8.92156541e-01 9.74536240e-01 -2.98505366e-01 5.46239793e-01 -1.73454896e-01 -3.15246344e-01 -1.20729113e+00 5.15516222e-01 -1.01157653e+00 2.04615280e-01 -2.24192888e-01 -1.35062861e+00 9.14318204e-01 -5.44196010e-01 -1.20384133e+00 -1.95932075e-01 -4.67149198e-01 -8.29260290e-01 1.26520228e+00 -1.76804650e+00 -1.03513384e+00 -2.11771086e-01 8.34901929e-01 4.43915129e-01 -4.78217095e-01 7.29599655e-01 7.11459219e-01 -9.44482386e-01 1.48287487e+00 2.72779584e-01 6.97605550e-01 7.17916548e-01 -8.80621672e-01 6.00456715e-01 8.66628826e-01 1.98563591e-01 5.33438802e-01 3.22178632e-01 4.29127477e-02 -1.39315927e+00 -8.59755099e-01 1.15762627e+00 -2.26285473e-01 4.57269311e-01 -8.10553253e-01 -9.24447715e-01 6.13105118e-01 2.28488326e-01 1.82969615e-01 8.94921243e-01 2.08562464e-02 -5.47469020e-01 -4.37702149e-01 -1.33520091e+00 3.51372734e-02 9.02691245e-01 -7.51746297e-01 -2.55481571e-01 2.36527741e-01 7.11920738e-01 -3.37038606e-01 -5.86846828e-01 2.44091481e-01 5.11619449e-01 -1.00557315e+00 8.02953124e-01 -2.39903361e-01 -2.85793282e-02 -1.15544222e-01 -3.10264915e-01 -1.14288211e+00 -3.84055101e-03 -7.65019119e-01 7.47413933e-02 1.64768314e+00 8.42488468e-01 -1.08010507e+00 7.19463646e-01 7.99974501e-01 -2.94505805e-01 -7.03283012e-01 -1.50313067e+00 -8.75919938e-01 2.21486256e-01 -6.37896061e-01 1.06355774e+00 8.54370236e-01 3.94105799e-02 2.11806849e-01 -3.37498695e-01 2.12337181e-01 7.05170512e-01 -7.68102631e-02 6.12158716e-01 -9.61304009e-01 -2.03920990e-01 -6.71987176e-01 -3.51694733e-01 -1.26937163e+00 4.69312519e-01 -8.88781369e-01 2.45376483e-01 -1.04652953e+00 -6.41004890e-02 -6.58945382e-01 -5.09469450e-01 5.55900872e-01 -1.06513865e-01 7.11155459e-02 5.85462861e-02 -1.05032831e-01 -1.92045793e-01 1.00250447e+00 6.46289110e-01 -6.36591971e-01 -1.35943368e-01 3.49644035e-01 -8.79919052e-01 3.55918974e-01 9.39286828e-01 -3.91066611e-01 -7.51054138e-02 -7.06627369e-01 -4.79693264e-01 2.21672729e-02 2.02908829e-01 -7.50455320e-01 2.79698253e-01 2.95377582e-01 1.26000065e-02 -4.42300260e-01 5.68838596e-01 -6.43708408e-01 -3.24326217e-01 2.25841835e-01 -3.02514732e-01 -8.15106481e-02 2.58790582e-01 2.19074413e-01 -6.26503825e-01 -1.25112921e-01 9.00241971e-01 2.49449372e-01 -2.96198338e-01 6.05023265e-01 -1.03864476e-01 2.64460314e-02 6.36955261e-01 -2.35235412e-03 -2.91603893e-01 -4.99326646e-01 -3.90732527e-01 2.24465653e-01 -1.26897603e-01 6.86481535e-01 7.65143991e-01 -1.33169687e+00 -1.13737738e+00 5.60860813e-01 8.63060802e-02 -1.99083865e-01 4.05094236e-01 8.66956532e-01 6.66419292e-05 4.12128061e-01 3.11405092e-01 -6.00541592e-01 -1.37900257e+00 1.71166927e-01 6.21362984e-01 1.01721659e-01 -3.17362964e-01 1.40570426e+00 2.35404652e-02 -7.48422265e-01 6.27281964e-01 -1.50457844e-01 7.62282982e-02 -1.61481053e-01 7.33385384e-01 2.25763485e-01 3.24109435e-01 -1.07158887e+00 -7.52074480e-01 4.58370894e-01 -2.69473046e-01 -1.95650890e-01 1.08834672e+00 -1.31198287e-01 1.67594433e-01 8.73322263e-02 1.68904960e+00 -7.62820318e-02 -9.28054750e-01 -5.03682852e-01 -3.20017785e-01 -3.03173065e-01 8.93402472e-02 -5.95933676e-01 -1.55258012e+00 1.19345212e+00 7.68488646e-01 1.05402358e-01 9.49879289e-01 -2.42207367e-02 1.05205536e+00 1.88047573e-01 1.89897910e-01 -8.61970127e-01 -2.23211408e-01 4.66023356e-01 9.92453218e-01 -1.74774349e+00 -3.78781587e-01 5.50319701e-02 -5.58635831e-01 6.63051367e-01 5.34240246e-01 4.94569391e-01 9.47097540e-01 1.99646503e-01 4.52627659e-01 5.68858422e-02 -3.57220113e-01 1.89460609e-02 1.84767872e-01 3.79134387e-01 6.54982448e-01 2.20391855e-01 3.54129747e-02 7.53255010e-01 -4.03296769e-01 -4.49161738e-01 -2.97694132e-02 7.53178835e-01 -1.01520725e-01 -1.13820231e+00 -4.50182229e-01 3.01843882e-01 -4.92221743e-01 -2.05605835e-01 -2.25321382e-01 3.48158568e-01 -1.47766382e-01 1.40809834e+00 -1.72688812e-01 -6.14278436e-01 2.70175487e-01 2.42467642e-01 2.08113436e-02 -1.74885213e-01 -5.59588552e-01 -1.08013846e-01 -8.60698968e-02 -3.07667524e-01 -1.14729173e-01 -8.02066386e-01 -1.18221045e+00 -5.75445473e-01 -6.42513275e-01 1.91932783e-01 1.11163747e+00 8.56445193e-01 4.11922157e-01 2.17004776e-01 1.14934695e+00 -6.28072739e-01 -8.01404119e-01 -1.22206044e+00 -7.53376901e-01 2.70307481e-01 7.20462978e-01 -5.05167961e-01 -7.40672827e-01 -1.89669311e-01]
[14.266674995422363, 6.049466133117676]
fcc99fa2-900c-4fce-8bbb-b77895014c8e
dynamic-vertex-replacement-grammars
2303.11553
null
https://arxiv.org/abs/2303.11553v2
https://arxiv.org/pdf/2303.11553v2.pdf
Dynamic Vertex Replacement Grammars
Context-free graph grammars have shown a remarkable ability to model structures in real-world relational data. However, graph grammars lack the ability to capture time-changing phenomena since the left-to-right transitions of a production rule do not represent temporal change. In the present work, we describe dynamic vertex-replacement grammars (DyVeRG), which generalize vertex replacement grammars in the time domain by providing a formal framework for updating a learned graph grammar in accordance with modifications to its underlying data. We show that DyVeRG grammars can be learned from, and used to generate, real-world dynamic graphs faithfully while remaining human-interpretable. We also demonstrate their ability to forecast by computing dyvergence scores, a novel graph similarity measurement exposed by this framework.
['Tim Weninger', 'Grant Boquet', 'Timothy La Fond', 'Justus Isaiah Hibshman', 'Daniel Gonzalez Cedre']
2023-03-21
null
null
null
null
['graph-similarity']
['graphs']
[ 2.52290457e-01 6.79096341e-01 -1.48190707e-01 -4.26399380e-01 7.63560832e-02 -7.86444604e-01 1.14697564e+00 3.07197928e-01 3.14072251e-01 7.78375030e-01 -1.34000391e-01 -6.43347144e-01 -3.70014668e-01 -1.29306889e+00 -8.29752445e-01 -2.94967264e-01 -6.78658009e-01 8.97728801e-01 7.14516699e-01 -5.32863140e-01 -1.90172255e-01 8.75272512e-01 -1.56769371e+00 3.48752961e-02 6.09028995e-01 1.63088098e-01 7.98715428e-02 9.26222265e-01 -2.13015839e-01 8.82253528e-01 -3.21879536e-01 -4.64404702e-01 1.55450972e-02 -5.91445863e-01 -6.23852491e-01 3.22727978e-01 3.14813673e-01 1.64538249e-01 -7.21471786e-01 6.37976944e-01 -3.58308941e-01 2.25671113e-01 5.10084450e-01 -1.57432055e+00 -6.13191843e-01 8.40370595e-01 8.22409317e-02 2.01429874e-01 7.10413754e-01 -8.56295601e-02 1.14250243e+00 -2.17527896e-01 1.36113989e+00 1.51451981e+00 4.45941359e-01 5.75104773e-01 -1.54283857e+00 -5.99431247e-02 6.81382537e-01 2.30883937e-02 -1.04070008e+00 -6.82485551e-02 9.22037840e-01 -3.56651932e-01 1.28462136e+00 5.06341219e-01 8.65199029e-01 9.99436677e-01 6.92536116e-01 9.60075781e-02 7.76841998e-01 -5.01525760e-01 6.71759173e-02 -5.06375492e-01 7.94064030e-02 1.11673117e+00 4.11809534e-01 5.14170289e-01 -3.62701088e-01 -2.25753397e-01 7.95788407e-01 -1.13694198e-01 7.46278390e-02 -6.84261978e-01 -1.12858260e+00 5.84380627e-01 2.89605588e-01 5.71740210e-01 -1.38246208e-01 7.00312436e-01 3.25647652e-01 6.16562843e-01 2.97526658e-01 1.37911081e-01 -3.90653193e-01 2.80999895e-02 -2.25147888e-01 4.57676142e-01 1.02811146e+00 1.24967766e+00 8.51170719e-01 3.51101846e-01 -5.57402037e-02 2.14112952e-01 3.17874849e-01 6.25502646e-01 2.33447999e-01 -7.00881243e-01 1.88371137e-01 7.48512328e-01 -2.12794542e-01 -1.17470825e+00 -5.62168062e-01 -2.01889008e-01 -6.44977033e-01 1.65666603e-02 3.07969034e-01 4.66195196e-01 -1.14310050e+00 2.15588784e+00 3.52275282e-01 3.63331616e-01 1.14927635e-01 3.57289538e-02 3.36769164e-01 6.59467220e-01 1.49798974e-01 -7.24505663e-01 8.46008480e-01 -4.19159144e-01 -6.86216950e-01 -5.48721142e-02 7.59389639e-01 -5.07548749e-02 1.01608908e+00 2.29040235e-01 -1.08007896e+00 -4.69866067e-01 -9.55146253e-01 2.23383754e-01 -5.27374923e-01 -8.46313298e-01 8.76688123e-01 5.64387441e-01 -1.39611602e+00 8.29631388e-01 -1.24535048e+00 -4.85602438e-01 -4.13275242e-01 2.69028544e-01 -5.16017675e-01 2.89594889e-01 -1.24274349e+00 7.61331797e-01 7.47302532e-01 -3.37489042e-03 -9.73906755e-01 -2.80495912e-01 -1.13532162e+00 -7.78447166e-02 7.11573124e-01 -6.02277398e-01 1.36112559e+00 -7.91563094e-01 -1.36517072e+00 8.81031454e-01 -7.21050575e-02 -7.57134199e-01 3.68129343e-01 6.32518709e-01 -1.11159635e+00 1.66055579e-02 -2.40525380e-01 6.84369206e-02 8.21523488e-01 -1.29205656e+00 -3.79688174e-01 -3.55803818e-01 2.39230916e-01 -2.81530380e-01 2.67706245e-01 -3.17424059e-01 -4.81943548e-01 -8.29430342e-01 3.79616678e-01 -1.17738664e+00 -2.63364941e-01 -4.15473551e-01 -4.01045561e-01 -2.82358050e-01 8.42757702e-01 -3.72989297e-01 1.63707280e+00 -1.69034243e+00 3.29034209e-01 5.83249509e-01 3.90990674e-01 -9.89915505e-02 -1.91443413e-01 1.14796710e+00 -2.46965811e-01 3.52844149e-01 -4.47567463e-01 9.28470269e-02 2.05908567e-01 9.85933661e-01 -5.17372787e-01 2.22671136e-01 -3.28880213e-02 1.03155494e+00 -1.26842546e+00 -4.02511090e-01 8.17223340e-02 -2.20197067e-02 -3.62579316e-01 2.76945974e-03 -7.67948687e-01 6.66051824e-03 -4.08090174e-01 3.02843362e-01 1.18775964e-01 -2.50554144e-01 1.12321854e+00 2.43526146e-01 2.55292892e-01 1.77011952e-01 -1.09915745e+00 1.45882118e+00 -4.71650809e-01 2.42587373e-01 -3.50415051e-01 -7.15401709e-01 1.06738079e+00 1.85556039e-01 3.22969437e-01 -8.23661387e-01 -3.75411153e-01 1.54401837e-02 2.01503724e-01 -1.78303182e-01 4.95423257e-01 -1.40280485e-01 -4.83793825e-01 7.51829386e-01 -1.27596438e-01 -4.44246411e-01 8.28839183e-01 7.20355213e-01 1.28513753e+00 2.80385375e-01 6.58769488e-01 -2.04782173e-01 5.52274644e-01 -1.52718425e-01 5.44973493e-01 7.42677033e-01 3.41588438e-01 1.43823788e-01 8.97090793e-01 -7.10117221e-01 -1.02398992e+00 -1.34959102e+00 3.51582915e-01 1.06171668e+00 -1.05051249e-01 -8.33725572e-01 -5.95719099e-01 -8.45453382e-01 2.06537902e-01 9.25959706e-01 -7.90071666e-01 -5.07782519e-01 -9.21525896e-01 -3.36909384e-01 4.77217197e-01 3.97612214e-01 -3.33248526e-01 -1.18444312e+00 -3.78899246e-01 7.77979493e-01 3.11249077e-01 -1.05815816e+00 -5.42706072e-01 2.78312284e-02 -1.04287243e+00 -1.42990351e+00 3.20293903e-01 -6.49089575e-01 7.85284877e-01 -1.04289740e-01 1.51938927e+00 4.84313965e-01 -7.68134296e-02 7.95875847e-01 -2.97838777e-01 -1.36534899e-01 -1.31962538e+00 1.10556573e-01 2.10692529e-02 -1.40020683e-01 -2.19544396e-01 -9.40670073e-01 2.73890942e-01 2.34152690e-01 -1.33651590e+00 2.09790826e-01 -6.74572513e-02 6.13261521e-01 8.27192724e-01 -7.12720491e-03 3.87145072e-01 -1.44312036e+00 7.46010780e-01 -1.59054369e-01 -1.04367399e+00 7.33517528e-01 -1.16182554e+00 6.69011414e-01 7.13363230e-01 -4.13725227e-01 -8.33845258e-01 -1.14874490e-01 2.96551704e-01 -1.81229979e-01 1.49118736e-01 7.39507794e-01 -1.72059789e-01 1.41164765e-01 4.69875216e-01 1.93930134e-01 -7.84629732e-02 -1.03851445e-01 6.38222873e-01 -2.64533758e-01 7.60922790e-01 -9.51005042e-01 1.08116853e+00 4.01187539e-01 6.88138723e-01 -3.36710304e-01 -2.32660234e-01 1.94902733e-01 -7.53542781e-01 -2.74960637e-01 3.85401964e-01 -2.76165515e-01 -6.17325842e-01 4.20938998e-01 -9.59274113e-01 -4.72244114e-01 -6.75397694e-01 -1.15448467e-01 -9.21785891e-01 5.48066437e-01 -5.27686715e-01 -6.54043198e-01 1.39113273e-02 -7.48931408e-01 7.71513402e-01 -2.43398979e-01 -1.85787559e-01 -1.46357977e+00 4.96558398e-01 -6.84724033e-01 -4.06535231e-02 8.12397182e-01 1.36666834e+00 -6.49496615e-01 -6.98520601e-01 -7.97530115e-02 4.09733117e-01 -1.49031013e-01 3.69488478e-01 6.14185333e-01 -2.76242644e-01 -3.48089248e-01 -4.17703688e-01 4.10729319e-01 2.61120766e-01 -4.19196635e-02 8.61110985e-01 -3.15444231e-01 -5.84267795e-01 4.30097967e-01 1.38349771e+00 5.99762738e-01 6.74587846e-01 -5.50270379e-02 6.40324771e-01 3.81628811e-01 3.45242649e-01 1.31860301e-01 5.45957685e-01 8.26417804e-01 4.62454319e-01 3.77749681e-01 -3.34025770e-01 -9.85050321e-01 2.77523905e-01 1.02664268e+00 -3.33193034e-01 -5.73350906e-01 -1.23479855e+00 3.88763666e-01 -2.03066254e+00 -1.06049967e+00 -4.38211262e-01 2.05959868e+00 5.98968089e-01 4.35727566e-01 2.70389952e-02 1.80662945e-02 8.22728634e-01 3.67390037e-01 -4.48022842e-01 -6.49704278e-01 -2.57138938e-01 2.95836806e-01 4.04030591e-01 8.25816095e-01 -4.68391716e-01 1.19212651e+00 7.43496084e+00 1.99215367e-01 -7.69223094e-01 -1.20249450e-01 2.39774138e-02 5.21555662e-01 -8.48252237e-01 6.18202746e-01 -1.40456453e-01 1.90933079e-01 1.39516664e+00 -8.92134368e-01 8.89942586e-01 4.41945046e-01 7.09374575e-03 2.83203363e-01 -1.34802103e+00 5.52571535e-01 -9.40170065e-02 -1.41419661e+00 4.69825208e-01 1.13162376e-01 6.42511010e-01 -3.06064814e-01 -3.64619762e-01 3.04885060e-01 9.57404971e-01 -8.07289541e-01 6.87434018e-01 1.02540541e+00 6.63350344e-01 -6.34595335e-01 2.85344701e-02 1.28281951e-01 -1.57173324e+00 9.76988971e-02 -8.34407657e-03 -6.87756836e-02 3.57057869e-01 3.43045682e-01 -1.05492067e+00 1.09687018e+00 3.61568928e-02 6.67396784e-01 -6.88699961e-01 3.17584783e-01 -4.23427075e-01 4.47942883e-01 -1.21029980e-01 1.30798087e-01 -1.91504825e-02 -5.30604303e-01 7.25555122e-01 8.67573798e-01 4.21232253e-01 6.42236173e-02 3.41171771e-01 5.69739878e-01 -5.75340390e-02 -2.20257267e-01 -1.08174670e+00 -3.95753264e-01 4.55395103e-01 7.57521331e-01 -9.52735543e-01 -4.03027326e-01 -2.14624941e-01 6.79198563e-01 3.66961718e-01 5.07482588e-01 -8.30244482e-01 -1.96721807e-01 5.12371421e-01 4.94019777e-01 1.59235775e-01 -7.18998849e-01 4.08376992e-01 -1.14017940e+00 1.82501192e-03 -1.00437152e+00 7.17861474e-01 -9.11606431e-01 -1.08130395e+00 7.73334265e-01 3.96534830e-01 -7.85280168e-01 -9.99893069e-01 -4.65594679e-01 -4.03958529e-01 4.51404542e-01 -7.55835891e-01 -1.24755776e+00 -6.49163797e-02 7.40399361e-01 1.04002133e-01 -6.24491721e-02 8.10649455e-01 -3.63011152e-01 -1.12693071e-01 1.54660553e-01 -2.97031432e-01 -1.38433173e-01 1.42024964e-01 -1.50662100e+00 1.10890079e+00 1.20638621e+00 7.70033777e-01 6.45902455e-01 9.90032554e-01 -1.05536890e+00 -1.77516162e+00 -1.14950085e+00 9.90309596e-01 -4.01232600e-01 1.03123581e+00 -5.50455749e-01 -1.07550132e+00 1.39099514e+00 -5.10917790e-02 2.14872375e-01 -9.07036290e-02 8.09868723e-02 -6.40313148e-01 -2.88776904e-01 -9.78603423e-01 6.56594396e-01 1.92096698e+00 -7.50778317e-01 -7.14249253e-01 3.16284269e-01 1.02778995e+00 -7.41124511e-01 -9.41276908e-01 2.66736031e-01 2.35073611e-01 -7.55797565e-01 7.27946103e-01 -1.05180848e+00 -2.09729686e-01 -4.27302301e-01 -5.38237654e-02 -1.29599881e+00 -4.88565266e-01 -1.13588119e+00 -6.65012896e-01 1.00756633e+00 3.40022862e-01 -1.07276487e+00 6.24028683e-01 3.70626360e-01 -2.27695465e-01 -1.93832502e-01 -8.43849123e-01 -1.16547751e+00 -1.89234108e-01 -6.20368600e-01 1.24523389e+00 8.33290935e-01 6.74516708e-02 1.53704122e-01 -1.49454728e-01 2.05250040e-01 3.73784900e-01 2.95748889e-01 6.66445851e-01 -1.58041656e+00 -4.29179937e-01 -3.09650928e-01 -6.47752941e-01 -3.45868498e-01 4.93314773e-01 -1.29455197e+00 -1.95022509e-01 -1.69026887e+00 -1.90987319e-01 -2.64728397e-01 -3.41785490e-03 6.39197946e-01 2.58072436e-01 -3.42344344e-01 1.79878697e-01 -9.39726457e-02 -3.26755583e-01 3.68265867e-01 1.34519124e+00 -1.12740070e-01 -1.77568614e-01 -1.05140701e-01 -2.73311943e-01 3.64486039e-01 6.03139162e-01 -6.11481667e-01 -1.14164078e+00 -1.01670623e-01 6.71098888e-01 4.78177875e-01 3.08919013e-01 -6.36408150e-01 2.43792180e-02 -6.60587609e-01 -4.22024041e-01 -3.94853115e-01 -1.73383638e-01 -7.88638353e-01 1.27040172e+00 6.57887638e-01 -2.10565090e-01 9.52409923e-01 8.59063193e-02 9.04133439e-01 -1.52289882e-01 4.24226746e-02 2.78794676e-01 -6.71678334e-02 -8.88442874e-01 5.54633379e-01 -3.40823323e-01 3.52332257e-02 1.19444478e+00 -1.66602001e-01 -3.90485913e-01 -4.60762382e-01 -1.23218989e+00 -4.49982621e-02 7.73790240e-01 6.24067664e-01 4.39908147e-01 -1.31005275e+00 -3.41997653e-01 1.66184187e-01 2.24134430e-01 -3.21030557e-01 2.43770182e-02 2.76456982e-01 -6.85745955e-01 3.43986928e-01 -6.96107596e-02 -4.24259305e-01 -1.10090673e+00 1.05544555e+00 5.63760757e-01 -7.33922958e-01 -8.45341444e-01 -6.32486269e-02 3.75989452e-02 -6.20281696e-01 -2.42032036e-01 -9.27067697e-01 2.76004583e-01 -3.11209202e-01 6.24151900e-02 9.69386846e-02 2.74813652e-01 -5.01666665e-01 -3.16948056e-01 2.92093873e-01 1.51361495e-01 -2.28712559e-01 1.19806206e+00 -2.79904120e-02 -3.14178705e-01 7.23807156e-01 5.39655745e-01 8.28585327e-02 -1.09589446e+00 -7.34386072e-02 4.62230176e-01 -2.13297635e-01 -5.21800578e-01 -5.99551797e-01 -8.90196204e-01 1.70078322e-01 1.80189330e-02 9.33038533e-01 1.03479552e+00 2.26708546e-01 3.80386919e-01 5.50838828e-01 8.93987060e-01 -9.39672232e-01 -1.76904395e-01 6.35456800e-01 1.14700162e+00 -3.75637859e-01 -4.08946872e-02 -7.28965223e-01 -4.33778673e-01 1.17886889e+00 3.70003581e-01 -8.83485749e-03 6.21716082e-01 3.83587062e-01 -2.39356741e-01 -3.65506709e-01 -1.29537761e+00 -2.16497317e-01 1.46602169e-01 1.07482708e+00 1.82333235e-02 3.99083883e-01 -3.56615663e-01 4.75971885e-02 -3.43559653e-01 -2.57522285e-01 9.20124650e-01 1.04272032e+00 -7.30675682e-02 -1.70924902e+00 5.99001767e-03 2.67707914e-01 9.38078985e-02 2.54565984e-01 -5.76183796e-01 1.43948901e+00 -2.29967549e-01 6.72661543e-01 2.60936528e-01 -3.61032188e-01 6.11779213e-01 1.53091535e-01 9.20828462e-01 -7.78548181e-01 -4.41523045e-01 -3.76333505e-01 5.17042339e-01 -7.08825588e-01 -4.63547975e-01 -8.32746148e-01 -1.56997049e+00 -3.51971030e-01 1.93495721e-01 2.12214589e-01 2.51647025e-01 6.27295911e-01 3.61562014e-01 7.00621903e-01 4.84295964e-01 -1.88038826e-01 -2.78704852e-01 -5.47035992e-01 -8.51425052e-01 7.01381683e-01 4.85768355e-02 -5.89450479e-01 -1.69226199e-01 3.44205260e-01]
[7.181717872619629, 6.07648229598999]
0f761c39-a855-4259-9032-347ef449ba8c
bottom-up-skill-discovery-from-unsegmented
2109.13841
null
https://arxiv.org/abs/2109.13841v2
https://arxiv.org/pdf/2109.13841v2.pdf
Bottom-Up Skill Discovery from Unsegmented Demonstrations for Long-Horizon Robot Manipulation
We tackle real-world long-horizon robot manipulation tasks through skill discovery. We present a bottom-up approach to learning a library of reusable skills from unsegmented demonstrations and use these skills to synthesize prolonged robot behaviors. Our method starts with constructing a hierarchical task structure from each demonstration through agglomerative clustering. From the task structures of multi-task demonstrations, we identify skills based on the recurring patterns and train goal-conditioned sensorimotor policies with hierarchical imitation learning. Finally, we train a meta controller to compose these skills to solve long-horizon manipulation tasks. The entire model can be trained on a small set of human demonstrations collected within 30 minutes without further annotations, making it amendable to real-world deployment. We systematically evaluated our method in simulation environments and on a real robot. Our method has shown superior performance over state-of-the-art imitation learning methods in multi-stage manipulation tasks. Furthermore, skills discovered from multi-task demonstrations boost the average task success by $8\%$ compared to those discovered from individual tasks.
['Yuke Zhu', 'Peter Stone', 'Yifeng Zhu']
2021-09-28
null
null
null
null
['robot-manipulation']
['robots']
[ 2.67984003e-01 3.47669154e-01 4.18865383e-02 -1.35298461e-01 -8.56552601e-01 -6.62386179e-01 4.70142096e-01 -2.49043837e-01 -5.00902891e-01 1.10992336e+00 -8.00365880e-02 -2.79380493e-02 -3.88083458e-01 -2.35646851e-02 -1.07031536e+00 -4.81798291e-01 -6.50917113e-01 1.02658367e+00 6.09313190e-01 -3.84318531e-01 3.39262754e-01 2.97524810e-01 -1.88718092e+00 2.72371858e-01 1.15066814e+00 4.33619201e-01 1.07545722e+00 8.42012763e-01 4.42506224e-01 1.07846975e+00 -7.42980182e-01 4.37607735e-01 5.29960632e-01 -4.25659031e-01 -9.75743890e-01 2.98677176e-01 -1.21569131e-02 -5.22217512e-01 -2.66389817e-01 7.84431040e-01 2.59490639e-01 4.57851321e-01 7.21699715e-01 -1.59780145e+00 -1.16498405e-02 7.47716427e-01 -1.41767696e-01 -1.26340330e-01 4.43538725e-01 6.18260860e-01 7.65553594e-01 -3.37075084e-01 7.69529283e-01 1.43681443e+00 3.60901207e-01 6.16341472e-01 -1.24780548e+00 -7.18853831e-01 2.95075893e-01 2.19918966e-01 -8.42921138e-01 -6.15151338e-02 1.96140096e-01 -6.99631453e-01 1.42139685e+00 -3.97066265e-01 8.08216512e-01 1.44169128e+00 3.47801268e-01 1.08640766e+00 1.23791254e+00 -1.81464881e-01 2.15740576e-01 -3.53319794e-01 -2.56234616e-01 9.28587079e-01 -4.36107107e-02 5.62991798e-01 -3.86281013e-01 -5.33012114e-02 9.36062038e-01 8.36758816e-04 -2.32859235e-02 -8.50401759e-01 -1.69901180e+00 5.49305677e-01 4.49565768e-01 1.68261185e-01 -5.67204893e-01 5.40061414e-01 4.22050148e-01 8.94127786e-01 -2.53728956e-01 1.13998091e+00 -7.60404229e-01 -4.25074726e-01 -3.78556192e-01 6.21039152e-01 9.53368485e-01 1.56672704e+00 7.67818511e-01 -8.60911384e-02 -9.82199684e-02 5.32617569e-01 -1.91032320e-01 4.67124492e-01 4.30508912e-01 -1.56658912e+00 4.11391795e-01 5.43459833e-01 4.83807504e-01 -6.18948750e-02 -6.59280062e-01 5.51227331e-02 -1.20380774e-01 8.99071991e-01 4.10406202e-01 -2.81981200e-01 -1.03977275e+00 1.58583570e+00 1.77936256e-01 1.80247258e-02 2.22405255e-01 8.05008650e-01 2.37829238e-01 3.85545194e-01 2.55343653e-02 -4.12736572e-02 9.75230217e-01 -1.60872149e+00 -3.36925477e-01 -1.87404558e-01 5.94946682e-01 -2.55707413e-01 1.13592958e+00 7.78917253e-01 -8.88714612e-01 -9.12196875e-01 -9.97143209e-01 2.80197352e-01 -1.55435503e-02 2.29588315e-01 5.81011176e-01 -1.62793353e-01 -1.01621628e+00 1.12276423e+00 -1.20080853e+00 -5.16520560e-01 1.01953462e-01 6.56611800e-01 -2.18762353e-01 2.35478535e-01 -5.74488461e-01 1.29172945e+00 7.72445917e-01 -2.26099283e-01 -2.20356441e+00 -2.25820497e-01 -8.80645454e-01 -1.02756113e-01 9.78535771e-01 -6.80588663e-01 1.89413905e+00 -5.93596756e-01 -1.89028704e+00 4.35238153e-01 2.92904764e-01 -4.39916998e-01 3.88957351e-01 -4.06448513e-01 3.86814207e-01 1.78003415e-01 4.89005923e-01 1.06927443e+00 1.09219265e+00 -1.50917089e+00 -1.25294662e+00 1.22162826e-01 4.36644793e-01 4.79613066e-01 1.01751946e-01 -1.54914007e-01 -1.30111635e-01 -2.08167717e-01 -2.84363955e-01 -1.59884405e+00 -4.92214173e-01 -4.26956922e-01 -1.06127918e-01 -6.40529275e-01 6.85184479e-01 -3.59063566e-01 2.80633003e-01 -1.88541889e+00 1.22949278e+00 -1.14214212e-01 1.83252379e-01 -2.08282828e-01 -3.60563129e-01 6.45469427e-01 2.68343180e-01 -5.31094491e-01 -1.65449455e-01 -2.84934312e-01 3.50000501e-01 5.14276564e-01 -2.33533978e-01 1.00465588e-01 -4.32457067e-02 9.90463734e-01 -1.50149441e+00 -3.84213358e-01 2.57224441e-01 -2.80187070e-01 -5.21313012e-01 7.06404030e-01 -7.57685184e-01 1.10686350e+00 -6.36513293e-01 6.05403721e-01 -2.96689957e-01 -1.12589151e-01 4.22897667e-01 4.52054322e-01 -1.05109088e-01 1.37158290e-01 -6.96697354e-01 2.35294366e+00 -6.02506697e-01 4.21473116e-01 2.97954977e-01 -9.84610319e-01 7.82307982e-01 3.01043987e-01 5.61874330e-01 -2.74420232e-01 2.33446993e-02 3.71463656e-01 4.37988997e-01 -8.57681036e-01 4.58188295e-01 2.30770960e-01 -5.87732196e-01 4.86744612e-01 5.59910536e-01 -9.73996043e-01 5.15932143e-01 2.14181673e-02 1.42711270e+00 9.67390537e-01 3.41826618e-01 -2.57120758e-01 8.76966193e-02 7.17728436e-01 3.22719634e-01 9.96571362e-01 -1.99119121e-01 -3.59667391e-02 4.45819378e-01 -1.89042568e-01 -1.21782732e+00 -1.08768487e+00 5.43898761e-01 1.51114285e+00 2.78815061e-01 -3.66748035e-01 -5.70213914e-01 -7.66544640e-01 1.77832142e-01 5.15514374e-01 -6.17404103e-01 -1.48649681e-02 -9.39082503e-01 1.97622851e-02 3.02380592e-01 5.06569743e-01 3.50482851e-01 -1.94607186e+00 -1.56767428e+00 3.64731550e-01 -5.38209220e-03 -1.06133831e+00 -1.53254882e-01 6.02478743e-01 -7.94224560e-01 -1.38012052e+00 -6.78028345e-01 -1.31050766e+00 6.36992395e-01 2.21569613e-01 9.15267587e-01 -9.42333043e-03 -3.58851880e-01 6.23574078e-01 -6.23513341e-01 -4.91662979e-01 -8.05240929e-01 1.45046964e-01 4.42696989e-01 -9.59327579e-01 -1.79201543e-01 -7.85088956e-01 -2.87496865e-01 3.33252400e-01 -1.78002715e-01 1.51803866e-01 1.11683261e+00 1.19231606e+00 3.42222452e-01 1.31962210e-01 5.15936255e-01 -2.81895638e-01 9.69993412e-01 -2.24319026e-01 -8.76470506e-01 1.79902583e-01 -3.33299845e-01 3.13844562e-01 6.63785577e-01 -8.39008689e-01 -9.63259757e-01 3.90189320e-01 4.35322434e-01 -6.19079947e-01 -3.84435207e-01 1.92652762e-01 4.27154541e-01 1.07010134e-01 7.01614976e-01 2.92786449e-01 2.34069258e-01 -3.69717270e-01 5.30412495e-01 4.30680007e-01 7.20169902e-01 -1.09206986e+00 6.92285597e-01 1.75958112e-01 -9.49477255e-02 -5.71033716e-01 -2.28127271e-01 -4.46418732e-01 -9.05137181e-01 -3.12786639e-01 8.62654448e-01 -9.71361637e-01 -1.33154130e+00 4.59181130e-01 -9.49119329e-01 -1.43116260e+00 -2.85089076e-01 5.55789590e-01 -1.47012281e+00 1.87928423e-01 -8.12481105e-01 -6.80699527e-01 -2.18133330e-02 -1.66957724e+00 1.31378317e+00 6.26536012e-02 -4.37647521e-01 -4.22352701e-01 1.39976218e-01 -2.11270973e-02 1.53006911e-01 2.38608867e-01 7.57475376e-01 -5.14690936e-01 -6.80144250e-01 3.69784474e-01 1.21702805e-01 2.91080717e-02 4.32991683e-02 -5.72925389e-01 -2.70802855e-01 -8.30219567e-01 -6.11401685e-02 -1.18986428e+00 5.78289807e-01 1.14228591e-01 8.72934043e-01 3.78084742e-02 -6.62280679e-01 1.93667457e-01 8.45054865e-01 4.66134787e-01 1.46669775e-01 5.74607432e-01 5.03894448e-01 7.72647858e-01 1.33926213e+00 2.81026185e-01 2.18584925e-01 7.36239254e-01 6.41471982e-01 5.03302097e-01 -6.06740564e-02 -4.56160367e-01 6.79400206e-01 5.20644903e-01 -1.87989518e-01 3.03180486e-01 -8.24885368e-01 7.07535088e-01 -2.33387232e+00 -7.51453340e-01 2.05737308e-01 1.61856151e+00 6.22296989e-01 2.71610290e-01 4.94137615e-01 -2.65816629e-01 2.04589188e-01 -4.25876766e-01 -9.40038025e-01 -1.65323287e-01 5.17775595e-01 1.85261905e-01 3.00808221e-01 3.46787423e-01 -1.05828726e+00 1.38780689e+00 6.40999174e+00 6.53812826e-01 -5.89678466e-01 7.73981512e-02 -3.99693906e-01 -2.55953491e-01 5.34651637e-01 1.45328920e-02 -4.97104704e-01 1.81238323e-01 6.70609355e-01 -1.29335433e-01 7.84158111e-01 1.29755080e+00 1.52801514e-01 -5.20845830e-01 -1.60717154e+00 5.95314622e-01 -1.93685219e-01 -6.97957158e-01 -4.72333252e-01 7.31883049e-02 8.85730386e-01 3.30210745e-01 -2.70933241e-01 1.08580959e+00 1.35177541e+00 -1.11723232e+00 8.31592619e-01 9.48940963e-02 6.52275324e-01 -3.82898510e-01 3.57408613e-01 1.01964104e+00 -1.12855458e+00 -6.87288165e-01 -2.63668329e-01 -4.45119917e-01 3.22853684e-01 -5.56925952e-01 -1.19685340e+00 5.66473126e-01 8.15729618e-01 8.33418727e-01 -2.23636851e-01 7.19712734e-01 -6.94852650e-01 6.11150218e-03 -1.83372855e-01 -3.85966659e-01 5.01475036e-01 -1.95193276e-01 6.38051629e-01 8.40038061e-01 2.75639832e-01 6.45334348e-02 7.46016741e-01 7.53333628e-01 3.77538890e-01 -5.14519095e-01 -8.74849141e-01 4.27276641e-02 4.18672562e-01 1.07891667e+00 -7.15417743e-01 -5.60715199e-01 -2.27932073e-02 1.22645962e+00 7.73313582e-01 3.15957218e-01 -8.84250879e-01 -3.16779852e-01 5.07632017e-01 -5.52057564e-01 4.25758541e-01 -6.04565263e-01 2.04426929e-01 -8.88509274e-01 -5.39216958e-02 -1.43846595e+00 1.02869332e-01 -8.93638849e-01 -8.96619081e-01 5.89112282e-01 5.24853289e-01 -1.50923169e+00 -7.63583958e-01 -7.35188544e-01 -3.46778065e-01 5.44845581e-01 -1.20102477e+00 -7.84523070e-01 -5.46027839e-01 5.04643798e-01 1.21676862e+00 -4.46338326e-01 1.04637778e+00 -3.69754523e-01 -1.32197231e-01 1.84599459e-02 -6.55560791e-02 -2.71538913e-01 5.62293887e-01 -1.41356456e+00 4.95745748e-01 3.77139151e-01 -1.86464444e-01 5.81644893e-01 9.23695028e-01 -7.84939945e-01 -1.22610319e+00 -6.88472033e-01 -1.82940364e-02 -6.66755915e-01 9.46003020e-01 -5.09129822e-01 -5.65999389e-01 1.02310431e+00 4.54887718e-01 -6.54404342e-01 -1.41688645e-01 1.61209684e-02 -5.51634039e-05 4.12073165e-01 -8.07161927e-01 7.72458971e-01 1.71537471e+00 -1.75487116e-01 -1.36842501e+00 6.13259673e-01 8.64380002e-01 -7.28962600e-01 -8.19911659e-01 5.82706332e-01 6.73372924e-01 -5.69484174e-01 7.93551326e-01 -7.51463473e-01 5.88574946e-01 -3.52987677e-01 3.06759149e-01 -1.95166957e+00 -2.92956591e-01 -8.02431107e-01 -1.47039115e-01 1.67468756e-01 4.00470406e-01 -4.15872365e-01 6.20256126e-01 -1.17991105e-01 -6.17936671e-01 -5.20842314e-01 -6.33821785e-01 -1.30577564e+00 2.56749354e-02 1.42433628e-01 1.58279777e-01 3.66734773e-01 7.94738352e-01 4.47134107e-01 -5.01929164e-01 -5.80082238e-02 7.36717463e-01 4.74523365e-01 1.49774218e+00 -1.12204194e+00 -7.37307608e-01 -4.02011156e-01 6.79023042e-02 -1.63737977e+00 6.77812934e-01 -8.10802221e-01 8.74925196e-01 -1.51030290e+00 2.27739319e-01 -5.85477293e-01 8.50319341e-02 6.69196963e-01 -5.16649112e-02 -3.53846699e-01 4.45160836e-01 4.84752566e-01 -1.11603534e+00 6.74691617e-01 1.70309997e+00 -1.52960420e-01 -4.85136002e-01 6.80829224e-04 -9.19755101e-02 6.60251617e-01 9.45768118e-01 -4.84023184e-01 -6.62500083e-01 -3.93196911e-01 -1.84487194e-01 3.11729968e-01 1.80077553e-01 -1.28404140e+00 2.51905173e-01 -3.12814862e-01 -7.77317658e-02 -3.41378361e-01 6.70322657e-01 -8.68044376e-01 1.42530715e-02 1.05796576e+00 -3.55587810e-01 1.71034440e-01 1.49280101e-01 7.87113369e-01 3.70165259e-02 -1.81458175e-01 3.07328731e-01 -6.50085747e-01 -1.21791172e+00 -2.04918653e-01 -8.01295340e-01 -2.08596423e-01 1.55083656e+00 -1.78845108e-01 -3.45616639e-01 -3.27100247e-01 -1.09174860e+00 1.04494488e+00 5.74174762e-01 7.00950205e-01 4.76787418e-01 -8.42538595e-01 -5.48154354e-01 -1.06167123e-01 2.46674985e-01 1.77509025e-01 -9.41689238e-02 7.44012296e-01 -2.23783433e-01 4.56084579e-01 -7.96788692e-01 -7.81341255e-01 -1.35154235e+00 7.21482396e-01 1.19846769e-01 -4.51486975e-01 -1.03700125e+00 6.78547442e-01 1.95727035e-01 -6.80324376e-01 4.60411966e-01 -6.53583884e-01 -7.11476654e-02 -4.47538942e-01 -8.32177997e-02 2.72288412e-01 -5.92847466e-01 -1.53837681e-01 -1.48982167e-01 4.18123126e-01 5.54131381e-02 -4.05304074e-01 1.29446161e+00 1.71266884e-01 9.09934491e-02 5.14128327e-01 5.80868185e-01 -6.62016392e-01 -2.02226424e+00 6.38671368e-02 2.72208810e-01 -7.59916678e-02 -8.91000867e-01 -7.44050920e-01 -1.52178794e-01 5.08709610e-01 -1.22148767e-02 -5.74372225e-02 6.62970603e-01 2.31412813e-01 5.03773510e-01 1.23419714e+00 1.31434047e+00 -1.36119926e+00 9.80904102e-01 8.64908636e-01 1.21969986e+00 -1.35993171e+00 -1.40721858e-01 -1.87108383e-01 -9.70587015e-01 8.81167233e-01 9.29344058e-01 -5.52959919e-01 4.49034758e-02 2.44433492e-01 -8.29658061e-02 -2.69059122e-01 -9.60400939e-01 -4.59550768e-01 -1.62949756e-01 8.99011374e-01 -3.49762887e-01 1.58003300e-01 -1.37048036e-01 2.10352913e-01 -3.89472157e-01 5.52172307e-03 5.30585229e-01 1.48506820e+00 -7.57811248e-01 -1.00556707e+00 -2.12562069e-01 3.49800348e-01 1.68035716e-01 4.46365148e-01 -3.90823573e-01 1.05169284e+00 -3.29851881e-02 1.03352046e+00 -2.14390665e-01 -6.02647066e-01 5.74307561e-01 1.18474595e-01 1.10929549e+00 -1.11591268e+00 -6.03376210e-01 -6.14583232e-02 3.10855985e-01 -8.40776443e-01 -4.33618367e-01 -9.06797588e-01 -1.54796028e+00 1.74863741e-01 6.24523088e-02 2.19246626e-01 5.32197654e-01 9.16989267e-01 1.86591849e-01 9.28621233e-01 2.39773765e-01 -1.59695578e+00 -1.04495573e+00 -1.42211044e+00 -2.86713660e-01 6.45830035e-01 2.66079158e-01 -1.25483119e+00 -1.96576878e-01 2.69258618e-02]
[4.476128101348877, 0.9703730344772339]
8260d72d-dd4e-42aa-ab15-f07239146990
findings-of-the-tsar-2022-shared-task-on
2302.02888
null
https://arxiv.org/abs/2302.02888v1
https://arxiv.org/pdf/2302.02888v1.pdf
Findings of the TSAR-2022 Shared Task on Multilingual Lexical Simplification
We report findings of the TSAR-2022 shared task on multilingual lexical simplification, organized as part of the Workshop on Text Simplification, Accessibility, and Readability TSAR-2022 held in conjunction with EMNLP 2022. The task called the Natural Language Processing research community to contribute with methods to advance the state of the art in multilingual lexical simplification for English, Portuguese, and Spanish. A total of 14 teams submitted the results of their lexical simplification systems for the provided test data. Results of the shared task indicate new benchmarks in Lexical Simplification with English lexical simplification quantitative results noticeably higher than those obtained for Spanish and (Brazilian) Portuguese.
['Marcos Zampieri', 'Kai North', 'Matthew Shardlow', 'Kim Cheng SHEANG', 'Daniel Ferrés', 'Sanja Štajner', 'Horacio Saggion']
2023-02-06
null
null
null
null
['lexical-simplification']
['natural-language-processing']
[-3.38413924e-01 3.85419667e-01 1.74098625e-03 -9.04462188e-02 -1.15823805e+00 -5.49300373e-01 6.18648231e-01 7.94238806e-01 -1.07775450e+00 9.99393463e-01 8.37576032e-01 -2.13042244e-01 1.81052878e-01 -3.64957154e-01 -3.79387975e-01 2.34182313e-01 5.21832824e-01 9.66589212e-01 -1.01454213e-01 -9.23870087e-01 7.11866096e-02 6.71690345e-01 -1.12522137e+00 5.64252853e-01 1.25522280e+00 -2.55177259e-01 2.72861660e-01 5.34543455e-01 -1.67196974e-01 7.49233246e-01 -8.24811935e-01 -9.41093326e-01 3.28360260e-01 9.79705453e-02 -1.07689750e+00 -7.14315414e-01 6.88028216e-01 2.43140712e-01 -3.82850081e-01 1.19257045e+00 8.17614794e-01 4.79394048e-01 5.84722519e-01 -7.42092371e-01 -7.18411326e-01 1.28966022e+00 -5.20816594e-02 3.45995873e-01 8.95837307e-01 2.45662197e-01 9.98080850e-01 -1.11420894e+00 1.32731616e+00 1.74975872e+00 1.07298636e+00 4.22721386e-01 -1.09994888e+00 -4.03989702e-01 6.95146769e-02 1.47657067e-01 -1.83022809e+00 -9.98963118e-01 -7.11185634e-02 -1.30338997e-01 1.98467517e+00 5.86429596e-01 3.72301877e-01 7.18078136e-01 2.66890854e-01 7.64605224e-01 6.90403283e-01 -8.30176353e-01 -5.38347185e-01 9.94038209e-02 3.01425904e-01 5.23526609e-01 6.37494028e-01 -4.90597248e-01 -5.12726426e-01 1.37516975e-01 -1.82697386e-01 -7.67829537e-01 3.59335542e-02 4.22321469e-01 -1.33556426e+00 6.76995516e-01 -1.08594246e-01 4.67474133e-01 -2.83639103e-01 -3.35171700e-01 1.10586298e+00 6.10840917e-01 8.06070387e-01 1.14204574e+00 -6.56087101e-01 -9.74601880e-02 -8.42339993e-01 7.44429171e-01 1.06481171e+00 1.38030505e+00 4.99587625e-01 7.12137967e-02 -6.03900313e-01 1.17614746e+00 -1.87538952e-01 1.06589460e+00 3.67633343e-01 -9.53415692e-01 1.14635599e+00 2.56898940e-01 8.76147300e-02 -7.55573630e-01 -7.41433144e-01 2.16763318e-01 -6.42781734e-01 -1.21212862e-01 2.11180702e-01 -2.67870635e-01 -5.45127034e-01 1.46431673e+00 -1.60635754e-01 -1.17781961e+00 1.54382691e-01 1.32393941e-01 1.18684471e+00 6.56682968e-01 5.44060528e-01 -5.08521438e-01 1.15203762e+00 -1.16738892e+00 -1.34238589e+00 1.33698899e-02 1.43919706e+00 -1.49000704e+00 1.38525462e+00 4.60095108e-01 -1.62132883e+00 -4.30152237e-01 -8.42133820e-01 -1.01566768e+00 -1.11788690e+00 -3.53418328e-02 2.16909915e-01 3.35564762e-01 -1.00644088e+00 5.79529941e-01 -5.33212900e-01 -8.25462520e-01 1.19127221e-01 1.88204974e-01 -7.10418046e-01 -2.59845257e-01 -1.43602920e+00 1.91627717e+00 6.13109291e-01 -3.17772895e-01 -3.08742315e-01 -1.06576526e+00 -1.12029409e+00 -1.78810775e-01 9.33368504e-02 -5.94521284e-01 1.31540477e+00 -3.44397038e-01 -1.24692810e+00 1.30375993e+00 -2.57114321e-01 -4.68944222e-01 7.28067636e-01 -7.89857030e-01 -6.72989011e-01 -7.03119040e-01 5.91707408e-01 7.70596087e-01 5.37290089e-02 -5.45099974e-01 -8.35729182e-01 4.47421223e-02 -1.22241877e-01 7.55208373e-01 -1.40803859e-01 6.82906985e-01 -3.83165807e-01 -9.51602221e-01 -7.67400742e-01 -7.82323062e-01 -1.85694978e-01 -8.96184504e-01 -5.60771525e-01 -5.62577367e-01 4.57229733e-01 -1.55347466e+00 1.75259471e+00 -1.86693692e+00 6.50931001e-01 -2.04507485e-01 1.65741280e-01 6.47952437e-01 -4.46610183e-01 8.92638266e-01 -1.98272675e-01 7.00961590e-01 -1.09813968e-02 -8.64310801e-01 3.52526039e-01 -1.07481945e-02 -1.65353134e-01 3.37114424e-01 -1.96699172e-01 1.07338119e+00 -8.58813226e-01 -5.60436964e-01 3.92847985e-01 2.69674689e-01 -4.76560235e-01 -2.49596268e-01 -4.47336994e-02 1.66312084e-01 1.98087245e-01 4.10098106e-01 4.42637771e-01 8.51337492e-01 1.18870832e-01 -3.97404999e-01 -6.66976988e-01 7.14187980e-01 -8.86077166e-01 1.91261351e+00 -7.77401924e-01 8.71472955e-01 -6.97058961e-02 7.64862150e-02 2.72171408e-01 3.67473632e-01 3.63685004e-02 -1.21787775e+00 9.32707563e-02 4.15192187e-01 5.09703010e-02 -4.33234423e-01 1.16439486e+00 4.49524671e-02 -6.17566288e-01 4.13104683e-01 2.39960402e-01 -8.65307927e-01 1.06678867e+00 3.84930760e-01 1.01630330e+00 -2.31388416e-02 9.92923260e-01 -8.72100711e-01 8.07345748e-01 1.00599647e-01 4.59682673e-01 6.90105557e-01 -1.94330603e-01 4.93471593e-01 -8.16156995e-03 -5.76174140e-01 -1.62113714e+00 -8.44665468e-01 3.07353027e-02 1.49643016e+00 -4.36686218e-01 -1.32883751e+00 -1.04887426e+00 -5.37516475e-01 -6.78624064e-02 1.31141758e+00 -4.36225504e-01 -2.48944573e-02 -1.19637096e+00 -3.54029417e-01 1.24193168e+00 7.67300744e-03 -1.79819912e-02 -1.64158762e+00 -1.64397404e-01 2.28344768e-01 -6.96521997e-01 -1.27187455e+00 -6.24168813e-01 2.58392747e-02 -3.71681660e-01 -9.80275869e-01 -2.65949398e-01 -8.14507961e-01 -2.94751162e-03 -1.83187842e-01 1.66208506e+00 4.29961570e-02 -3.23598683e-01 1.07581511e-01 -2.37489313e-01 -9.05347764e-01 -7.47127295e-01 9.72661078e-01 2.53946662e-01 -1.18772995e+00 5.13852596e-01 -1.07905462e-01 3.25449675e-01 -3.62718165e-01 -8.67169261e-01 -8.27659480e-03 1.89068407e-01 4.56304342e-01 6.39012039e-01 -2.11989492e-01 3.75135064e-01 -1.29494715e+00 1.38679719e+00 -1.35870561e-01 -3.28290671e-01 5.35547912e-01 -3.16229969e-01 -7.85811171e-02 9.68518913e-01 3.60093167e-04 -1.26829112e+00 -3.16199273e-01 -3.82228136e-01 3.11627448e-01 -5.28908931e-02 6.47190988e-01 -6.06441617e-01 4.02993150e-02 7.60044277e-01 -4.10619438e-01 -5.66912949e-01 -7.87266076e-01 8.68154049e-01 4.02520239e-01 6.42528236e-01 -7.07128763e-01 7.12522805e-01 -8.79967660e-02 -2.11992383e-01 -1.21429110e+00 -9.25479412e-01 -2.15041563e-01 -9.87489998e-01 3.57352555e-01 8.48364592e-01 -9.87118125e-01 7.28773847e-02 4.51734096e-01 -1.60723543e+00 -3.95383656e-01 -7.62287259e-01 5.40376492e-02 -1.60319373e-01 3.77129495e-01 -8.24518085e-01 2.19580121e-02 -8.27933252e-01 -1.04395819e+00 1.00742567e+00 -2.38552377e-01 -1.18231511e+00 -1.43157494e+00 6.26065791e-01 1.89144388e-01 5.96292019e-01 -3.30813928e-03 1.28589427e+00 -1.02518809e+00 2.62288809e-01 -2.52662420e-01 -1.34985626e-01 3.19585860e-01 -4.89183925e-02 4.18349415e-01 -4.17007089e-01 -2.18908697e-01 -4.27301645e-01 -3.43465030e-01 4.21531171e-01 2.35491276e-01 5.76136172e-01 -2.52504051e-01 -1.25428408e-01 5.86402774e-01 9.09114838e-01 -2.55705714e-01 7.89475977e-01 6.54628217e-01 1.15418875e+00 5.67871094e-01 7.11929202e-01 1.82044953e-01 6.80487216e-01 7.65086353e-01 -2.98799485e-01 -5.30746728e-02 -9.06082332e-01 -6.16144799e-02 3.24289680e-01 1.87568653e+00 2.27452904e-01 -2.64720619e-01 -1.18823659e+00 5.38821578e-01 -1.73926473e+00 -5.00987828e-01 -5.34526765e-01 2.05732059e+00 1.28523052e+00 -2.88262405e-02 -9.73059908e-02 -1.00844078e-01 5.69509268e-01 1.16736636e-01 9.38463509e-02 -1.21975088e+00 -8.96082580e-01 4.92139876e-01 4.75395679e-01 1.43207204e+00 -1.03321600e+00 1.76807797e+00 7.02368879e+00 1.25466526e+00 -5.66641212e-01 5.18201888e-01 -4.15313430e-02 -2.64684469e-01 -5.40107489e-01 -2.12394461e-01 -1.07093573e+00 -1.83993950e-02 1.19460607e+00 -9.87519681e-01 7.79380620e-01 1.33296683e-01 2.90357500e-01 1.16148323e-01 -1.13507032e+00 8.76959860e-01 2.81281352e-01 -9.65550125e-01 6.69549882e-01 -4.60655421e-01 9.81817484e-01 4.25708383e-01 -1.35305718e-01 8.73032391e-01 5.00105798e-01 -1.38025177e+00 9.05710757e-01 7.97403097e-01 8.70110452e-01 -1.16399443e+00 8.23680580e-01 2.12032229e-01 -1.08728290e+00 3.63070428e-01 -4.83062863e-01 -2.26058513e-01 3.98996562e-01 5.36191344e-01 -4.38976198e-01 5.00922620e-01 5.55482566e-01 8.46312463e-01 -1.23351538e+00 7.37555265e-01 -4.22069013e-01 1.40466949e-03 -4.00005251e-01 2.74751425e-01 -1.33104861e-01 1.73499901e-02 1.12761712e+00 2.15338635e+00 -2.49033958e-01 -1.56527430e-01 1.55278265e-01 2.95104552e-02 -5.07239342e-01 9.65382993e-01 -8.74204814e-01 2.18057297e-02 8.62850010e-01 1.17625713e+00 -4.17324036e-01 -6.41008735e-01 -6.31314218e-02 8.45700741e-01 8.28418970e-01 1.03493743e-01 -6.30442679e-01 -8.87619197e-01 4.38873380e-01 -2.82786906e-01 -1.51451409e-01 -1.01082094e-01 -8.98409247e-01 -1.32922637e+00 -1.35408551e-01 -1.58271432e+00 2.93355823e-01 -6.56597376e-01 -1.30765867e+00 7.41847277e-01 2.12310299e-01 -5.18010676e-01 9.56325829e-02 -3.67510676e-01 -1.73997238e-01 1.26551330e+00 -1.10373688e+00 -1.16248751e+00 1.77145645e-01 4.84165162e-01 8.88410091e-01 -3.31696004e-01 1.29432225e+00 6.96249962e-01 -6.50244176e-01 9.18591976e-01 1.10343747e-01 -2.13250235e-01 1.18724895e+00 -1.28137732e+00 9.68890905e-01 1.03581250e+00 -4.03813608e-02 6.47283137e-01 9.30323601e-01 -9.46685016e-01 -8.18281591e-01 -1.24619031e+00 2.02952909e+00 -9.52569723e-01 7.91814446e-01 -3.03029656e-01 -8.02230060e-01 9.80996966e-01 9.78499234e-01 -6.28579617e-01 3.99577379e-01 2.40283191e-01 -1.74959511e-01 7.06212297e-02 -1.12779021e+00 1.16739082e+00 1.17720461e+00 -6.69975758e-01 -1.15257108e+00 7.48755991e-01 1.03832757e+00 -6.58448875e-01 -1.08015466e+00 3.74983996e-01 2.45174602e-01 -1.40309677e-01 6.83695853e-01 -7.57783890e-01 2.23451972e-01 -2.84141414e-02 -1.63419381e-01 -1.74482894e+00 -4.35785592e-01 -9.44833755e-01 3.21072727e-01 1.22119915e+00 7.18864143e-01 -3.71035397e-01 -1.59329727e-01 3.68139029e-01 -5.81151962e-01 9.10077095e-02 -9.33757424e-01 -5.73465466e-01 8.19445550e-01 -3.56358171e-01 4.08695936e-01 1.02980781e+00 3.10944729e-02 8.51442695e-01 2.22548395e-02 -3.27532202e-01 4.27713901e-01 -9.47366416e-01 9.34637189e-01 -1.26641345e+00 3.98597091e-01 -7.58457363e-01 3.50057594e-02 -2.72632331e-01 6.95509970e-01 -1.31403983e+00 -9.28493030e-03 -1.76489830e+00 7.79874548e-02 -6.04352579e-02 2.48152658e-01 7.54719138e-01 -3.49772125e-01 2.54925787e-01 6.66184127e-01 1.77633390e-01 -1.16661108e+00 4.98346329e-01 9.20404434e-01 -1.29006520e-01 -1.80122554e-01 -7.21715927e-01 -6.45449698e-01 8.50688934e-01 5.75418055e-01 -7.35716879e-01 9.21301618e-02 -1.16514504e+00 6.34880006e-01 -5.11099339e-01 -4.76587981e-01 -9.88009930e-01 2.07611859e-01 1.27658278e-01 -6.01411834e-02 -5.96384287e-01 3.29371430e-02 -5.59324622e-01 1.64133564e-01 3.87134969e-01 -3.25385809e-01 8.89414310e-01 7.37775028e-01 -5.18978119e-01 -1.66538090e-01 -3.06034952e-01 7.23912179e-01 -1.25298560e-01 -4.23097372e-01 -1.18182950e-01 -8.48472774e-01 8.84783983e-01 6.82332158e-01 3.11896980e-01 -5.75253785e-01 -1.59296498e-01 -7.65090585e-01 2.35288054e-01 4.65758204e-01 5.16437173e-01 6.45967051e-02 -1.13486171e+00 -1.23969924e+00 -1.44868717e-01 -1.75378080e-02 -2.44549915e-01 -2.00747073e-01 6.84408367e-01 -1.14857948e+00 1.01295483e+00 -3.61140788e-01 1.22891605e-01 -1.20451391e+00 4.59896505e-01 2.44056240e-01 -9.57024455e-01 -5.52567840e-01 5.18309414e-01 -2.69608408e-01 -1.20712554e+00 1.35479838e-01 -1.89100191e-01 -2.26169363e-01 2.05822706e-01 5.86229503e-01 1.05931079e+00 7.89526761e-01 -9.23784375e-01 -2.78586894e-01 3.92017782e-01 -5.09902477e-01 -1.79836214e-01 1.22948837e+00 -4.83726382e-01 -7.52715290e-01 4.74945068e-01 1.09363866e+00 7.62439728e-01 -1.02633744e-01 -1.11566655e-01 5.15870214e-01 2.73039162e-01 -2.93342590e-01 -1.21391690e+00 -4.83875483e-01 6.09385610e-01 -2.81549036e-03 -4.08388115e-02 7.08197653e-01 -5.44534087e-01 9.03701425e-01 1.03471661e+00 1.27119631e-01 -1.60147548e+00 -8.36699843e-01 1.69660640e+00 1.20014560e+00 -9.93146062e-01 3.49263519e-01 -4.72222418e-01 -5.92773080e-01 1.01502466e+00 3.89512420e-01 -2.03215316e-01 4.85484868e-01 3.28088403e-01 1.06551051e-01 -1.26745284e-01 -6.82672262e-01 -6.65763533e-03 3.35855156e-01 9.16636884e-01 8.21994603e-01 3.55439425e-01 -1.07886779e+00 6.52334690e-01 -6.93955064e-01 -3.84347737e-01 4.05360103e-01 7.82476068e-01 -1.65844992e-01 -1.36542118e+00 -1.29656374e-01 1.56911418e-01 -6.20275497e-01 -9.38483834e-01 -7.18461871e-01 1.33632696e+00 1.20876111e-01 6.98012054e-01 -2.50990182e-01 7.17108473e-02 1.21726406e+00 9.14644003e-02 6.79767907e-01 -1.14732981e+00 -1.37906325e+00 -3.06072265e-01 8.20556223e-01 -5.35825908e-01 2.07277715e-01 -7.96204805e-01 -1.08373284e+00 -9.15561259e-01 2.33437642e-01 2.54689097e-01 3.70984614e-01 9.25824881e-01 2.02602670e-01 7.02304006e-01 -1.66038260e-01 -1.09961998e+00 -1.68399781e-01 -1.36020589e+00 -1.05643809e-01 6.20246172e-01 -2.10859939e-01 -1.26567513e-01 -1.33391500e-01 1.28747031e-01]
[10.941949844360352, 10.409415245056152]
a2e60cb4-0d4f-4f87-9bb5-b2f85add512c
improving-the-robustness-of-federated
2204.13414
null
https://arxiv.org/abs/2204.13414v1
https://arxiv.org/pdf/2204.13414v1.pdf
Improving the Robustness of Federated Learning for Severely Imbalanced Datasets
With the ever increasing data deluge and the success of deep neural networks, the research of distributed deep learning has become pronounced. Two common approaches to achieve this distributed learning is synchronous and asynchronous weight update. In this manuscript, we have explored very simplistic synchronous weight update mechanisms. It has been seen that with an increasing number of worker nodes, the performance degrades drastically. This effect has been studied in the context of extreme imbalanced classification (e.g. outlier detection). In practical cases, the assumed conditions of i.i.d. may not be fulfilled. There may also arise global class imbalance situations like that of outlier detection where the local servers receive severely imbalanced data and may not get any samples from the minority class. In that case, the DNNs in the local servers will get completely biased towards the majority class that they receive. This would highly impact the learning at the parameter server (which practically does not see any data). It has been observed that in a parallel setting if one uses the existing federated weight update mechanisms at the parameter server, the performance degrades drastically with the increasing number of worker nodes. This is mainly because, with the increasing number of nodes, there is a high chance that one worker node gets a very small portion of the data, either not enough to train the model without overfitting or having a highly imbalanced class distribution. The chapter, hence, proposes a workaround to this problem by introducing the concept of adaptive cost-sensitive momentum averaging. It is seen that for the proposed system, there was no to minimal degradation in performance while most of the other methods hit their bottom performance before that.
['Ashish Ghosh', 'Debasrita Chakraborty']
2022-04-28
null
null
null
null
['imbalanced-classification']
['miscellaneous']
[-3.30753893e-01 2.83508766e-02 3.47614795e-01 -2.07746252e-01 -1.74381167e-01 -4.93253656e-02 2.30075568e-01 4.68539655e-01 -8.85158002e-01 9.47214842e-01 -3.45663309e-01 -7.43663907e-02 -1.63978890e-01 -8.43254507e-01 -6.74558342e-01 -1.26237428e+00 -1.06952399e-01 8.97803903e-01 5.09050786e-01 4.25331369e-02 7.38400891e-02 5.83374918e-01 -1.81091380e+00 2.81108379e-01 5.80653250e-01 9.88070130e-01 2.24604551e-02 7.26933777e-01 -4.27763432e-01 8.04201722e-01 -1.17001688e+00 -4.18399692e-01 3.60551298e-01 -2.20983207e-01 -4.66878742e-01 -2.75529563e-01 3.65714490e-01 -5.11870444e-01 7.72308409e-02 1.15176475e+00 9.61408794e-01 1.50317445e-01 2.21762091e-01 -1.55917108e+00 4.51444834e-01 7.55995154e-01 -7.60362089e-01 3.26221406e-01 -3.40923458e-01 -7.39647523e-02 7.15895295e-01 -6.51255727e-01 5.42109132e-01 9.66095626e-01 7.72597551e-01 2.99702406e-01 -1.12989569e+00 -6.93315327e-01 2.18064874e-01 2.54044414e-01 -1.09950745e+00 -1.67399034e-01 6.27358019e-01 -2.19823599e-01 8.65644395e-01 2.49959826e-01 6.21656656e-01 8.15639257e-01 3.30978572e-01 5.21881640e-01 8.16774487e-01 -2.62417048e-01 7.73080468e-01 2.42879719e-01 1.22281030e-01 8.78425911e-02 8.35478008e-01 -1.00050591e-01 -6.38971329e-01 -3.67139041e-01 2.26141885e-01 2.98175961e-01 1.31220117e-01 -4.86045122e-01 -5.66890955e-01 6.96836531e-01 3.45871419e-01 5.64732015e-01 -6.85647130e-01 1.92498773e-01 9.21500981e-01 8.19601119e-01 8.55100870e-01 4.49291468e-02 -6.02050900e-01 -3.99225980e-01 -1.24769616e+00 3.78921241e-01 1.11300087e+00 3.23631555e-01 6.82810426e-01 1.15043916e-01 3.21482897e-01 7.19280839e-01 2.57070828e-02 1.37585938e-01 7.79483557e-01 -5.43691039e-01 4.94447380e-01 6.25268400e-01 6.77149668e-02 -1.00935221e+00 -6.76967919e-01 -8.45090747e-01 -9.80618298e-01 7.69008875e-01 8.50468636e-01 -5.03822625e-01 -5.09246171e-01 1.60683584e+00 6.06209040e-01 -3.40848342e-02 8.54519606e-02 9.84883010e-01 2.58242428e-01 4.66815770e-01 -1.27712250e-01 -2.91229695e-01 9.24720049e-01 -6.16016805e-01 -6.98903501e-01 1.06292553e-02 5.54364443e-01 -8.36610913e-01 6.07424259e-01 6.94957316e-01 -1.00405157e+00 -3.16808969e-01 -1.11283040e+00 5.40372729e-01 -3.60024571e-01 -3.23568135e-01 5.28887928e-01 6.81232691e-01 -9.93656039e-01 9.11454558e-01 -1.13195086e+00 -4.63534266e-01 4.22090381e-01 5.60365081e-01 -1.40791446e-01 -7.69869611e-02 -9.01671827e-01 7.87149191e-01 3.23283583e-01 2.12633967e-01 -7.45988667e-01 -5.82927942e-01 1.37803882e-01 1.90201312e-01 1.03845946e-01 -3.94142359e-01 1.24576592e+00 -1.51039624e+00 -1.19743228e+00 5.31934381e-01 2.59018809e-01 -6.72286212e-01 1.13626587e+00 -2.88742244e-01 -5.23528531e-02 -2.89163470e-01 -4.02268320e-01 2.23889407e-02 8.02688479e-01 -1.11300850e+00 -7.44749427e-01 -8.21370721e-01 -1.99563801e-01 2.77408659e-01 -6.20408535e-01 -8.83358419e-02 9.93684754e-02 -2.73104280e-01 1.72018662e-01 -7.37771213e-01 -7.56635517e-02 -1.59361348e-01 -1.56043470e-01 -1.75790995e-01 8.84801924e-01 -1.10283889e-01 9.22594905e-01 -2.24262738e+00 -2.06993237e-01 3.58013660e-01 4.94747937e-01 3.38410795e-01 2.46964648e-01 6.66435897e-01 -1.52103111e-01 -3.26118767e-01 1.47328883e-01 -5.36441565e-01 -8.77818614e-02 3.95507991e-01 -1.80093184e-01 6.04912281e-01 -3.16999346e-01 -5.19525679e-03 -7.85623908e-01 -1.72749043e-01 -2.12606475e-01 3.95443380e-01 -5.33149898e-01 1.23840518e-01 -1.88276637e-02 4.28982712e-02 -2.43461639e-01 1.82059973e-01 8.81782174e-01 -2.17309389e-02 2.16515169e-01 7.54516721e-02 -8.69484320e-02 -4.12964709e-02 -1.71448600e+00 1.06047094e+00 -4.29402888e-01 5.84051907e-01 2.22979173e-01 -1.33218992e+00 9.56991971e-01 5.77059209e-01 7.94551790e-01 -5.69158912e-01 4.71531227e-02 6.45662248e-01 4.40715373e-01 -2.92779803e-01 3.67521793e-01 -1.88215733e-01 3.89925390e-01 6.60352349e-01 -1.16280252e-02 3.85559648e-01 1.35480672e-01 7.27123469e-02 1.41304100e+00 -2.57954150e-01 -2.23251462e-01 -1.93605170e-01 2.40218744e-01 -7.03673586e-02 5.26434600e-01 9.10223365e-01 -2.53273040e-01 3.65879864e-01 8.28150809e-01 -7.30748951e-01 -1.20674479e+00 -9.25386012e-01 -7.88035709e-03 1.41428339e+00 -1.71984538e-01 -1.08253144e-01 -7.73847044e-01 -3.37162465e-01 8.75826403e-02 2.53142059e-01 -3.03747922e-01 -3.03238481e-01 -3.16564500e-01 -1.11410248e+00 4.46647435e-01 1.54785439e-01 3.94448966e-01 -1.26895654e+00 -1.10666001e+00 4.86989081e-01 2.09549412e-01 -5.64262509e-01 2.27233097e-01 7.18405962e-01 -1.35978544e+00 -9.29468989e-01 -7.43308246e-01 -2.79603928e-01 8.00928116e-01 1.14399157e-02 1.11942637e+00 1.10497683e-01 -3.16740870e-01 -1.66720129e-04 -2.00065106e-01 -7.95791328e-01 -3.16772312e-01 3.52033287e-01 1.53595939e-01 1.61640748e-01 4.59488273e-01 -7.10616827e-01 -7.26334393e-01 8.58075097e-02 -1.15991712e+00 -3.64191979e-01 3.58489335e-01 7.14282751e-01 1.41309693e-01 3.62472981e-01 8.68320346e-01 -1.08688211e+00 6.67908907e-01 -5.68090737e-01 -4.94585663e-01 -2.56212354e-01 -7.34829605e-01 -5.96439689e-02 9.04656649e-01 -5.60259163e-01 -8.37171137e-01 -2.58924067e-01 -9.09570456e-02 -3.08227599e-01 -1.47845402e-01 3.12346131e-01 2.61315499e-02 2.61972994e-01 7.84351349e-01 -1.31634980e-01 2.49919146e-01 -6.29704118e-01 -2.17645556e-01 8.32315505e-01 -9.58302543e-02 -4.66943502e-01 3.97276074e-01 4.66999143e-01 -8.81614462e-02 -7.92532206e-01 -3.97992224e-01 -4.11509067e-01 -1.58889338e-01 -3.68823498e-01 2.49762133e-01 -6.74653590e-01 -6.71969533e-01 1.03735471e+00 -1.09392405e+00 -4.59597617e-01 -6.19253635e-01 4.30005580e-01 -2.32131451e-01 1.83874723e-02 -6.33651257e-01 -1.01180196e+00 -4.66810137e-01 -1.06810915e+00 4.70462024e-01 2.86301345e-01 -2.25937515e-01 -8.02365005e-01 1.80149958e-01 -9.55379978e-02 9.19702530e-01 1.51272058e-01 6.70533299e-01 -1.22916806e+00 -2.11303204e-01 -5.17922580e-01 1.72880199e-02 4.58330035e-01 -1.16839483e-01 1.55806839e-01 -1.01771009e+00 -4.95544076e-01 2.54565209e-01 -2.85581976e-01 4.52984661e-01 1.00713715e-01 8.29829574e-01 -1.75127029e-01 -1.89633407e-02 2.27167383e-01 1.53011370e+00 2.03288376e-01 4.50655669e-01 4.42144424e-01 3.78892064e-01 5.57208657e-01 2.62378007e-01 6.76107347e-01 -1.14606723e-01 5.72910249e-01 7.29088247e-01 -1.03717566e-01 1.33388042e-01 3.49678069e-01 4.56069380e-01 8.17664742e-01 -1.10623591e-01 -2.00218037e-01 -9.41845179e-01 5.29094517e-01 -2.03774548e+00 -8.32433224e-01 -2.52232164e-01 2.53450346e+00 9.62169647e-01 3.49235564e-01 1.76846191e-01 3.78553450e-01 7.40451038e-01 -9.23747495e-02 -6.74730420e-01 -8.08719218e-01 1.95507184e-01 1.10133857e-01 5.30921221e-01 1.94559291e-01 -6.49275422e-01 2.94713587e-01 5.05025530e+00 6.49586916e-01 -1.40900064e+00 4.26756889e-01 5.60460925e-01 -5.91767073e-01 1.47836000e-01 -3.15315843e-01 -8.02040875e-01 8.54851961e-01 1.08823431e+00 -5.41566461e-02 1.61061838e-01 9.47481394e-01 2.45965809e-01 -4.72352356e-01 -6.92845345e-01 8.84402156e-01 -2.59646147e-01 -6.90361440e-01 -4.09522474e-01 2.63500422e-01 6.81204319e-01 5.60749590e-01 -3.85382116e-01 2.01159865e-01 3.32593322e-01 -5.58294356e-01 4.43887860e-01 5.68049133e-01 1.06765382e-01 -1.11347818e+00 1.31607795e+00 7.15483785e-01 -3.21270794e-01 -2.80045211e-01 -6.74172342e-01 -2.83764333e-01 -9.40468982e-02 1.40717101e+00 -5.33945501e-01 3.36336702e-01 8.72492313e-01 -1.01739347e-01 -3.15250695e-01 1.49285424e+00 2.93653131e-01 7.43200302e-01 -8.98227453e-01 -9.22093913e-02 1.65533036e-01 -6.53165281e-02 5.13427675e-01 8.35488498e-01 2.72025347e-01 -6.17492378e-01 -8.72608200e-02 1.54415116e-01 -1.46942973e-01 2.94688582e-01 -5.52907407e-01 5.89033484e-01 1.48072928e-01 1.23495758e+00 -9.54852998e-01 -4.99814868e-01 -3.18220824e-01 8.28083158e-01 3.19917202e-01 2.05064252e-01 -6.51200116e-01 -4.59073424e-01 6.79664791e-01 2.84459412e-01 2.78712958e-01 1.36801839e-01 -2.86220968e-01 -7.95550823e-01 1.25767753e-01 -8.60086799e-01 3.75871420e-01 -2.48368695e-01 -1.38233232e+00 4.56052601e-01 -1.34504631e-01 -7.52944648e-01 -1.41137496e-01 -3.77029896e-01 -7.88257420e-01 5.97247541e-01 -9.32413816e-01 -2.78184325e-01 -2.52756715e-01 4.75803167e-01 2.60090172e-01 -1.51194334e-01 5.93329012e-01 7.81652331e-01 -5.02827883e-01 6.15923107e-01 7.14459240e-01 -1.96063638e-01 8.54827762e-01 -1.18927181e+00 -2.09374323e-01 6.98755205e-01 1.58207342e-01 2.55810916e-01 1.11991930e+00 -5.23733020e-01 -1.00761580e+00 -7.72586942e-01 8.88284802e-01 1.78370089e-03 6.59413934e-01 -3.63328218e-01 -1.12188745e+00 2.23003522e-01 3.93257551e-02 1.60502300e-01 4.74817872e-01 2.53955096e-01 3.99506092e-02 -8.61724496e-01 -1.28384256e+00 3.02179307e-01 6.44246459e-01 -1.00262957e-02 -2.60744542e-01 5.64902127e-01 -6.28082547e-03 -3.06423545e-01 -5.79629183e-01 2.00926494e-02 4.47874308e-01 -1.43071425e+00 5.33209920e-01 -5.39968431e-01 1.07073620e-01 -1.37367010e-01 2.02371240e-01 -1.38839138e+00 2.35771805e-01 -3.07184815e-01 -3.30380231e-01 1.23366749e+00 2.90498674e-01 -1.04378498e+00 1.01080716e+00 5.93821347e-01 2.70156592e-01 -7.52834022e-01 -1.36094677e+00 -8.27435076e-01 1.57111660e-01 -2.87225962e-01 4.16821897e-01 6.75130308e-01 -2.09429279e-01 8.91463831e-02 -7.49520734e-02 -1.89595923e-01 5.41197240e-01 -7.14896992e-02 8.39254141e-01 -1.40819848e+00 -4.02268529e-01 -3.78680617e-01 -5.96485019e-01 -4.03241247e-01 -4.84182954e-01 -5.34771919e-01 -6.58872426e-02 -1.15671563e+00 7.62102157e-02 -5.86661875e-01 -5.88766336e-01 3.83564085e-01 2.73601990e-02 2.04059616e-01 1.84107020e-01 3.42145771e-01 -6.06607676e-01 7.13782310e-02 8.07383358e-01 2.84173548e-01 -2.22238153e-01 4.24308300e-01 -2.23073706e-01 6.87295794e-01 1.04434431e+00 -1.00172210e+00 -4.58357692e-01 -2.94352770e-01 8.37343037e-01 -2.69740552e-01 2.86318094e-01 -1.32708001e+00 5.42787611e-01 3.82290810e-01 4.88132298e-01 -4.24182117e-01 -4.89145257e-02 -1.28580058e+00 2.74236917e-01 6.95654988e-01 -5.85976467e-02 5.04860759e-01 2.55927946e-02 4.04493541e-01 -1.32678509e-01 -4.92645353e-01 7.14811802e-01 -3.07840314e-02 -3.79078351e-02 1.05466284e-02 -5.47299385e-01 -1.38349710e-02 1.13550794e+00 -1.92396507e-01 -1.91868797e-01 -3.80074769e-01 -8.01873624e-01 1.45040721e-01 4.86385643e-01 2.11335361e-01 -1.14147142e-02 -9.53050673e-01 -4.93185371e-01 2.15804204e-01 -4.14898187e-01 2.93385476e-01 4.35456246e-01 1.13239133e+00 -5.89816153e-01 -1.90828279e-01 -1.97859451e-01 -4.84611571e-01 -1.15105021e+00 2.13118628e-01 4.52108622e-01 -5.19452989e-01 -7.49345124e-01 6.44021392e-01 -3.45260084e-01 -2.77036905e-01 5.60129941e-01 -2.07516864e-01 1.00216292e-01 7.06207514e-01 4.29230601e-01 6.03272200e-01 7.67644286e-01 -1.35239074e-02 -3.30486208e-01 1.53552210e-02 -3.36155593e-01 -8.12105369e-03 1.70597243e+00 2.01429218e-01 -2.99418539e-01 8.85770142e-01 1.04372430e+00 -1.72633547e-02 -1.29109073e+00 -9.10201967e-02 2.30806824e-02 -2.23721296e-01 -1.49433166e-01 -5.71848214e-01 -1.32420754e+00 9.40881431e-01 9.69082952e-01 7.09660351e-01 1.01962388e+00 -4.02911276e-01 6.19085908e-01 3.09931785e-01 4.75316674e-01 -1.54634905e+00 -1.27679467e-01 5.61003149e-01 4.34575647e-01 -1.00871193e+00 1.29530817e-01 5.48631430e-01 -4.36538011e-01 1.18964696e+00 6.54852271e-01 -3.63066167e-01 6.13190353e-01 6.13426566e-01 3.50766182e-01 -3.01974058e-01 -8.79906952e-01 3.27334970e-01 -5.60813010e-01 2.71509230e-01 3.12705338e-01 -6.71784207e-02 -5.00647187e-01 2.95291215e-01 -1.07387267e-01 6.32173941e-02 5.41526318e-01 9.08122718e-01 -5.41800141e-01 -1.12165546e+00 -5.22089005e-01 7.48783112e-01 -8.91978443e-01 1.76012859e-01 -6.27128333e-02 7.42246807e-01 2.90327251e-01 5.18035114e-01 5.37289619e-01 2.07599536e-01 2.65808433e-01 4.70150352e-01 1.42880559e-01 -3.70204359e-01 -9.27498281e-01 -1.82400554e-01 -3.09835255e-01 -5.03969133e-01 -2.77478546e-01 -5.31800687e-01 -1.24282074e+00 -5.88957489e-01 -2.51836121e-01 2.51528382e-01 9.93273139e-01 8.04426551e-01 2.17417330e-01 6.53530538e-01 3.44252765e-01 -7.40197659e-01 -1.02538514e+00 -9.95871305e-01 -7.26385593e-01 3.75422418e-01 2.65538782e-01 -4.21307564e-01 -8.04254532e-01 -4.32405502e-01]
[8.173552513122559, 3.3303709030151367]
d7932d0e-504e-4666-8860-f44148003c8b
emotion-distribution-learning-from-texts
null
null
https://aclanthology.org/D16-1061
https://aclanthology.org/D16-1061.pdf
Emotion Distribution Learning from Texts
null
['Xin Geng', 'Yin Zhou', 'Xuan Zhang', 'Quan Zhao', 'Deyu Zhou']
2016-11-01
null
null
null
emnlp-2016-11
['product-recommendation']
['miscellaneous']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.254841327667236, 3.8090314865112305]
b7f1c56a-db2f-4dea-b9f1-39ad3c2cbd7d
situation-recognition-visual-semantic-role
null
null
http://openaccess.thecvf.com/content_cvpr_2016/html/Yatskar_Situation_Recognition_Visual_CVPR_2016_paper.html
http://openaccess.thecvf.com/content_cvpr_2016/papers/Yatskar_Situation_Recognition_Visual_CVPR_2016_paper.pdf
Situation Recognition: Visual Semantic Role Labeling for Image Understanding
This paper introduces situation recognition, the problem of producing a concise summary of the situation an image depicts including: (1) the main activity (e.g., clipping), (2) the participating actors, objects, substances, and locations (e.g., man, shears, sheep, wool, and field) and most importantly (3) the roles these participants play in the activity (e.g., the man is clipping, the shears are his tool, the wool is being clipped from the sheep, and the clipping is in a field). We use FrameNet, a verb and role lexicon developed by linguists, to define a large space of possible situations and collect a large-scale dataset containing over 500 activities, 1,700 roles, 11,000 objects, 125,000 images, and 200,000 unique situations. We also introduce structured prediction baselines and show that, in activity-centric images, situation-driven prediction of objects and activities outperforms independent object and activity recognition.
['Luke Zettlemoyer', 'Ali Farhadi', 'Mark Yatskar']
2016-06-01
null
null
null
cvpr-2016-6
['grounded-situation-recognition', 'situation-recognition']
['computer-vision', 'computer-vision']
[ 6.02801263e-01 -9.83706303e-03 -2.22268566e-01 -2.23587275e-01 -2.96771795e-01 -7.82499731e-01 1.16434228e+00 2.28742674e-01 -2.97648996e-01 5.10882616e-01 1.02519739e+00 2.28762716e-01 -7.01304302e-02 -2.56467670e-01 -7.26272941e-01 -6.31146371e-01 -8.37909356e-02 4.54978764e-01 4.35140908e-01 3.52512486e-02 4.19537097e-01 4.15486872e-01 -1.76347983e+00 7.17164099e-01 -1.69440284e-02 1.03391910e+00 2.34844178e-01 3.95021081e-01 2.52572238e-01 1.54264498e+00 -9.45764959e-01 -5.18284380e-01 -1.98075950e-01 -4.89613771e-01 -7.73866653e-01 7.13650823e-01 4.67084348e-01 -4.91579950e-01 -2.04437867e-01 5.48678994e-01 1.00578070e-01 3.85044664e-01 6.33554220e-01 -1.44290781e+00 -4.21930939e-01 5.37996888e-01 -7.15397239e-01 2.26940066e-01 8.29804778e-01 2.17118979e-01 8.36166859e-01 -9.05184150e-01 1.24043226e+00 1.25170445e+00 7.65368491e-02 3.74843568e-01 -1.05763054e+00 -3.21259618e-01 4.03187633e-01 3.06234658e-01 -9.93063033e-01 -7.46522725e-01 7.84899771e-01 -6.10810161e-01 9.22141910e-01 3.14232051e-01 1.05315983e+00 1.63569605e+00 4.85101976e-02 9.17966306e-01 6.32270634e-01 -4.62546557e-01 3.29325050e-01 7.39189833e-02 1.80486720e-02 4.52494174e-01 7.68828392e-02 -4.48497027e-01 -9.33918715e-01 -4.37607020e-01 7.18440831e-01 1.52884379e-01 2.53894133e-03 -6.65158212e-01 -1.81499016e+00 2.34024495e-01 1.25294337e-02 2.20156282e-01 -5.27164996e-01 2.73449510e-01 3.46807033e-01 -2.25751385e-01 1.33822709e-01 3.95833939e-01 -2.17201382e-01 -4.23033953e-01 -6.10362887e-01 5.97993851e-01 7.54159153e-01 9.72589433e-01 4.51825291e-01 -2.53264457e-01 -4.00877893e-01 7.58788764e-01 -2.65273415e-02 2.84250498e-01 1.74557507e-01 -1.39667892e+00 6.72921896e-01 8.32042992e-01 6.79014981e-01 -9.16496813e-01 -2.95375168e-01 1.85353160e-01 -3.27886015e-01 -2.37460539e-01 5.91216505e-01 -1.46382209e-02 -7.73442507e-01 1.54243004e+00 2.77670056e-01 2.24825427e-01 -2.39412948e-01 8.02341163e-01 8.83046746e-01 6.56375647e-01 5.72109044e-01 -2.89251298e-01 1.92130208e+00 -1.18445444e+00 -7.19839752e-01 -5.96472144e-01 3.85560930e-01 -4.72601771e-01 8.16887140e-01 1.78962514e-01 -1.20507896e+00 -4.59896237e-01 -7.14988649e-01 -5.91052100e-02 -3.29889923e-01 8.63781497e-02 6.82541251e-01 6.95885867e-02 -5.70496142e-01 2.81410336e-01 -6.52929842e-01 -8.19737017e-01 3.53586197e-01 -1.37122750e-01 -8.86274934e-01 -4.42573018e-02 -7.79546738e-01 9.99482632e-01 3.32554668e-01 -1.28217101e-01 -1.16494381e+00 -3.54898363e-01 -1.00558603e+00 -6.19412176e-02 7.14306474e-01 -2.45989338e-01 1.05416083e+00 -1.09095240e+00 -7.97820687e-01 1.17409563e+00 -2.23799348e-01 -4.42868292e-01 3.39987755e-01 -1.53886959e-01 -6.60927415e-01 3.77303123e-01 5.50450385e-01 9.31220949e-01 7.68852472e-01 -1.22342336e+00 -9.78127182e-01 -3.43278050e-01 4.18292075e-01 5.87849736e-01 -4.64770012e-02 6.96920216e-01 -6.66971385e-01 -6.21019244e-01 -3.70068592e-03 -7.63227642e-01 1.03475556e-01 2.23516718e-01 -3.78161371e-01 -1.78349271e-01 6.70618355e-01 -6.68625951e-01 1.10240996e+00 -2.50713372e+00 2.91102409e-01 -8.11921805e-02 1.25566542e-01 -2.90416449e-01 -3.64750437e-03 6.79312646e-01 -2.43113652e-01 -4.24992479e-02 -4.70037498e-02 -2.07908526e-01 -9.60366521e-03 3.97299826e-01 -1.44911245e-01 4.46848094e-01 8.53575487e-03 8.26874197e-01 -1.04456866e+00 -7.44424164e-01 4.44971144e-01 3.58146131e-01 -5.12762666e-02 -4.29898389e-02 -1.35756016e-01 3.09199721e-01 -3.55925083e-01 7.53911912e-01 1.14615820e-01 -3.41017246e-01 5.00904560e-01 -3.51019442e-01 -1.27935871e-01 2.19038382e-01 -1.22922421e+00 1.44938767e+00 -5.46900905e-04 6.31764591e-01 1.10787514e-03 -4.49401647e-01 4.32066560e-01 3.24832767e-01 6.61917508e-01 -5.50311685e-01 -2.50485420e-01 -3.42897594e-01 -1.06402762e-01 -6.08676791e-01 5.45351028e-01 1.82491720e-01 -4.61768508e-01 7.31417239e-01 -1.16070986e-01 1.64026961e-01 6.52917683e-01 3.43966097e-01 1.32610106e+00 4.71458644e-01 7.01616466e-01 5.55943996e-02 2.52206802e-01 1.30386993e-01 5.36754251e-01 5.92423856e-01 -5.85825980e-01 3.32342654e-01 9.23854947e-01 -7.74291575e-01 -8.99007440e-01 -1.01484227e+00 2.73872703e-01 1.69232261e+00 4.92907673e-01 -4.12664324e-01 -4.97769386e-01 -5.09483159e-01 1.16062902e-01 7.24127829e-01 -8.87789607e-01 1.02670901e-02 -9.82741892e-01 -3.26241851e-02 4.10186887e-01 8.36481094e-01 4.71463799e-01 -1.58384883e+00 -1.03124309e+00 2.00788215e-01 -6.34313881e-01 -1.09003556e+00 -6.63463175e-01 2.18406215e-01 -5.07982075e-01 -1.26731133e+00 -4.20892090e-01 -7.93730557e-01 6.81630671e-01 2.35023960e-01 1.44743359e+00 -2.11718813e-01 -8.74353424e-02 3.94145817e-01 -3.79747897e-01 -4.22000378e-01 5.78088723e-02 -6.63092196e-01 -1.34524420e-01 4.00249869e-01 2.22810313e-01 -4.08089519e-01 -4.86548275e-01 6.30808592e-01 -4.24248695e-01 3.58486831e-01 2.80192703e-01 2.76628077e-01 5.45505583e-01 3.21664661e-02 1.60924606e-02 -7.18002319e-01 1.15113683e-01 -6.02992952e-01 4.29747328e-02 6.84716046e-01 3.90292346e-01 -3.29471499e-01 1.01941548e-01 -7.06369162e-01 -1.27758360e+00 4.39673126e-01 5.37746549e-01 -2.56295979e-01 -6.10039771e-01 7.36051649e-02 -3.49441648e-01 4.92401898e-01 9.12398100e-01 2.23833293e-01 -3.06860685e-01 -2.77150661e-01 4.95359540e-01 3.06539625e-01 7.13172138e-01 -6.90318406e-01 4.01475757e-01 9.21259105e-01 -2.53053904e-01 -6.83020234e-01 -6.52646482e-01 -5.43204844e-01 -7.35393226e-01 -6.88748240e-01 9.92432058e-01 -9.86128807e-01 -4.58543241e-01 4.97348189e-01 -1.21668518e+00 -5.03756881e-01 -4.03206766e-01 2.32903093e-01 -5.51010549e-01 6.72410429e-02 -4.69796240e-01 -9.72016394e-01 1.92009225e-01 -7.98540950e-01 1.26786780e+00 6.42867982e-02 -7.18101203e-01 -5.61580956e-01 -4.96288329e-01 7.10999191e-01 -5.91286905e-02 6.75351501e-01 7.57769346e-01 -4.76744562e-01 -4.12777334e-01 3.08525246e-02 -2.05526933e-01 -2.00326487e-01 3.29892367e-01 1.37895063e-01 -3.29256803e-01 1.64975807e-01 -1.53413519e-01 -3.79863799e-01 3.76319706e-01 2.54481345e-01 1.02681565e+00 -3.17289174e-01 -6.52506053e-01 8.51222649e-02 7.83860743e-01 7.12741017e-01 8.32493544e-01 7.99215734e-02 5.97120821e-01 1.00949609e+00 7.94750094e-01 4.29072201e-01 3.96018744e-01 7.33253300e-01 3.85351539e-01 3.00954524e-02 -1.17493503e-01 -3.09133679e-01 4.60231870e-01 -2.20644131e-01 -3.98876578e-01 -3.43616039e-01 -8.43146682e-01 7.79254735e-01 -1.92963827e+00 -1.50996530e+00 -3.36677879e-02 1.72140193e+00 4.48139608e-01 7.97971711e-02 3.96566361e-01 -6.86218217e-02 9.57190990e-01 6.51167154e-01 -6.75560236e-01 6.36493042e-02 -5.25668673e-02 -2.84542859e-01 1.30384952e-01 -8.90174806e-02 -1.34191918e+00 8.67742717e-01 6.58129072e+00 6.50767565e-01 -4.51043993e-01 1.49536699e-01 6.04655921e-01 -5.97450912e-01 5.64132333e-02 -2.38245409e-02 -6.63280368e-01 6.73744619e-01 3.83766860e-01 5.71750104e-02 4.29379016e-01 9.73010898e-01 2.26326674e-01 -5.50981998e-01 -1.44526267e+00 1.17061865e+00 6.41463161e-01 -1.13315511e+00 5.61938807e-02 -9.10109580e-02 4.61938560e-01 -4.71086711e-01 -3.37902367e-01 9.38704237e-03 2.45770290e-01 -7.81524777e-01 1.54754937e+00 4.59311426e-01 6.54598355e-01 -2.87669390e-01 1.18213020e-01 3.82635206e-01 -1.23525870e+00 -3.40339482e-01 2.66040653e-01 -4.46645468e-01 4.79332924e-01 1.38978943e-01 -3.26067716e-01 9.82672796e-02 8.10606003e-01 7.88211703e-01 -5.00518382e-01 4.08943295e-01 -5.64952374e-01 4.53932822e-01 -1.55262515e-01 -9.34113637e-02 5.32367043e-02 -1.74156819e-02 5.26696980e-01 9.86058831e-01 -1.64083332e-01 3.55761141e-01 1.01040155e-01 4.41280037e-01 -5.80829643e-02 -2.27810994e-01 -6.61546469e-01 -1.53416932e-01 5.67464769e-01 1.17841578e+00 -1.06157184e+00 -5.57934046e-01 -3.70946378e-01 8.53749454e-01 7.13227838e-02 3.91165048e-01 -1.01354706e+00 -7.96336755e-02 7.67400324e-01 7.64838219e-01 8.06455538e-02 -8.96417070e-03 1.03077911e-01 -1.00550735e+00 3.81165087e-01 -8.39356542e-01 4.83173400e-01 -1.41151559e+00 -8.96594167e-01 1.36309594e-01 4.57311094e-01 -1.08354878e+00 -1.31280795e-01 -3.54290098e-01 -5.18669724e-01 4.14941788e-01 -5.63149035e-01 -1.41763783e+00 -4.55148697e-01 6.58271015e-01 7.70460546e-01 -6.59590820e-03 3.71068239e-01 2.41302416e-01 -4.12328005e-01 -2.45510980e-01 -6.41669095e-01 3.85468006e-01 6.22282088e-01 -8.13028991e-01 4.01888430e-01 6.60151780e-01 3.84826809e-01 5.83754838e-01 4.56150323e-01 -8.48241985e-01 -1.09439540e+00 -9.76511002e-01 1.14090860e+00 -1.03061950e+00 6.82326317e-01 -7.77067959e-01 -4.79075104e-01 1.09621239e+00 6.28943667e-02 -8.47402140e-02 5.67573965e-01 -9.16441903e-02 -2.53761858e-01 3.65582779e-02 -1.11454964e+00 8.04597080e-01 1.67580485e+00 -3.17157835e-01 -9.05307233e-01 6.94114625e-01 3.24723005e-01 -5.81743300e-01 -3.66596788e-01 -1.81169972e-01 1.02765262e+00 -1.05952322e+00 1.10418689e+00 -8.43135059e-01 7.39299953e-01 -2.49313027e-01 -2.05045149e-01 -8.24311733e-01 -4.98747706e-01 -1.91439852e-01 -4.72978622e-01 1.14631152e+00 2.85176635e-01 -1.32456645e-01 6.66940033e-01 1.05463517e+00 -6.24998510e-02 -5.64258993e-01 -7.75198340e-01 -4.69389647e-01 -6.55706227e-01 -3.13848227e-01 7.08448410e-01 8.71505678e-01 3.01776361e-02 1.26903251e-01 -7.94268131e-01 -3.21508855e-01 3.34340721e-01 -3.79147939e-02 6.66972995e-01 -1.03795516e+00 -2.86340564e-01 -1.56313226e-01 -3.69095594e-01 -8.95665407e-01 -7.51263797e-02 -3.68358493e-01 -3.18920575e-02 -1.87926412e+00 5.07962108e-01 2.28955820e-02 -1.12204142e-01 1.08222437e+00 4.37206253e-02 3.10433488e-02 3.94071132e-01 3.75263304e-01 -1.18040693e+00 -1.29303530e-01 1.15773165e+00 -1.71877563e-01 -1.65449217e-01 -4.08901006e-01 -7.64026821e-01 1.08274972e+00 3.18450361e-01 -2.78533012e-01 -3.93971175e-01 -5.49985409e-01 2.27784514e-01 3.34874779e-01 5.80409288e-01 -9.36292052e-01 2.70480782e-01 -5.44043541e-01 6.89776540e-01 -4.49496597e-01 6.55331910e-01 -8.59993219e-01 6.68019831e-01 3.64029497e-01 -5.00268042e-01 -2.56303996e-01 -3.01509053e-01 6.68395162e-01 -1.82991512e-02 -5.48189059e-02 4.31580812e-01 -4.35545951e-01 -1.14089978e+00 -1.82182103e-01 -5.54817319e-01 -1.73613995e-01 1.50622296e+00 -5.95863342e-01 -4.82224613e-01 -4.32627290e-01 -9.73335564e-01 3.12669575e-01 5.48562169e-01 6.90206051e-01 2.21630007e-01 -1.24599731e+00 -4.06195283e-01 -1.25751802e-04 5.05491257e-01 -2.23010942e-01 3.30866724e-01 5.69747210e-01 -2.71242082e-01 -1.17305458e-01 -3.36772174e-01 -2.97165543e-01 -1.25623047e+00 3.65402162e-01 -7.06475675e-02 -1.68975234e-01 -5.25196731e-01 6.70316398e-01 4.95982736e-01 1.09896511e-01 3.10249835e-01 -3.07983726e-01 -4.90337044e-01 5.63465714e-01 6.22900188e-01 6.13940656e-01 -2.59228587e-01 -1.10244298e+00 -7.96035290e-01 4.11014140e-01 1.14628360e-01 -2.16551125e-01 1.06053174e+00 1.18156850e-01 -1.83389544e-01 5.24160981e-01 6.19204283e-01 -1.88250080e-01 -1.43209910e+00 -4.43411022e-02 -7.43275657e-02 -5.52244246e-01 -6.43744409e-01 -8.76900852e-01 -7.54714310e-01 3.93924057e-01 6.09527789e-02 2.31927902e-01 9.49410141e-01 5.11560619e-01 4.05900419e-01 3.17433327e-01 8.13508213e-01 -1.25603235e+00 4.90368396e-01 1.84059069e-01 1.30743492e+00 -8.24748993e-01 1.62050128e-01 -3.89667541e-01 -1.21570766e+00 6.20043695e-01 6.10583544e-01 1.65176675e-01 3.31544191e-01 1.69207707e-01 -2.67950207e-01 -4.68729436e-01 -9.41358566e-01 -8.79961327e-02 -1.80987939e-01 7.07916677e-01 1.76576674e-01 2.45966345e-01 -1.33204043e-01 5.98106205e-01 1.08496226e-01 -1.89842388e-01 3.59045535e-01 1.15358937e+00 -3.01407218e-01 -6.09272659e-01 -3.62020493e-01 8.54201436e-01 -3.61384153e-01 2.94785768e-01 -6.56927824e-01 7.86421299e-01 7.70392120e-01 1.13352382e+00 4.71087247e-01 -1.80435348e-02 7.89256573e-01 1.36967108e-01 4.43933398e-01 -8.53605390e-01 -7.02283502e-01 -3.30898538e-02 5.79474688e-01 -7.74928153e-01 -8.97263587e-01 -1.12450838e+00 -1.13476944e+00 1.31783515e-01 2.79836476e-01 -1.47272527e-01 4.97477472e-01 9.01304722e-01 3.60539794e-01 7.83535093e-02 1.54840991e-01 -9.14014637e-01 2.13885054e-01 -9.91686463e-01 -7.88859844e-01 1.01218379e+00 -1.01499967e-01 -9.06147540e-01 -5.27392209e-01 7.22024679e-01]
[8.249226570129395, 0.6669843196868896]
b5e03f68-fa12-47f6-b129-bb0d837d9e2b
dudornext-a-hybrid-model-for-dual-domain
2303.10611
null
https://arxiv.org/abs/2303.10611v1
https://arxiv.org/pdf/2303.10611v1.pdf
DuDoRNeXt: A hybrid model for dual-domain undersampled MRI reconstruction
Undersampled MRI reconstruction is crucial for accelerating clinical scanning procedures. Recent deep learning methods for MRI reconstruction adopt CNN or ViT as backbone, which lack in utilizing the complementary properties of CNN and ViT. In this paper, we propose DuDoRNeXt, whose backbone hybridizes CNN and ViT in an domain-specific, intra-stage way. Besides our hybrid vertical layout design, we introduce domain-specific modules for dual-domain reconstruction, namely image-domain parallel local detail enhancement and k-space global initialization. We evaluate different conventions of MRI reconstruction including image-domain, k-space-domain, and dual-domain reconstruction with a reference protocol on the IXI dataset and an in-house multi-contrast dataset. DuDoRNeXt achieves significant improvements over competing deep learning methods.
['S. Kevin Zhou', 'Ziqi Gao']
2023-03-19
null
null
null
null
['layout-design', 'mri-reconstruction']
['computer-vision', 'computer-vision']
[ 5.69995120e-02 8.86670500e-02 -6.62828833e-02 -4.62088317e-01 -1.13113058e+00 -2.41610199e-01 3.76543313e-01 3.73556130e-02 -5.34423053e-01 6.43839180e-01 5.46986938e-01 -4.37489420e-01 -1.79643512e-01 -5.40364623e-01 -5.08440733e-01 -6.97760582e-01 -2.69882590e-01 3.93690497e-01 4.39417899e-01 -2.02946663e-01 -9.86268073e-02 7.38586247e-01 -5.14369011e-01 7.49524176e-01 2.86675364e-01 8.24916959e-01 3.85066122e-01 7.48429060e-01 7.89652467e-02 9.52193916e-01 -1.41865924e-01 -3.35290320e-02 1.91868171e-01 -4.18387115e-01 -1.19748199e+00 -1.84933767e-01 3.75824511e-01 -6.91760004e-01 -6.92745209e-01 5.23316324e-01 1.18343306e+00 -1.54840991e-01 5.34419060e-01 -5.68593383e-01 -1.96855918e-01 6.61752760e-01 -6.84217513e-01 9.41446304e-01 -2.66626835e-01 3.65833789e-01 7.10943416e-02 -5.64726889e-01 8.06606352e-01 4.48193014e-01 1.19106400e+00 6.48172438e-01 -1.38924146e+00 -5.92931628e-01 -3.57598901e-01 7.95356110e-02 -1.12679970e+00 -2.21899047e-01 9.09099698e-01 -3.96499097e-01 1.20154488e+00 1.24900294e-02 5.81262469e-01 1.06366837e+00 7.17042267e-01 7.32864380e-01 1.46482098e+00 -2.16312915e-01 -1.04877390e-02 -4.68620986e-01 -4.00696807e-02 5.19315481e-01 -6.84085339e-02 2.21451849e-01 -1.23181067e-01 -7.70938909e-03 1.39756227e+00 -1.01074003e-01 -3.93566906e-01 -5.76392889e-01 -1.48322284e+00 7.70173252e-01 7.60703921e-01 5.81112981e-01 -5.76704264e-01 3.95420015e-01 8.21398795e-01 1.52080029e-01 4.71549511e-01 2.59651065e-01 -4.56081718e-01 7.63334185e-02 -1.37841940e+00 -2.12702304e-02 2.50409961e-01 7.26000667e-01 2.83382237e-01 2.41286263e-01 -2.85620391e-01 7.37833679e-01 -1.64087504e-01 1.87644269e-03 9.86938655e-01 -7.75897384e-01 4.45576273e-02 5.90093294e-03 -3.90823871e-01 -5.43989003e-01 -1.00502694e+00 -8.95978510e-01 -1.16926122e+00 7.73499310e-02 1.86598450e-01 -1.89744741e-01 -1.20431364e+00 1.43860269e+00 4.22513634e-01 4.55436558e-01 -1.44945368e-01 1.08425772e+00 1.29738569e+00 2.27543384e-01 1.81031436e-01 -1.53496861e-01 1.36512518e+00 -1.24769914e+00 -6.55780375e-01 5.24677709e-02 7.48364687e-01 -6.45009696e-01 8.38139832e-01 4.85240966e-01 -1.57186115e+00 -4.49194729e-01 -1.00698411e+00 -1.64994806e-01 6.05012141e-02 -1.62758573e-03 6.15746081e-01 6.36779785e-01 -1.51564860e+00 6.47394836e-01 -1.09022593e+00 1.62627563e-01 5.24354994e-01 5.84617078e-01 -6.22976243e-01 -3.87417339e-02 -1.05184555e+00 1.06611097e+00 3.31850410e-01 6.92322617e-03 -1.40897250e+00 -1.21398437e+00 -7.90460646e-01 -3.58723342e-01 1.92125961e-01 -7.78178692e-01 1.55129468e+00 -5.50832927e-01 -1.37020636e+00 1.11537313e+00 2.81918824e-01 -6.72172189e-01 7.94181287e-01 1.61077991e-01 -4.76518899e-01 6.30945206e-01 -7.10188150e-02 7.79604733e-01 9.34852719e-01 -1.09369922e+00 1.06101610e-01 -2.79053330e-01 -1.49734959e-01 1.07605502e-01 2.43104413e-01 -1.23687452e-02 -4.97007817e-01 -9.02529955e-01 2.49759138e-01 -6.76643252e-01 -6.87719941e-01 9.49973837e-02 -1.68026432e-01 5.53576350e-01 5.09282708e-01 -9.93197680e-01 1.05992615e+00 -1.91363871e+00 -1.54218048e-01 3.15381698e-02 8.46602738e-01 6.80313110e-02 -1.39054537e-01 -1.38608381e-01 -7.61353970e-01 -2.78413445e-01 -4.67555821e-01 -2.14542508e-01 -5.67396104e-01 3.06774884e-01 2.07402512e-01 8.37059438e-01 -9.29886941e-03 1.19458330e+00 -1.02068853e+00 -6.51857734e-01 5.31076908e-01 7.29557991e-01 -8.25591087e-01 -5.02494024e-03 4.28549647e-01 9.21098173e-01 -1.94928169e-01 3.21757138e-01 1.03350401e+00 -3.35505664e-01 5.51204026e-01 -9.02508676e-01 -2.35695183e-01 2.71584421e-01 -7.32684076e-01 2.39384151e+00 -7.87291229e-01 4.30838466e-01 4.49445486e-01 -1.22965252e+00 3.43857020e-01 5.54141462e-01 9.41940606e-01 -1.18401337e+00 2.75743425e-01 3.41043144e-01 -5.29932566e-02 -4.47829336e-01 4.44479287e-01 -6.33901119e-01 2.09040865e-01 4.47853476e-01 3.79779726e-01 -3.05018932e-01 -3.16512614e-01 1.88428089e-01 1.13258600e+00 9.50110704e-02 1.33393899e-01 -6.00297749e-01 2.64345586e-01 9.00834575e-02 1.12667792e-01 9.50013876e-01 -4.31307375e-01 1.33203518e+00 4.33591038e-01 -6.52732015e-01 -1.38044238e+00 -1.16304731e+00 -5.16209304e-01 5.11467576e-01 -1.50525033e-01 -1.67644680e-01 -8.11157525e-01 -7.39865661e-01 -5.00169456e-01 2.10554376e-01 -9.34053063e-01 1.24393709e-01 -1.14518499e+00 -9.11245346e-01 6.34536624e-01 7.00921118e-01 3.38628858e-01 -1.03387356e+00 -9.11271989e-01 5.95525265e-01 -3.40254873e-01 -1.14411974e+00 -2.63721853e-01 7.26577044e-01 -1.28185701e+00 -8.87116551e-01 -1.23407841e+00 -7.83452868e-01 4.64013606e-01 1.85645834e-01 1.54471707e+00 5.39527498e-02 -5.15876412e-01 1.09903120e-01 -1.61530763e-01 1.27912745e-01 -1.86672106e-01 3.08943272e-01 -2.93084949e-01 -6.93210363e-01 -5.02177954e-01 -7.47376084e-01 -1.18879318e+00 9.08008032e-03 -1.15209520e+00 3.40305120e-01 5.78723252e-01 1.05683172e+00 6.81307614e-01 -2.99719512e-01 1.94982767e-01 -9.03642297e-01 4.12738591e-01 -4.87416059e-01 -7.30459243e-02 8.84156227e-02 -2.60235101e-01 6.73220158e-02 3.49700361e-01 -1.93820849e-01 -7.76860833e-01 -3.66173163e-02 -8.38215649e-01 -5.03928483e-01 -4.18816656e-02 3.38070333e-01 4.02933270e-01 -5.07461607e-01 6.59483790e-01 3.90631586e-01 2.07869560e-01 -4.23879534e-01 2.19937846e-01 4.96489331e-02 7.18032181e-01 -5.23905277e-01 2.26063117e-01 7.61300981e-01 -1.08890153e-01 -5.02902210e-01 -4.69198585e-01 -2.76278615e-01 -9.21088755e-01 -3.03860039e-01 1.17493701e+00 -9.36663389e-01 -4.20316458e-01 5.31359315e-01 -1.07509243e+00 -8.13593030e-01 -5.00064313e-01 6.29556060e-01 -6.29212201e-01 3.95012498e-01 -9.93430555e-01 2.18768463e-01 -6.28307581e-01 -1.92377794e+00 1.11431968e+00 -1.48576319e-01 9.61156115e-02 -1.11668396e+00 2.57074356e-01 2.05774397e-01 9.19935882e-01 5.24696469e-01 8.56378257e-01 -2.87672430e-01 -2.83297777e-01 3.76368731e-01 -4.53113735e-01 3.24732363e-01 -1.95801213e-01 -8.18577170e-01 -8.62211108e-01 -5.21269679e-01 2.49209970e-01 -4.28950816e-01 6.44667685e-01 1.07395983e+00 1.53679216e+00 2.65837759e-01 1.96937323e-02 1.30126750e+00 1.67633891e+00 2.65792222e-03 9.13713217e-01 6.04038239e-01 5.71001649e-01 8.55008960e-02 -1.38843000e-01 3.58079106e-01 2.96013087e-01 7.22523272e-01 3.97174954e-01 -8.62013996e-01 -5.79752624e-01 1.90181777e-01 -1.94093928e-01 1.02787483e+00 1.88460976e-01 2.64670402e-01 -1.16405046e+00 7.36597657e-01 -1.22314095e+00 -5.76092660e-01 -1.99626520e-01 1.54308593e+00 9.53603923e-01 -1.34463146e-01 1.11535028e-01 -1.26436323e-01 2.63553947e-01 1.53180972e-01 -4.56471771e-01 -2.95796663e-01 -2.66834702e-02 8.95757794e-01 8.70578468e-01 4.10036355e-01 -1.06806898e+00 5.94808578e-01 7.39917946e+00 1.00644219e+00 -1.59545469e+00 1.04410863e+00 9.19963479e-01 -2.28000253e-01 -2.76523054e-01 -3.91194522e-01 -2.00055450e-01 7.88897499e-02 8.02617788e-01 2.71769226e-01 1.16389886e-01 5.52098811e-01 6.41813176e-03 -4.80019301e-02 -7.32585847e-01 1.08279777e+00 -7.53404647e-02 -1.87379169e+00 -2.48336226e-01 -9.44576114e-02 6.42733455e-01 6.26435697e-01 1.01568863e-01 2.45080158e-01 5.71984760e-02 -1.15303314e+00 6.28047049e-01 6.59213364e-02 1.37077677e+00 -7.39219308e-01 6.15496516e-01 -8.52157772e-02 -9.73501623e-01 4.86390233e-01 -5.45850322e-02 4.32291955e-01 2.79147446e-01 5.26195765e-01 -7.34113216e-01 1.04417813e+00 8.43343079e-01 4.92096663e-01 -4.55590308e-01 8.74483228e-01 1.95269734e-01 5.28572500e-01 -6.21358901e-02 9.20485020e-01 7.50907958e-01 1.36500403e-01 1.99429706e-01 1.63299537e+00 -3.01469564e-02 3.62772405e-01 -4.33027260e-02 5.22397161e-01 1.39159605e-01 -1.26056328e-01 -3.25169355e-01 6.99059725e-01 -2.64275432e-01 1.34057343e+00 -1.11744308e+00 -5.40654182e-01 -3.26566696e-01 9.90609646e-01 1.35840371e-01 1.04557984e-01 -9.58535969e-01 5.06103709e-02 3.32555264e-01 4.59732383e-01 2.91823596e-01 -4.22485203e-01 -5.58236718e-01 -1.13040376e+00 -3.54402184e-01 -8.79057586e-01 2.85573930e-01 -8.70929182e-01 -9.93386030e-01 9.87987518e-01 2.03150034e-01 -1.12521911e+00 -1.80118233e-02 -5.39348900e-01 -4.70780283e-01 8.70971262e-01 -1.87189519e+00 -1.09511065e+00 -4.05130655e-01 9.15003181e-01 3.47776383e-01 1.42192483e-01 6.09139025e-01 8.29443276e-01 -1.08522028e-01 6.05020106e-01 1.99739248e-01 3.34343106e-01 5.84761679e-01 -1.11232996e+00 4.28118557e-01 5.32795012e-01 -5.06062686e-01 5.03799438e-01 3.23288888e-01 -5.84966302e-01 -1.21740639e+00 -9.75442588e-01 2.08354592e-01 -1.60924584e-01 6.18125021e-01 -8.44595730e-02 -7.42966115e-01 6.45035446e-01 3.81510705e-01 5.87581873e-01 6.16008699e-01 -3.32032353e-01 7.70135894e-02 3.00566871e-02 -1.60128593e+00 3.51260632e-01 8.30104768e-01 -6.24284089e-01 -4.77094710e-01 4.04529750e-01 7.58612692e-01 -1.10709989e+00 -1.22921014e+00 5.71646035e-01 5.57884574e-01 -1.08002484e+00 1.41067231e+00 -3.58669877e-01 7.64412344e-01 -1.26260594e-01 8.30947980e-02 -1.24144959e+00 -5.65983772e-01 -2.55755514e-01 2.50803530e-01 3.29229504e-01 7.92132095e-02 -3.57106775e-01 9.61269915e-01 1.06143780e-01 -7.13634670e-01 -9.13008392e-01 -1.15066648e+00 -3.89853656e-01 5.17093360e-01 -7.15632319e-01 4.50061053e-01 1.17460847e+00 -4.95271772e-01 -1.70246735e-01 -3.84040236e-01 -3.28317434e-02 7.24514544e-01 -2.48741880e-01 2.62153924e-01 -3.77257079e-01 -5.01060426e-01 -4.28054124e-01 -2.23103464e-01 -8.96864951e-01 -2.14198217e-01 -1.10648584e+00 -2.85059065e-01 -1.55081248e+00 4.19050872e-01 -6.40144825e-01 -6.44173086e-01 4.76388246e-01 3.38115960e-01 8.53451312e-01 9.03316960e-03 7.12221488e-02 -3.99131238e-01 2.70922750e-01 1.77613783e+00 -1.35396838e-01 1.15036361e-01 -5.34651518e-01 -4.32292581e-01 4.29600030e-01 7.05345452e-01 -4.73825097e-01 -3.30089599e-01 -9.31988358e-01 -2.74149179e-01 5.91104805e-01 6.42160833e-01 -1.17949212e+00 7.19395727e-02 3.25808167e-01 6.57955408e-01 -7.44954586e-01 2.66434193e-01 -6.75927341e-01 2.73785770e-01 8.21384907e-01 -3.00441027e-01 4.42191064e-01 5.22157550e-01 -9.60556492e-02 -1.98352054e-01 -1.00070864e-01 1.25070012e+00 -5.96906185e-01 -5.82026422e-01 5.21193147e-01 -3.33415657e-01 1.63852885e-01 6.68757856e-01 -2.85641015e-01 1.53508455e-01 -1.72025207e-02 -1.26675594e+00 -1.44108817e-01 2.85967112e-01 2.77070031e-02 9.44696248e-01 -1.11427808e+00 -7.58213401e-01 2.23617896e-01 -2.53633887e-01 1.09085161e-03 9.88833189e-01 1.54136634e+00 -1.13221049e+00 3.79934579e-01 -7.11961508e-01 -9.49668348e-01 -7.95252383e-01 3.98657888e-01 9.17617321e-01 -8.91789556e-01 -1.15888214e+00 9.17777956e-01 5.09772360e-01 -6.78358257e-01 -1.52011439e-01 -4.73206341e-01 8.82491916e-02 -3.22578400e-01 6.12424433e-01 -1.93580776e-01 7.65272379e-01 -4.23151284e-01 -5.76615572e-01 3.88318181e-01 -4.04326826e-01 -2.41430297e-01 1.53669560e+00 4.78684381e-02 2.88879499e-02 -8.80769864e-02 1.40116060e+00 -4.69468802e-01 -1.19818568e+00 -2.75701433e-01 -3.05950284e-01 -1.29895672e-01 7.63014317e-01 -9.76586580e-01 -1.62357795e+00 1.02198279e+00 1.12925541e+00 -5.57385206e-01 1.30509591e+00 -1.41412184e-01 1.12969756e+00 -4.63807523e-01 4.86105472e-01 -6.25173032e-01 6.54612184e-02 4.75966722e-01 7.76988506e-01 -1.10563207e+00 1.69612214e-01 -6.66298494e-02 -7.20938623e-01 1.11302078e+00 4.71179247e-01 -3.01706463e-01 8.56326520e-01 8.70189428e-01 2.78538838e-02 -5.52934349e-01 -4.24377203e-01 1.51348591e-01 1.61184743e-01 7.03092456e-01 8.58437896e-01 9.88909006e-02 -3.49051028e-01 2.42755473e-01 6.78470358e-02 3.13756764e-01 3.62569690e-01 1.09321153e+00 1.23756953e-01 -1.01756883e+00 -2.55234629e-01 3.44211280e-01 -7.17012703e-01 -4.20124322e-01 4.52762336e-01 9.03836787e-01 2.06352964e-01 1.10242285e-01 -5.89309074e-02 -1.54536709e-01 2.55066544e-01 -4.60448742e-01 9.78112996e-01 -3.23401958e-01 -1.36596715e+00 2.24327669e-01 -1.47459000e-01 -7.30673432e-01 -3.53299439e-01 -3.78588915e-01 -1.30164659e+00 -2.92514205e-01 1.80132434e-01 -5.94529323e-02 8.36440086e-01 8.53367329e-01 1.71423674e-01 9.93206024e-01 3.36150795e-01 -1.16026366e+00 -1.75201520e-01 -7.03662276e-01 -4.67926621e-01 2.04113573e-01 3.86136830e-01 -5.37684619e-01 2.75637686e-01 -3.12787503e-01]
[13.99292278289795, -2.5054984092712402]
fd8e6417-71d3-4e7c-b5e7-b64f9b790d66
one-shot-scene-graph-generation
2202.10824
null
https://arxiv.org/abs/2202.10824v2
https://arxiv.org/pdf/2202.10824v2.pdf
One-shot Scene Graph Generation
As a structured representation of the image content, the visual scene graph (visual relationship) acts as a bridge between computer vision and natural language processing. Existing models on the scene graph generation task notoriously require tens or hundreds of labeled samples. By contrast, human beings can learn visual relationships from a few or even one example. Inspired by this, we design a task named One-Shot Scene Graph Generation, where each relationship triplet (e.g., "dog-has-head") comes from only one labeled example. The key insight is that rather than learning from scratch, one can utilize rich prior knowledge. In this paper, we propose Multiple Structured Knowledge (Relational Knowledge and Commonsense Knowledge) for the one-shot scene graph generation task. Specifically, the Relational Knowledge represents the prior knowledge of relationships between entities extracted from the visual content, e.g., the visual relationships "standing in", "sitting in", and "lying in" may exist between "dog" and "yard", while the Commonsense Knowledge encodes "sense-making" knowledge like "dog can guard yard". By organizing these two kinds of knowledge in a graph structure, Graph Convolution Networks (GCNs) are used to extract knowledge-embedded semantic features of the entities. Besides, instead of extracting isolated visual features from each entity generated by Faster R-CNN, we utilize an Instance Relation Transformer encoder to fully explore their context information. Based on a constructed one-shot dataset, the experimental results show that our method significantly outperforms existing state-of-the-art methods by a large margin. Ablation studies also verify the effectiveness of the Instance Relation Transformer encoder and the Multiple Structured Knowledge.
['Heng Tao Shen', 'Lianli Gao', 'Jingkuan Song', 'Yuyu Guo']
2022-02-22
null
null
null
null
['scene-graph-generation']
['computer-vision']
[ 1.54842883e-01 3.66942376e-01 -1.33435667e-01 -5.51668406e-01 -1.72189802e-01 -3.39441121e-01 6.25498116e-01 2.14394823e-01 -8.13430026e-02 5.46276033e-01 2.78886139e-01 -1.44239590e-01 7.17236400e-02 -1.18056071e+00 -1.04200304e+00 -4.65677530e-01 1.89190269e-01 1.11281544e-01 1.12820901e-01 -3.16254199e-01 -6.30343258e-02 1.19124599e-01 -1.53161955e+00 5.41538239e-01 6.65800631e-01 9.72266495e-01 4.79641706e-01 1.28325716e-01 -5.61390281e-01 1.46018469e+00 -6.41483009e-01 -6.33415461e-01 -1.24218747e-01 -5.89991629e-01 -9.55777705e-01 4.77149695e-01 3.95464301e-01 -3.13745648e-01 -6.58220708e-01 1.28093624e+00 1.67694092e-02 4.27184850e-01 4.06047255e-01 -1.43476033e+00 -1.52614164e+00 8.21294606e-01 -5.33618331e-01 8.74626637e-02 6.79220438e-01 3.49233925e-01 1.19583821e+00 -7.80990422e-01 8.83269370e-01 1.30583370e+00 2.23452881e-01 3.10687184e-01 -1.03587830e+00 -5.14302492e-01 4.65021551e-01 6.13844097e-01 -1.54532981e+00 -1.41931042e-01 1.09403729e+00 -4.45443392e-01 9.59162116e-01 -8.28049108e-02 9.60371077e-01 1.07939601e+00 -1.78126737e-01 9.71837282e-01 9.42242622e-01 -4.09108847e-01 6.99335039e-02 2.37288311e-01 1.22642606e-01 1.03783739e+00 3.45110923e-01 -2.46519465e-02 -4.85021770e-01 2.81749755e-01 7.73624718e-01 2.91187674e-01 -3.67852628e-01 -5.05384505e-01 -1.30665672e+00 7.89295673e-01 1.11462343e+00 2.57007211e-01 -2.37089902e-01 3.14339697e-01 3.39311093e-01 2.69262996e-02 1.45335525e-01 2.54982352e-01 -1.05913110e-01 3.93535227e-01 -4.21568125e-01 -6.38271719e-02 6.50044680e-01 1.52050769e+00 1.31084514e+00 6.37254566e-02 -2.81174093e-01 6.85198665e-01 2.65679777e-01 4.42078441e-01 2.95242786e-01 -5.67051113e-01 5.54774284e-01 1.05933976e+00 -1.10105701e-01 -1.26445353e+00 -1.75979584e-01 -1.57021746e-01 -1.02407146e+00 -2.64629066e-01 3.55070084e-02 -1.08024850e-02 -1.23983335e+00 1.70030487e+00 2.90502012e-01 3.17943305e-01 2.01457396e-01 9.16562974e-01 1.53793323e+00 7.01588333e-01 2.05214024e-01 -4.39006947e-02 1.70550692e+00 -9.88472760e-01 -7.50838459e-01 -6.01109147e-01 2.99260467e-01 -2.46829078e-01 1.35472703e+00 -1.86579347e-01 -6.22702062e-01 -6.33186758e-01 -9.00114477e-01 -4.72299039e-01 -8.72842610e-01 5.56418905e-03 8.42619777e-01 3.80918160e-02 -7.93381095e-01 2.65307605e-01 -3.94898266e-01 -5.46018362e-01 6.74133897e-01 -1.58587545e-01 -5.83418906e-01 -4.62535113e-01 -1.35739744e+00 7.67243683e-01 7.60441661e-01 1.72382429e-01 -8.91399086e-01 -4.32321012e-01 -1.53801632e+00 2.33275071e-01 8.06216598e-01 -1.05861723e+00 8.39612365e-01 -1.04869628e+00 -1.08124888e+00 1.20340133e+00 -2.68255472e-01 -3.65305841e-01 4.37343121e-02 -1.10069267e-01 -4.84354019e-01 3.65586609e-01 4.77437884e-01 5.50141096e-01 7.27957189e-01 -1.60934329e+00 -2.95784116e-01 -3.30487460e-01 5.86772382e-01 1.70702577e-01 2.68180054e-02 6.20154776e-02 -5.44267714e-01 -6.03273392e-01 6.29988462e-02 -5.13003051e-01 -1.36624664e-01 -9.11419466e-02 -8.78185272e-01 -9.63054448e-02 7.47587919e-01 -5.43541789e-01 1.01824760e+00 -2.34134388e+00 -1.33181617e-01 1.26478449e-01 4.26246524e-01 -1.00816913e-01 -2.70579066e-02 4.03564006e-01 -2.63055623e-01 9.84267518e-02 -1.65453807e-01 1.00297451e-01 -2.61560064e-02 5.26015699e-01 -4.11375970e-01 3.12005058e-02 3.26628983e-01 1.32890320e+00 -1.33609724e+00 -5.87676048e-01 2.35021040e-01 3.62164468e-01 -2.37743646e-01 4.13848490e-01 -3.76591206e-01 1.98871017e-01 -4.69803959e-01 5.80037355e-01 4.53326792e-01 -9.42355514e-01 1.93362236e-01 -6.64239228e-01 3.11836153e-01 1.42399639e-01 -9.52257752e-01 1.83623004e+00 -3.57817709e-01 5.62567174e-01 -5.71762204e-01 -1.06258464e+00 9.69036996e-01 1.45554394e-01 -1.79066719e-03 -7.39895642e-01 2.37437472e-01 -4.64686573e-01 -1.64928168e-01 -8.24316621e-01 3.38850319e-01 -4.13607955e-01 -2.03919262e-01 1.13250934e-01 3.38296831e-01 -1.44446909e-01 2.70313174e-01 6.63909674e-01 9.16268229e-01 1.47546828e-01 7.54223108e-01 1.31489351e-01 2.65136987e-01 1.71229109e-01 6.06086195e-01 5.42439163e-01 -1.08880483e-01 3.50834101e-01 6.96998656e-01 -4.30730283e-01 -6.62996173e-01 -1.35060072e+00 4.74806637e-01 9.13545907e-01 7.57181525e-01 -6.47816002e-01 -3.50785762e-01 -5.83878160e-01 -1.16766199e-01 1.03042567e+00 -8.02305639e-01 -3.81735027e-01 -3.52829009e-01 -2.23288044e-01 2.54094005e-01 6.99310601e-01 8.99936438e-01 -1.35294127e+00 -4.92913127e-01 -3.49567495e-02 -2.64239341e-01 -1.55641019e+00 -2.41712138e-01 -7.42650032e-02 -4.01035428e-01 -1.25695086e+00 -2.03066781e-01 -9.81720686e-01 9.20392931e-01 5.53682446e-01 1.51815104e+00 1.03158340e-01 -4.48153526e-01 5.88063717e-01 -6.21943533e-01 -3.15558195e-01 -3.31769139e-02 -5.96784353e-01 -3.84696841e-01 1.96691051e-01 5.48990667e-01 -7.37151325e-01 -5.39150953e-01 -1.35965496e-01 -8.69339406e-01 4.61388737e-01 5.76804459e-01 7.64701068e-01 7.80038595e-01 2.55464166e-01 4.26802963e-01 -1.18082786e+00 4.61740911e-01 -7.60440946e-01 -2.56147295e-01 4.94963348e-01 -1.64013356e-01 1.51736572e-01 8.80854547e-01 -3.71813148e-01 -1.23425543e+00 9.15485770e-02 4.74970043e-01 -8.55415821e-01 -3.88965726e-01 6.60740793e-01 -6.03820860e-01 4.54180092e-01 5.60948133e-01 4.64305401e-01 -2.86639124e-01 -9.13041234e-02 1.01865172e+00 1.10953860e-01 6.66773200e-01 -5.85122705e-01 8.35365176e-01 7.09401429e-01 -1.50025889e-01 -7.27918148e-01 -1.31009853e+00 -4.05577868e-01 -5.64539254e-01 -7.96299577e-02 1.21376550e+00 -1.03340209e+00 -6.06078684e-01 1.56752855e-01 -1.25404394e+00 -1.70396298e-01 -4.28479195e-01 1.60945699e-01 -5.65506041e-01 3.66627485e-01 -4.99179095e-01 -4.82689738e-01 -6.41474351e-02 -7.23225534e-01 1.03938305e+00 5.33510327e-01 1.38574898e-01 -9.61646914e-01 -4.49856758e-01 3.95969182e-01 -4.71482202e-02 5.51922202e-01 1.16326082e+00 -4.68993306e-01 -8.40004802e-01 3.12469862e-02 -6.92561090e-01 2.86214203e-01 3.79362762e-01 -5.64695150e-02 -7.26509690e-01 7.44511485e-02 -2.19384044e-01 -4.00596797e-01 6.94765210e-01 5.69441617e-02 1.27997625e+00 -4.94917393e-01 -3.60009581e-01 4.85915512e-01 1.63479698e+00 1.86005592e-01 7.37568796e-01 -1.79304760e-02 1.21902311e+00 5.56816936e-01 4.34499323e-01 3.71525586e-01 9.38819051e-01 2.70897895e-01 4.02225822e-01 -6.57838061e-02 -3.88611466e-01 -8.26603234e-01 1.76785991e-01 4.93888646e-01 -9.19433609e-02 -6.54794872e-02 -9.07723844e-01 6.89586580e-01 -1.92653334e+00 -1.29391003e+00 4.05901633e-02 1.85589671e+00 7.39177942e-01 7.66520482e-03 -1.54344335e-01 -3.33364576e-01 8.86022925e-01 3.10837835e-01 -5.14100850e-01 -9.12029948e-03 -2.30798796e-01 2.06722692e-02 6.16680607e-02 2.50481039e-01 -9.22430754e-01 1.44034505e+00 5.01207066e+00 5.94375253e-01 -7.90623903e-01 -6.80253580e-02 3.43352169e-01 2.84480810e-01 -4.33352381e-01 4.77046639e-01 -4.85128492e-01 3.50005835e-01 1.76408753e-01 -4.47402060e-01 5.05864859e-01 1.00905609e+00 -9.59247351e-02 -1.46693632e-01 -1.21403670e+00 1.27187777e+00 3.87479633e-01 -1.41059721e+00 6.15922093e-01 -2.27453440e-01 7.01820791e-01 -3.45636249e-01 -3.01248819e-01 5.50217628e-01 6.80297256e-01 -1.14547956e+00 6.74131393e-01 6.52869165e-01 8.14582527e-01 -4.57867175e-01 3.88218224e-01 3.50539595e-01 -1.70699430e+00 2.92339567e-02 -5.13395011e-01 -9.80525315e-02 2.88964957e-01 7.08727658e-01 -6.67491436e-01 8.33862662e-01 6.65952623e-01 1.02818990e+00 -6.40081704e-01 6.32995844e-01 -9.67985988e-01 2.02349305e-01 6.33709356e-02 -7.50672519e-02 1.88698187e-01 -1.24812670e-01 2.42985040e-01 1.08553278e+00 2.45842915e-02 4.94685113e-01 3.57831150e-01 1.46566153e+00 -3.29997629e-01 -1.23751678e-01 -1.14568937e+00 -1.40062034e-01 4.64031905e-01 1.26020467e+00 -6.57043695e-01 -7.15731680e-01 -5.95814049e-01 1.01796687e+00 6.71519697e-01 7.18724012e-01 -9.12047267e-01 -6.13796473e-01 3.59283894e-01 3.65183912e-02 4.33590889e-01 -8.40937346e-03 7.64380321e-02 -1.43011391e+00 1.21765383e-01 -5.15075624e-01 4.84069377e-01 -1.42484736e+00 -1.56838226e+00 5.12689054e-01 2.37488821e-01 -1.20926571e+00 -1.14645816e-01 -5.93411207e-01 -8.08660686e-01 6.34579062e-01 -1.42988312e+00 -1.56799960e+00 -7.36056089e-01 9.70349252e-01 4.45122123e-01 6.58266470e-02 6.79395139e-01 -8.28382969e-02 -2.95407027e-01 3.31208557e-01 -6.33990288e-01 6.09806538e-01 1.93575099e-01 -1.23558486e+00 3.07764113e-01 9.13531184e-01 5.32153845e-01 8.62228453e-01 4.24249440e-01 -7.21227288e-01 -1.32542980e+00 -1.19019997e+00 6.29416943e-01 -3.02855015e-01 8.76034915e-01 -4.50563669e-01 -1.03197205e+00 1.08943546e+00 2.86566138e-01 5.26187122e-01 7.34010994e-01 2.03684941e-01 -9.20747638e-01 -1.51586272e-02 -8.91236246e-01 8.08049500e-01 1.41395485e+00 -8.28992188e-01 -1.08834577e+00 4.08066183e-01 1.08426166e+00 -2.48216152e-01 -6.25880063e-01 2.26625293e-01 3.28041390e-02 -9.45033193e-01 9.91426408e-01 -9.91721153e-01 7.50351012e-01 -5.73706746e-01 -1.85078710e-01 -1.37659419e+00 -4.01080668e-01 -2.68106014e-01 -2.47499526e-01 1.25025177e+00 2.07744926e-01 -3.76677930e-01 4.82522488e-01 5.19552648e-01 -1.09603703e-01 -5.80705225e-01 -5.52951038e-01 -8.17382634e-01 -3.44885230e-01 -3.27715427e-01 7.13356793e-01 1.30432117e+00 1.45632118e-01 9.36191678e-01 -2.75009871e-01 2.83578336e-01 5.98409891e-01 5.90507925e-01 8.24707389e-01 -9.13635254e-01 -4.01138574e-01 -1.11701116e-01 -7.02074885e-01 -1.02884901e+00 3.46382022e-01 -9.90170658e-01 1.26521138e-03 -2.06597304e+00 4.99339283e-01 -9.99911353e-02 -2.59008646e-01 7.96478271e-01 -4.60515440e-01 -2.19957635e-01 3.51688236e-01 -2.11873904e-01 -8.44917059e-01 6.44151330e-01 1.59724736e+00 -4.26011473e-01 1.12450905e-01 -6.84516132e-01 -9.75985885e-01 8.16685557e-01 5.19565701e-01 -1.68014616e-01 -8.08737516e-01 -4.72728521e-01 5.30304909e-01 7.53359050e-02 9.14688706e-01 -6.52971029e-01 2.46681482e-01 -5.00310779e-01 2.90914267e-01 -2.73841739e-01 2.52511859e-01 -8.05521309e-01 8.51556137e-02 -2.91794981e-03 -1.34086415e-01 -2.17053726e-01 -2.11198851e-02 7.63495266e-01 -4.38979805e-01 3.96286361e-02 4.70038444e-01 -5.79739511e-01 -1.39347029e+00 3.97666216e-01 1.84062779e-01 3.82550567e-01 1.14089906e+00 -2.99628377e-01 -7.61068225e-01 -4.30934340e-01 -8.13767433e-01 2.85414457e-01 2.77012378e-01 5.09984076e-01 9.31356132e-01 -1.51424813e+00 -3.95892799e-01 7.31221661e-02 7.36084342e-01 4.22982901e-01 4.68570471e-01 3.94526064e-01 -2.69984812e-01 -8.64777789e-02 -2.55641013e-01 -3.53301644e-01 -9.48702872e-01 1.12785888e+00 1.84889585e-01 3.40037718e-02 -9.13971305e-01 9.83000457e-01 6.54932320e-01 -1.69934094e-01 -1.53868729e-02 -4.21428293e-01 -2.84582615e-01 -1.10543393e-01 5.16513109e-01 -1.63744256e-01 -4.96292382e-01 -7.21751630e-01 -2.99136519e-01 3.73519063e-01 -9.23957601e-02 3.70808214e-01 1.05863035e+00 -1.03166610e-01 -3.92452121e-01 6.67683780e-01 1.15751410e+00 -1.22396760e-01 -9.31635559e-01 -4.48509961e-01 -1.45656735e-01 -6.11805797e-01 -3.71231079e-01 -7.09273696e-01 -1.09660172e+00 9.58328009e-01 1.39029538e-02 3.12201023e-01 1.19629228e+00 5.28458953e-01 6.49738967e-01 5.18351495e-01 5.07222831e-01 -7.99689651e-01 4.06208873e-01 4.52975661e-01 1.01644945e+00 -1.42881870e+00 1.85219329e-02 -7.17466474e-01 -1.02207971e+00 8.53605211e-01 9.66399968e-01 -2.20548436e-01 5.48596501e-01 -9.31810662e-02 -1.24822259e-01 -8.08277071e-01 -5.84618270e-01 -7.03303754e-01 3.70831817e-01 8.17303419e-01 1.95997894e-01 3.85672241e-01 2.71911263e-01 7.29972541e-01 -2.63642639e-01 -9.80903432e-02 4.40001905e-01 8.26832891e-01 -4.02905077e-01 -5.77067673e-01 7.67076537e-02 4.97651666e-01 8.22149366e-02 -3.50576162e-01 -6.21138513e-01 8.90374005e-01 3.68292660e-01 9.17629063e-01 3.53988446e-02 -4.39546496e-01 4.24037874e-01 -1.61387056e-01 4.80499297e-01 -1.05057919e+00 -1.65784627e-01 -4.76353198e-01 6.72779158e-02 -6.41734004e-01 -6.56057596e-01 -5.70206344e-02 -1.70954192e+00 -1.85496926e-01 -9.05811340e-02 -1.39723614e-01 7.66096711e-02 1.13625133e+00 2.20220640e-01 7.16042519e-01 3.36607903e-01 -3.56281847e-01 1.55825643e-02 -6.16238177e-01 -6.88154995e-01 9.72778916e-01 1.11938529e-01 -8.37338209e-01 -1.79963946e-01 4.14526790e-01]
[10.426462173461914, 1.7129364013671875]
9a6ce385-2923-4a05-8ccd-e4ac9ce9ecf4
learnability-with-pac-semantics-for-multi
2306.0549
null
https://arxiv.org/abs/2306.05490v1
https://arxiv.org/pdf/2306.05490v1.pdf
Learnability with PAC Semantics for Multi-agent Beliefs
The tension between deduction and induction is perhaps the most fundamental issue in areas such as philosophy, cognition and artificial intelligence. In an influential paper, Valiant recognised that the challenge of learning should be integrated with deduction. In particular, he proposed a semantics to capture the quality possessed by the output of Probably Approximately Correct (PAC) learning algorithms when formulated in a logic. Although weaker than classical entailment, it allows for a powerful model-theoretic framework for answering queries. In this paper, we provide a new technical foundation to demonstrate PAC learning with multi-agent epistemic logics. To circumvent the negative results in the literature on the difficulty of robust learning with the PAC semantics, we consider so-called implicit learning where we are able to incorporate observations to the background theory in service of deciding the entailment of an epistemic query. We prove correctness of the learning procedure and discuss results on the sample complexity, that is how many observations we will need to provably assert that the query is entailed given a user-specified error bound. Finally, we investigate under what circumstances this algorithm can be made efficient. On the last point, given that reasoning in epistemic logics especially in multi-agent epistemic logics is PSPACE-complete, it might seem like there is no hope for this problem. We leverage some recent results on the so-called Representation Theorem explored for single-agent and multi-agent epistemic logics with the only knowing operator to reduce modal reasoning to propositional reasoning.
['Brendan Juba', 'Vaishak Belle', 'Ionela G. Mocanu']
2023-06-08
null
null
null
null
['philosophy']
['miscellaneous']
[ 2.46229053e-01 9.70591068e-01 -1.33045629e-01 -2.79103726e-01 -1.12591672e+00 -7.26016283e-01 5.82064092e-01 3.63630503e-01 -4.87421334e-01 1.06364775e+00 1.06627457e-01 -6.37925506e-01 -7.50399649e-01 -1.12175405e+00 -1.13963759e+00 -8.08570504e-01 -1.83676302e-01 7.52882898e-01 4.02168185e-01 -1.91814750e-02 3.25876437e-02 2.38906398e-01 -1.59095514e+00 2.89035529e-01 7.57248998e-01 9.13936377e-01 -3.63702297e-01 5.39941967e-01 2.61282213e-02 1.32110846e+00 -2.95132771e-02 -7.08541930e-01 1.21523649e-01 -3.54276627e-01 -1.28852487e+00 2.80434894e-03 2.87583619e-01 -2.21352383e-01 9.25398152e-03 1.55558598e+00 9.62223113e-02 -1.53018460e-01 4.99012500e-01 -1.52732253e+00 -9.61704627e-02 1.32853031e+00 5.57964705e-02 -1.05607003e-01 6.43742740e-01 1.55069605e-01 1.52208495e+00 -1.57919992e-02 4.50921804e-01 1.38158381e+00 5.16122222e-01 5.28687119e-01 -1.21178448e+00 -1.25267208e-01 1.50157914e-01 5.95540822e-01 -9.53057289e-01 -2.70891160e-01 4.12779003e-01 -3.44102532e-01 5.38684130e-01 5.24511218e-01 5.02098083e-01 7.25295663e-01 7.71881416e-02 9.44844842e-01 1.52600300e+00 -9.67025101e-01 6.27773106e-01 4.21431571e-01 3.44667614e-01 7.60761738e-01 6.47927463e-01 3.05793256e-01 -4.48084503e-01 -4.21772301e-01 2.55508929e-01 -3.99746776e-01 -2.28702217e-01 -4.85871911e-01 -9.81565714e-01 1.17009890e+00 -8.64723027e-02 8.35576355e-02 -2.24089429e-01 2.67196029e-01 4.02544320e-01 7.32659221e-01 -2.36671790e-02 4.45068628e-01 -6.12211287e-01 -9.34443027e-02 -4.12855119e-01 4.04417604e-01 1.48769665e+00 6.61133647e-01 4.09925789e-01 -3.84110242e-01 2.54800379e-01 -1.63959980e-01 5.80339432e-01 5.15240312e-01 -1.68470412e-01 -1.53255236e+00 1.04937382e-01 3.79378408e-01 4.86807555e-01 -3.43014807e-01 -2.27978498e-01 6.36877343e-02 -1.18417330e-01 5.60149193e-01 9.36757088e-01 -2.07165018e-01 -1.43773183e-01 2.03303146e+00 4.89225149e-01 1.55092388e-01 6.01550817e-01 5.87332428e-01 1.97358504e-01 4.81672376e-01 7.87136182e-02 -6.83057129e-01 1.38326442e+00 -5.62828302e-01 -5.38963735e-01 -4.58107069e-02 7.86667943e-01 -8.89175832e-02 9.79639173e-01 7.50544548e-01 -1.22130752e+00 4.10209149e-01 -1.07755721e+00 2.09807500e-01 -2.52983719e-02 -7.33166456e-01 1.13699329e+00 7.11613357e-01 -6.97440863e-01 3.89425665e-01 -8.02539110e-01 -2.59628028e-01 1.70048445e-01 3.64823908e-01 -3.85573357e-01 -9.92197841e-02 -1.35694158e+00 1.19660389e+00 8.25405121e-01 -9.73083079e-02 -1.10402489e+00 -1.82290703e-01 -7.59798408e-01 8.06111917e-02 1.08911526e+00 -5.99447846e-01 1.65596735e+00 -1.09413099e+00 -1.65848458e+00 7.07902849e-01 -1.53580913e-02 -9.32976127e-01 7.63207316e-01 3.96552030e-03 -1.76285878e-01 3.65820080e-01 -9.06983986e-02 -1.36344567e-01 4.53748256e-01 -1.12315106e+00 -9.09650803e-01 -5.10593057e-01 1.09690440e+00 1.67263493e-01 3.81093085e-01 -7.63241053e-02 3.36731672e-01 3.95095915e-01 2.35825881e-01 -1.04431105e+00 -2.08392471e-01 -1.26684874e-01 -1.43749386e-01 -5.91321051e-01 2.03041002e-01 8.65716264e-02 5.94958723e-01 -2.00387740e+00 -5.35802618e-02 4.47537690e-01 6.81900559e-03 -1.28612980e-01 1.17123850e-01 4.86160010e-01 1.18174866e-01 4.37521413e-02 -1.21241458e-01 3.23798805e-01 7.30316460e-01 5.63761115e-01 -8.05217206e-01 7.87993133e-01 -7.29870796e-02 7.60203302e-01 -9.53699529e-01 -5.65753222e-01 -1.52147477e-02 -4.72065173e-02 -7.92108715e-01 -1.41525967e-02 -9.35279071e-01 1.34594783e-01 -5.88158488e-01 3.11045051e-01 5.21869361e-01 -3.17116410e-01 5.30431747e-01 3.50343257e-01 1.17625341e-01 6.26028955e-01 -1.78156853e+00 1.28138328e+00 -2.39601165e-01 2.43528917e-01 2.26582855e-01 -1.02044427e+00 8.81646425e-02 5.90195596e-01 6.59122989e-02 -3.16557884e-01 2.06240937e-01 4.08658385e-01 2.77893730e-02 -7.49060214e-01 8.71282965e-02 -8.79584908e-01 -3.01986843e-01 7.24302948e-01 4.09591896e-03 -2.50324100e-01 1.27129510e-01 9.65642706e-02 9.07404482e-01 2.26628825e-01 1.99725732e-01 -3.84887338e-01 8.91015768e-01 2.42089495e-01 5.95487595e-01 1.10802293e+00 -8.48946422e-02 -2.31697261e-01 1.10735404e+00 -4.08785254e-01 -5.47356963e-01 -1.02479529e+00 -1.98934004e-01 1.19259501e+00 1.65853709e-01 -4.00883019e-01 -6.98413193e-01 -6.35312319e-01 -1.83496520e-01 1.05594289e+00 -5.25954962e-01 -1.21743046e-01 -1.39830530e-01 -4.47428018e-01 7.44225025e-01 2.02218086e-01 2.65979379e-01 -1.00758040e+00 -8.61779213e-01 4.26991098e-02 -8.74360874e-02 -8.58953536e-01 4.01150256e-01 6.98585689e-01 -8.06992471e-01 -1.48630321e+00 2.90994853e-01 -4.05077994e-01 4.42273825e-01 -4.73771870e-01 8.92526865e-01 -2.39334125e-02 6.09336972e-01 7.64286160e-01 -1.89500466e-01 -8.33522320e-01 -6.77818894e-01 -2.12999597e-01 9.84873772e-02 -3.51802498e-01 6.59239709e-01 -3.66228402e-01 5.73206693e-02 -1.92964017e-01 -1.13800430e+00 -1.69980183e-01 3.32005054e-01 6.93620801e-01 4.12462294e-01 6.24596715e-01 3.40442091e-01 -1.24504995e+00 4.57639217e-01 -2.65190452e-01 -1.02712607e+00 5.96201897e-01 -5.92677951e-01 6.06741011e-01 5.94445944e-01 -2.72668511e-01 -1.01944292e+00 -5.34818545e-02 1.46633625e-01 3.82883787e-01 -9.25510526e-02 9.13092375e-01 -5.00504553e-01 7.56436959e-02 7.11574018e-01 2.22012416e-01 1.17236376e-01 7.25983679e-02 4.78757620e-01 4.73785877e-01 4.33749348e-01 -1.33270955e+00 8.49356890e-01 6.87347233e-01 4.59741235e-01 -4.34807152e-01 -1.41530681e+00 -1.20464601e-02 -3.20884287e-01 7.50252232e-02 4.37229604e-01 -7.40397871e-01 -1.40611935e+00 1.13023520e-02 -9.24936056e-01 -4.93928254e-01 -6.83269083e-01 6.58754945e-01 -1.25404775e+00 4.24936146e-01 -4.98673171e-01 -1.47700918e+00 1.47014439e-01 -1.08514905e+00 4.90952224e-01 5.18848039e-02 -6.32026196e-02 -1.11113584e+00 9.36508849e-02 3.14810783e-01 2.88711712e-02 5.38156480e-02 1.13865018e+00 -1.22307491e+00 -6.47885203e-01 -2.02364907e-01 1.42662287e-01 3.11307251e-01 -4.17844027e-01 -1.64824247e-01 -1.15498495e+00 -1.01353385e-01 3.64119172e-01 -6.74445629e-01 4.75503922e-01 2.33306959e-01 6.66845024e-01 -7.94720232e-01 1.48231849e-01 7.57259410e-03 1.64979506e+00 -1.08169943e-01 6.94320440e-01 7.69730270e-01 -1.41001493e-01 7.34067202e-01 6.28242493e-01 2.73615092e-01 4.87689137e-01 1.99101761e-01 3.60777080e-01 8.78727496e-01 6.57980323e-01 -2.45220378e-01 4.80810583e-01 2.55174696e-01 -1.32369995e-01 3.25598419e-01 -6.43836558e-01 1.80874616e-01 -2.04081941e+00 -1.53981531e+00 2.42040157e-02 2.46989870e+00 1.50268185e+00 4.96944189e-01 1.06995508e-01 3.94642502e-01 3.69633019e-01 -1.96379423e-01 -3.18613827e-01 -5.16961813e-01 -2.17772678e-01 1.65307641e-01 4.60061133e-01 1.09227073e+00 -9.15843248e-01 6.58901811e-01 6.37110519e+00 5.42831838e-01 -7.13217735e-01 2.43994161e-01 4.05618660e-02 1.25956908e-01 -7.39311576e-01 6.88950658e-01 -6.31642818e-01 2.01409906e-01 1.34811199e+00 -4.44774419e-01 5.93882680e-01 1.03729045e+00 -1.87356129e-01 -6.35085166e-01 -1.59772837e+00 4.29820091e-01 -1.35805160e-01 -1.07803440e+00 -2.19586208e-01 8.69581848e-02 5.12395322e-01 -2.70690650e-01 -1.88160583e-01 4.73728180e-01 7.97519624e-01 -1.07166755e+00 9.81448412e-01 4.77432519e-01 3.49981189e-01 -9.44085360e-01 9.04596448e-01 8.12553704e-01 -5.59515417e-01 -4.63199407e-01 -4.13007498e-01 -4.83503997e-01 -6.39132112e-02 3.79663110e-01 -5.64201832e-01 4.88570869e-01 7.72318244e-02 2.91209370e-02 -1.04454614e-01 1.08932757e+00 -6.15472317e-01 5.70017457e-01 -7.33071625e-01 -8.57845973e-03 3.10550600e-01 -2.56826907e-01 4.58174884e-01 8.05499732e-01 -2.32158434e-02 1.95316628e-01 -4.40271161e-02 6.49726272e-01 1.90136209e-01 -3.31193566e-01 -5.12668967e-01 7.77478293e-02 3.53679925e-01 8.72713447e-01 -3.55886668e-01 -4.74955678e-01 -3.89779210e-01 2.71931887e-01 1.41110808e-01 1.02232732e-01 -7.32094526e-01 3.94151397e-02 1.75540417e-01 -9.51360017e-02 2.18515441e-01 2.31915787e-01 -2.34817397e-02 -1.07138634e+00 9.06560794e-02 -1.10308325e+00 9.01466012e-01 -5.98351240e-01 -1.22048795e+00 8.18557069e-02 2.86354125e-01 -7.52519190e-01 -4.58671391e-01 -8.05985510e-01 -3.58124256e-01 4.97404754e-01 -1.79117441e+00 -1.00411165e+00 3.16149622e-01 7.27798939e-01 -1.18180066e-01 3.32623810e-01 1.12788057e+00 -3.03365231e-01 3.52231115e-02 2.06208959e-01 -1.34853959e-01 -1.85992628e-01 4.44521755e-01 -1.56396902e+00 -7.45912910e-01 9.08894181e-01 2.87861288e-01 6.06658936e-01 1.24970019e+00 -2.01197386e-01 -1.82591653e+00 -7.38837779e-01 9.44179893e-01 -7.09658861e-01 1.09659970e+00 2.62134075e-01 -6.69621587e-01 1.22777283e+00 1.69567481e-01 -6.05999418e-02 6.81728959e-01 3.06045443e-01 -6.77913964e-01 -1.67046130e-01 -1.26226950e+00 5.14566064e-01 5.71440756e-01 -7.88677871e-01 -1.30547452e+00 4.17418092e-01 6.94278121e-01 -3.32003027e-01 -7.12562442e-01 2.12472200e-01 5.34699261e-01 -9.13774848e-01 4.97730136e-01 -8.91358852e-01 6.21488206e-02 -6.41510844e-01 -5.05832732e-01 -8.31174433e-01 3.27644683e-02 -7.75324464e-01 -4.28483456e-01 7.39347398e-01 4.50268537e-01 -9.49418843e-01 5.60813487e-01 1.13631022e+00 1.41686082e-01 -4.51093644e-01 -1.34279895e+00 -7.21668124e-01 5.37370503e-01 -8.81981969e-01 4.43863481e-01 5.72990119e-01 7.07059801e-01 2.42180109e-01 6.59686886e-03 5.33959687e-01 8.06600392e-01 2.26804733e-01 4.89349455e-01 -1.60372436e+00 -6.61895335e-01 -2.72658587e-01 -3.30486774e-01 -5.80383122e-01 5.98558843e-01 -9.29567814e-01 2.40906492e-01 -1.41030335e+00 3.98557603e-01 -4.73215312e-01 -2.54873693e-01 5.11598647e-01 1.59442812e-01 -3.30681264e-01 1.87357858e-01 2.74016932e-02 -1.19280422e+00 7.21587092e-02 1.03322339e+00 -1.47521228e-01 9.02990922e-02 2.79140383e-01 -1.03915846e+00 1.05984485e+00 6.88369572e-01 -4.28165376e-01 -3.87053788e-01 -3.37187350e-02 1.12446189e+00 1.56949222e-01 6.50129616e-01 -7.79511333e-01 6.06381655e-01 -5.98496556e-01 -4.99419689e-01 -6.25899062e-02 4.02066596e-02 -1.12636554e+00 2.31576234e-01 6.94922388e-01 -8.09038162e-01 -5.03273726e-01 -2.29699478e-01 6.20643377e-01 5.04854433e-02 -9.05831277e-01 6.72029257e-01 -2.74096817e-01 -5.80689251e-01 -1.16129657e-02 -4.76420522e-01 3.63028347e-01 1.10221148e+00 2.61226267e-01 -3.61579984e-01 -4.24622089e-01 -8.60914588e-01 1.58926532e-01 4.45308656e-01 -5.13824880e-01 2.62320876e-01 -8.61615062e-01 -5.81958652e-01 -2.98756927e-01 1.47197351e-01 6.99238330e-02 1.48284237e-03 1.23470807e+00 -3.24134707e-01 6.32113516e-01 1.14817493e-01 -2.45253608e-01 -9.77784336e-01 7.21913159e-01 4.75643069e-01 -1.99310437e-01 -5.56598663e-01 5.81953943e-01 -1.80461898e-01 -3.66507173e-01 6.14254057e-01 -3.16107303e-01 6.77711815e-02 -8.62325504e-02 8.87414515e-01 1.57910213e-01 -2.09177032e-01 -2.62474597e-01 -3.69662404e-01 1.22473471e-01 1.45493940e-01 -4.67799157e-01 1.44940102e+00 -2.07585648e-01 -4.63118196e-01 7.98001528e-01 4.58596855e-01 2.45840237e-01 -1.14921141e+00 -4.89667654e-01 4.17560667e-01 1.47011727e-02 -2.26388305e-01 -8.51723313e-01 -4.65715736e-01 5.91808438e-01 2.06371143e-01 5.87063134e-01 7.49186337e-01 4.96209055e-01 1.77245900e-01 1.02713752e+00 9.16438162e-01 -1.22840130e+00 -4.95064497e-01 5.57251096e-01 4.80906993e-01 -1.11886299e+00 2.39547864e-01 -8.38905796e-02 -7.05031097e-01 1.14877939e+00 9.32110995e-02 -1.60373356e-02 2.95977145e-01 3.62190157e-01 -2.25270510e-01 -2.31242150e-01 -1.13526022e+00 -3.31040800e-01 -2.74543256e-01 5.34619987e-01 -5.99546842e-02 2.60797441e-01 -2.32658431e-01 7.84911096e-01 -3.84616017e-01 2.17390716e-01 8.54584157e-01 1.04512334e+00 -8.83536696e-01 -1.07493711e+00 -4.71219838e-01 1.46431744e-01 -8.93704474e-01 6.48702960e-03 -4.75261826e-03 6.88029110e-01 -7.41135180e-02 1.14997375e+00 -4.16334182e-01 2.22307324e-01 -3.91045064e-02 3.11972231e-01 9.12027538e-01 -5.36480486e-01 -7.19685182e-02 -3.01057518e-01 3.86631668e-01 -4.13396984e-01 -9.01211977e-01 -5.98878384e-01 -1.34762144e+00 -5.38589954e-01 -4.68194008e-01 7.98362076e-01 2.27769554e-01 1.42580390e+00 -3.05819154e-01 -4.82096970e-02 2.91190058e-01 -1.24763452e-01 -1.50923324e+00 -6.58566952e-01 -8.99956167e-01 4.85011190e-02 4.37810749e-01 -3.64461362e-01 -9.70235646e-01 5.85577115e-02]
[8.593092918395996, 6.605257034301758]
6412cb91-b5b2-43ba-88ac-416292e7e9e0
pre-trained-language-models-with-domain
null
null
https://www.sciencedirect.com/science/article/pii/S0950705122007328
https://reader.elsevier.com/reader/sd/pii/S0950705122007328?token=6B0D860FD9A7EA3A7BFE32EF631BCD4F592ADB7606EF0F28451448E7470CE1EF7C7DFE161EB4A1F743EC314FCDD5608C&originRegion=eu-west-1&originCreation=20220724191534
Pre-trained language models with domain knowledge for biomedical extractive summarization
Biomedical text summarization is a critical task for comprehension of an ever-growing amount of biomedical literature. Pre-trained language models (PLMs) with transformer-based architectures have been shown to greatly improve performance in biomedical text mining tasks. However, existing methods for text summarization generally fine-tune PLMs on the target corpora directly and do not consider how fine-grained domain knowledge, such as PICO elements used in evidence-based medicine, can help to identify the context needed for generating coherent summaries. To fill the gap, we propose KeBioSum, a novel knowledge infusion training framework, and experiment using a number of PLMs as bases, for the task of extractive summarization on biomedical literature. We investigate generative and discriminative training techniques to fuse domain knowledge (i.e., PICO elements) into knowledge adapters and apply adapter fusion to efficiently inject the knowledge adapters into the basic PLMs for fine-tuning the extractive summarization task. Experimental results from the extractive summarization task on three biomedical literature datasets show that existing PLMs (BERT, RoBERTa, BioBERT, and PubMedBERT) are improved by incorporating the KeBioSum knowledge adapters, and our model outperforms the strong baselines.
['QianqianXie;Jennifer Amy Bishop;PrayagTiwari;Sophia Ananiadoua']
2022-07-19
null
null
null
knowledge-based-systems-2022-7
['pico', 'extractive-summarization']
['natural-language-processing', 'natural-language-processing']
[ 5.21376073e-01 4.95454967e-01 -4.83701050e-01 -2.01669946e-01 -1.19739354e+00 -3.15245748e-01 3.72050494e-01 7.47588396e-01 -3.46740365e-01 1.06043828e+00 1.09912467e+00 -2.02319950e-01 -1.47676080e-01 -5.19764781e-01 -8.55431736e-01 -4.43835229e-01 2.65461624e-01 7.67251492e-01 -3.01418975e-02 -2.33299717e-01 3.79661739e-01 1.28260195e-01 -7.78010368e-01 9.02456105e-01 1.43623149e+00 3.57647836e-01 2.12253228e-01 9.15239036e-01 -3.90411198e-01 8.62056494e-01 -1.05423772e+00 -4.50891376e-01 -5.66627979e-01 -7.30918348e-01 -1.24488306e+00 -4.37949777e-01 3.41450691e-01 -1.31583288e-01 -2.53537208e-01 7.46583581e-01 9.53586161e-01 -8.14801008e-02 9.74575460e-01 -6.69700265e-01 -5.82612455e-01 1.27833843e+00 -3.72607052e-01 4.68027830e-01 2.56995559e-01 -5.83771765e-02 9.51441526e-01 -5.45839131e-01 9.16819692e-01 1.27698958e+00 6.70329511e-01 7.92659461e-01 -1.10651028e+00 -4.50488627e-01 4.97484878e-02 3.02248389e-01 -1.00871146e+00 -6.94324851e-01 4.81848329e-01 -2.21236661e-01 1.46913409e+00 4.45699573e-01 4.96953249e-01 1.30550754e+00 7.24608600e-01 1.18631959e+00 3.29346389e-01 -2.24347293e-01 1.55621156e-01 -7.86357597e-02 4.13659453e-01 6.76189721e-01 6.15846753e-01 -6.94052815e-01 -8.44967544e-01 -5.30439019e-01 2.39219904e-01 -2.09545895e-01 -4.39701557e-01 3.91187578e-01 -1.46262002e+00 7.33332872e-01 2.65427798e-01 3.68203431e-01 -5.74381709e-01 -5.69588728e-02 9.63058591e-01 2.84479726e-02 6.17873371e-01 1.07970083e+00 -7.00719297e-01 -1.32247239e-01 -1.33965707e+00 1.74443156e-01 1.03940547e+00 1.13800967e+00 3.14384997e-01 -1.40972883e-01 -8.71999979e-01 9.32839990e-01 -2.65466928e-01 3.30747426e-01 9.18045104e-01 -7.36779034e-01 6.39386773e-01 8.22982669e-01 -3.86110812e-01 -7.48124659e-01 -6.83729351e-01 -7.32876182e-01 -1.13792777e+00 -1.02791679e+00 -2.30844975e-01 -2.98441917e-01 -9.24998224e-01 1.48634541e+00 1.28535494e-01 2.67552793e-01 5.07015288e-01 3.15822810e-01 1.67362320e+00 6.12666786e-01 1.45198524e-01 -4.34957623e-01 1.71057105e+00 -1.01890695e+00 -1.02745974e+00 -3.01134408e-01 8.61279547e-01 -5.40794015e-01 6.27218604e-01 2.50646770e-01 -1.25821292e+00 -1.67769551e-01 -1.05097950e+00 -6.18988991e-01 -2.52040744e-01 2.77310878e-01 3.69945914e-01 1.69570088e-01 -9.49403763e-01 6.12722516e-01 -1.03225052e+00 -4.16944593e-01 7.38737941e-01 1.12364702e-01 -1.41428441e-01 1.15746923e-01 -1.24646366e+00 1.03220415e+00 7.97586560e-01 -1.05824873e-01 -7.67316997e-01 -1.29981351e+00 -8.60611975e-01 3.57414782e-01 2.58765936e-01 -1.69603908e+00 1.22692502e+00 -2.00630009e-01 -1.32721901e+00 8.91510367e-01 -4.17365044e-01 -8.25169921e-01 2.00986907e-01 -2.71136671e-01 -5.20740114e-02 5.96903205e-01 4.34803039e-01 7.28206694e-01 3.14403117e-01 -7.62890637e-01 -4.44720209e-01 -2.80620903e-01 -2.97324896e-01 1.85615137e-01 -3.97355109e-01 -1.06319472e-01 -4.50638235e-01 -8.50857317e-01 -4.43170995e-01 -3.97011667e-01 -3.63089234e-01 -8.37130845e-01 -1.10029519e+00 -4.10827607e-01 2.28144854e-01 -1.03498662e+00 1.63095641e+00 -1.57358170e+00 2.92764246e-01 -2.65176296e-01 3.40489417e-01 3.99645627e-01 -2.12273806e-01 5.87921202e-01 -2.83061415e-02 2.05046877e-01 -4.57334131e-01 -2.19940841e-01 -1.09011039e-01 2.59560287e-01 -4.68729407e-01 -1.03174090e-01 3.77071768e-01 1.30276489e+00 -1.30951953e+00 -1.03255928e+00 -3.12815696e-01 2.50691533e-01 -6.80444658e-01 1.56191319e-01 -5.27828038e-01 3.98940802e-01 -7.32948422e-01 5.51651597e-01 1.83335468e-01 -4.97815013e-01 2.11175561e-01 -5.16396999e-01 3.89303535e-01 8.12946081e-01 -2.43870243e-01 1.95870781e+00 -3.04160953e-01 2.93693125e-01 -3.21884751e-01 -1.21967876e+00 5.66354990e-01 5.82954049e-01 7.17233181e-01 -3.44976991e-01 1.05207562e-01 2.34681681e-01 -1.79892361e-01 -6.14956439e-01 6.05011106e-01 -1.14914261e-01 -1.84857711e-01 4.28867251e-01 4.07598764e-01 -2.25641936e-01 5.64357638e-01 6.50037467e-01 1.54299915e+00 -9.67051983e-02 8.40549290e-01 -2.68159479e-01 5.70757329e-01 2.92340189e-01 6.24848783e-01 8.63461018e-01 3.36027294e-01 5.56813598e-01 6.75855100e-01 -1.03217829e-02 -6.60249770e-01 -7.60433555e-01 -1.05120838e-01 9.76353884e-01 -3.78242970e-01 -9.45899189e-01 -9.82576787e-01 -7.80390739e-01 -1.36357218e-01 1.09150517e+00 -5.48911929e-01 -6.89461172e-01 -7.09416628e-01 -1.13371384e+00 1.06688702e+00 5.70958018e-01 3.29152912e-01 -1.06726587e+00 -5.44426262e-01 4.56869990e-01 -8.17525923e-01 -1.19535470e+00 -7.19298244e-01 1.23853050e-01 -1.11442280e+00 -9.72846270e-01 -7.46507823e-01 -6.68797135e-01 6.37675047e-01 -6.27112314e-02 1.51947880e+00 -1.30456582e-01 -3.42454195e-01 4.46281254e-01 -3.32162231e-01 -8.99276316e-01 -8.89268875e-01 7.46150374e-01 -3.75836372e-01 -5.29338241e-01 1.81018129e-01 -5.19315898e-01 -6.19585276e-01 -3.45622450e-01 -1.08626461e+00 6.23298049e-01 9.91686046e-01 1.02164948e+00 6.72324121e-01 -4.03887928e-01 1.20001709e+00 -1.34404826e+00 9.85814333e-01 -5.35611808e-01 2.36187354e-01 5.48771262e-01 -4.27641958e-01 4.67050374e-01 5.96120000e-01 -2.01914087e-01 -9.43689108e-01 -5.03032327e-01 -4.65091735e-01 1.45149156e-01 2.04984009e-01 1.07975399e+00 -7.64268711e-02 5.95983744e-01 9.22669530e-01 5.95927000e-01 -9.78406668e-02 -4.29107308e-01 5.55168748e-01 6.72965884e-01 7.29929268e-01 -6.30251825e-01 7.98041373e-03 2.86067754e-01 -8.89421776e-02 -7.92220235e-01 -1.20934308e+00 -5.87450266e-01 -3.62319529e-01 4.07394499e-01 8.03255737e-01 -9.08313572e-01 -4.02844906e-01 -6.93570971e-02 -1.45528483e+00 -1.04760289e-01 -4.33670133e-01 1.69564947e-01 -6.34064674e-01 5.65996170e-01 -7.76444912e-01 -5.09982407e-02 -1.40039968e+00 -9.06108975e-01 1.46905839e+00 1.79980770e-01 -7.95836449e-01 -1.07931900e+00 1.69654340e-01 4.63563979e-01 1.08092546e-01 2.62131780e-01 1.49838424e+00 -1.09097290e+00 -7.57165849e-02 1.72330011e-02 -2.58025452e-02 1.83288023e-01 3.40253711e-01 -3.97383243e-01 -6.62949681e-01 1.30151911e-02 -3.39531563e-02 -2.16643527e-01 1.39210784e+00 6.46368563e-01 1.28488600e+00 -8.52303922e-01 -8.83918524e-01 4.71100211e-01 8.11935365e-01 -2.10072964e-01 5.18158495e-01 7.06327036e-02 6.64424598e-01 3.01221758e-01 2.94571668e-01 3.89491290e-01 7.18761921e-01 2.63722301e-01 -2.69324243e-01 -5.81677780e-02 -4.25609350e-01 -1.07604109e-01 3.46674651e-01 1.46836424e+00 1.06179751e-01 -2.03015476e-01 -8.00793409e-01 7.13342071e-01 -1.93951631e+00 -8.35748971e-01 1.84402198e-01 1.56987298e+00 1.95918870e+00 -1.12093784e-01 -1.76695138e-01 -3.34215462e-01 2.93078303e-01 -8.43370557e-02 -7.44177818e-01 -4.06560600e-01 -1.36690423e-01 4.07255113e-01 2.13369027e-01 2.95946389e-01 -7.42173672e-01 8.79447520e-01 6.18358517e+00 1.09634399e+00 -8.83064747e-01 5.70102595e-02 5.21304905e-01 -2.54125208e-01 -4.01192248e-01 -2.69571215e-01 -1.01105690e+00 3.65993381e-01 1.11001933e+00 -8.05479765e-01 -2.51773089e-01 4.52095836e-01 2.47977510e-01 1.21755218e-02 -1.48418796e+00 7.27925062e-01 4.06217873e-01 -1.91682899e+00 7.19557047e-01 -3.19865048e-01 8.48927975e-01 4.36244421e-02 -3.64126027e-01 4.93916899e-01 3.90509218e-01 -9.34221923e-01 2.41585314e-01 8.24503660e-01 4.57903028e-01 -4.07656252e-01 9.29671466e-01 3.89339000e-01 -5.67082286e-01 3.06528986e-01 -4.30647433e-01 6.51284695e-01 1.49756625e-01 9.69759107e-01 -1.56584978e+00 1.17265522e+00 2.03893945e-01 9.26705301e-01 -6.06643677e-01 9.01593089e-01 -2.49374300e-01 8.92415881e-01 -4.47833315e-02 -7.87272155e-02 4.43944242e-03 3.90975535e-01 7.11639524e-01 1.93829811e+00 1.75386906e-01 8.40282589e-02 -1.70598924e-02 8.63591790e-01 -4.36688721e-01 3.13909978e-01 -1.53889686e-01 -4.81057882e-01 2.16525018e-01 1.15767634e+00 -5.02751827e-01 -8.43276799e-01 6.17571399e-02 7.43543923e-01 1.74839452e-01 1.93678081e-01 -6.21149302e-01 -4.71783221e-01 2.20061123e-01 -2.83382267e-01 9.35030058e-02 2.68915206e-01 -5.22605002e-01 -1.30541623e+00 -2.30790287e-01 -1.15960824e+00 7.86530972e-01 -6.14124119e-01 -1.25464070e+00 3.68789285e-01 1.34302750e-01 -8.19717169e-01 -4.45065379e-01 -1.60050079e-01 -4.97873276e-01 7.05095708e-01 -1.55348670e+00 -1.33283174e+00 1.28279850e-01 2.18520030e-01 8.67416024e-01 -1.28904432e-01 1.06052172e+00 7.21824542e-02 -5.80220997e-01 5.93939722e-01 6.33192286e-02 1.01030901e-01 9.45928037e-01 -1.37017965e+00 4.07087415e-01 4.81983662e-01 -1.44770563e-01 1.05991435e+00 7.78855264e-01 -9.41617906e-01 -1.35032058e+00 -1.38084781e+00 1.10148335e+00 -6.23509288e-01 4.34984088e-01 -3.74464057e-02 -1.12888253e+00 6.74781799e-01 4.72797215e-01 -8.14014733e-01 1.12775171e+00 2.66153753e-01 -1.00370176e-01 -1.45920543e-02 -7.82374203e-01 6.43757105e-01 9.69022751e-01 -2.34891236e-01 -1.24574292e+00 6.26233757e-01 1.04946434e+00 -5.32711267e-01 -1.08263493e+00 5.66603243e-01 3.62253547e-01 -2.09920987e-01 1.02901244e+00 -9.93469477e-01 8.67616236e-01 -8.42257291e-02 4.92309988e-01 -1.64387071e+00 -2.39887759e-01 -5.19111872e-01 -4.11364555e-01 1.17795038e+00 5.88245511e-01 -4.79421794e-01 5.08204341e-01 6.48865104e-02 -7.37321913e-01 -9.13519263e-01 -7.47960508e-01 -2.91448265e-01 3.13198477e-01 7.78162405e-02 4.57199305e-01 7.94709265e-01 5.79406023e-01 9.94556248e-01 -1.20428437e-02 -2.35522121e-01 2.63082117e-01 1.89844981e-01 6.57595575e-01 -1.03946304e+00 -1.68496311e-01 -8.08109224e-01 -4.63926680e-02 -9.21658576e-01 2.78682381e-01 -1.49947286e+00 1.92313250e-02 -2.35499787e+00 8.59641969e-01 1.37140319e-01 -2.50043333e-01 8.20909083e-01 -8.39390755e-01 -3.51570278e-01 -2.94651568e-01 4.35025012e-03 -7.79806614e-01 4.93299633e-01 1.24870265e+00 -6.32226169e-01 -2.00293913e-01 -1.99948207e-01 -1.36061692e+00 5.29946029e-01 5.65238416e-01 -4.28865969e-01 -3.71484518e-01 -3.92041445e-01 4.63529170e-01 1.44137502e-01 -4.10116017e-02 -7.24651873e-01 6.10215306e-01 4.31117937e-02 2.93808758e-01 -8.79885972e-01 -1.24248125e-01 6.57339469e-02 -8.78174677e-02 5.70395410e-01 -7.97898054e-01 -2.84902245e-01 5.96786976e-01 3.64856452e-01 -2.21055269e-01 -2.74625808e-01 3.83767515e-01 -2.71125048e-01 -1.49080753e-01 5.55032827e-02 -4.07683372e-01 4.96676177e-01 2.98622012e-01 1.33463994e-01 -6.44453228e-01 -1.15575202e-01 -6.40841186e-01 5.98244250e-01 -7.25476518e-02 2.98053086e-01 6.68048799e-01 -9.59142447e-01 -1.30049908e+00 -2.61083990e-01 1.30912900e-01 2.14031771e-01 4.62303758e-01 1.08194947e+00 -4.09409881e-01 8.03592801e-01 4.82084639e-02 -5.12032270e-01 -1.31710351e+00 4.42336917e-01 1.14300936e-01 -1.05721653e+00 -7.18149602e-01 7.64532447e-01 4.68572527e-01 -3.88125718e-01 3.32261585e-02 -7.97069848e-01 -2.93300360e-01 1.93644896e-01 7.55202472e-01 3.42224836e-01 4.91381228e-01 1.20105781e-02 -3.85209113e-01 1.43229336e-01 -5.87680399e-01 2.19421655e-01 1.52345216e+00 1.14390075e-01 -5.64553618e-01 2.66942084e-01 1.02510858e+00 -4.13931943e-02 -3.07209909e-01 -3.82629633e-01 3.14439327e-01 5.74731588e-01 -9.84732583e-02 -1.20061743e+00 -5.81777096e-01 7.55171955e-01 -2.90847778e-01 -2.52589494e-01 1.23750412e+00 1.83572665e-01 1.08454287e+00 6.78423822e-01 -1.18571162e-01 -1.13273680e+00 2.74495929e-01 6.34795308e-01 1.19376969e+00 -6.74064398e-01 4.39885497e-01 -3.57746542e-01 -6.65557683e-01 1.12479317e+00 2.01042667e-01 3.06221098e-01 1.84780568e-01 2.96282768e-01 -2.23591417e-01 -3.85630012e-01 -1.11376834e+00 2.80704964e-02 5.83631456e-01 3.13520819e-01 5.58233976e-01 4.84086461e-02 -5.62330723e-01 1.16622639e+00 -4.72635388e-01 2.64978796e-01 4.64162052e-01 8.29891503e-01 -3.87645155e-01 -8.70088160e-01 -1.62116736e-01 9.85134482e-01 -8.14392567e-01 -5.75411379e-01 -5.46945810e-01 2.63438791e-01 -1.47623643e-01 8.38787556e-01 -3.41114163e-01 1.95030831e-02 3.98470640e-01 2.26776794e-01 5.83815932e-01 -1.05089259e+00 -9.26330566e-01 1.84409544e-01 4.19666797e-01 -2.41131380e-01 -4.07599717e-01 -4.81357038e-01 -1.42617583e+00 1.46902157e-02 -1.29019901e-01 4.47765231e-01 1.68447256e-01 1.06165862e+00 8.86623621e-01 1.24685013e+00 -1.24582700e-01 -2.80063570e-01 -5.06658494e-01 -1.33037889e+00 -7.20195845e-02 8.25636461e-02 3.61401916e-01 -9.48080644e-02 1.46583349e-01 4.98906910e-01]
[12.140429496765137, 9.342761993408203]
272b393a-1ad1-4448-9396-0a89ce2214a9
bipartite-graph-reasoning-gans-for-person
2008.04381
null
https://arxiv.org/abs/2008.04381v2
https://arxiv.org/pdf/2008.04381v2.pdf
Bipartite Graph Reasoning GANs for Person Image Generation
We present a novel Bipartite Graph Reasoning GAN (BiGraphGAN) for the challenging person image generation task. The proposed graph generator mainly consists of two novel blocks that aim to model the pose-to-pose and pose-to-image relations, respectively. Specifically, the proposed Bipartite Graph Reasoning (BGR) block aims to reason the crossing long-range relations between the source pose and the target pose in a bipartite graph, which mitigates some challenges caused by pose deformation. Moreover, we propose a new Interaction-and-Aggregation (IA) block to effectively update and enhance the feature representation capability of both person's shape and appearance in an interactive way. Experiments on two challenging and public datasets, i.e., Market-1501 and DeepFashion, show the effectiveness of the proposed BiGraphGAN in terms of objective quantitative scores and subjective visual realness. The source code and trained models are available at https://github.com/Ha0Tang/BiGraphGAN.
['Nicu Sebe', 'Hao Tang', 'Philip H. S. Torr', 'Song Bai']
2020-08-10
null
null
null
null
['pose-transfer']
['computer-vision']
[-2.03063980e-01 3.26580018e-01 4.97549683e-01 -3.57413620e-01 -4.81259674e-01 -4.02517885e-01 5.78747869e-01 -4.37538803e-01 1.55462578e-01 6.78371549e-01 2.46432111e-01 4.40229386e-01 1.60394665e-02 -1.00740123e+00 -7.79328108e-01 -4.97766227e-01 2.83203304e-01 4.42390293e-01 1.08684123e-01 -4.82215941e-01 -3.76670629e-01 1.37157679e-01 -1.08124435e+00 -1.41558588e-01 8.66906464e-01 6.93204284e-01 -7.97918215e-02 4.04192150e-01 5.04058003e-01 4.37600851e-01 -5.77929378e-01 -1.04711521e+00 5.08102655e-01 -6.81422234e-01 -4.19545710e-01 4.80709851e-01 5.11949956e-01 -4.15321410e-01 -7.58802533e-01 1.04289174e+00 8.06333721e-01 1.10815398e-01 3.53058606e-01 -1.46993518e+00 -6.78663433e-01 3.55553776e-01 -8.19226921e-01 -2.87639946e-01 8.10132802e-01 4.76310521e-01 7.32602000e-01 -5.11176646e-01 7.60348141e-01 1.45828807e+00 6.16541147e-01 6.37285829e-01 -1.02886367e+00 -7.09396601e-01 1.22019552e-01 1.88435927e-01 -1.42695749e+00 -7.44522437e-02 9.73491013e-01 -4.36240107e-01 1.35109201e-01 4.70175534e-01 1.09369314e+00 1.16797853e+00 3.90903614e-02 5.26202381e-01 9.85857904e-01 -2.13105038e-01 -2.73257792e-01 -2.30862513e-01 -3.10151339e-01 1.03331435e+00 4.98049736e-01 1.45389542e-01 -4.39606667e-01 1.79211780e-01 1.07686210e+00 -1.09128999e-02 -3.10711026e-01 -5.26827633e-01 -9.71084476e-01 6.15271568e-01 8.29968452e-01 -1.07130863e-01 -2.68189579e-01 3.13512474e-01 1.26524404e-01 -1.78131506e-01 1.85338110e-01 4.85582724e-02 -9.56437588e-02 2.35375926e-01 -2.12924808e-01 4.94480401e-01 3.97232652e-01 1.23214543e+00 6.15063488e-01 7.07930997e-02 -6.00352108e-01 8.49240363e-01 4.56957340e-01 8.05714726e-01 -3.35661545e-02 -6.93705261e-01 5.30213535e-01 8.67897689e-01 3.54138277e-02 -1.49122941e+00 -4.66203451e-01 -5.04710138e-01 -1.15212250e+00 -2.20572799e-01 3.00685406e-01 -4.10409510e-01 -1.08261371e+00 1.92651236e+00 5.86285770e-01 1.79966707e-02 -2.66411752e-01 1.23411047e+00 1.43446493e+00 4.82819825e-01 -5.57158068e-02 2.60520488e-01 1.58535969e+00 -1.05811310e+00 -6.60175383e-01 -4.24259216e-01 1.44869328e-01 -6.08195662e-01 9.96665955e-01 7.89906755e-02 -1.30873644e+00 -8.03807616e-01 -7.80540526e-01 -9.76559743e-02 -6.71183318e-02 5.72340012e-01 5.78670502e-01 5.97478330e-01 -8.46518517e-01 1.53740913e-01 -4.37701553e-01 -3.73615563e-01 3.08961034e-01 3.27028304e-01 -5.46615422e-01 -1.96148589e-01 -1.30925560e+00 3.56023043e-01 3.02945256e-01 5.83472848e-01 -7.10367441e-01 -4.91229147e-01 -1.03350580e+00 -6.92990050e-02 5.71346045e-01 -1.08589458e+00 7.72646785e-01 -7.51812339e-01 -1.27843559e+00 9.01929259e-01 3.51869196e-01 -2.93457340e-02 8.50648165e-01 -2.18892410e-01 -3.40361804e-01 1.66890562e-01 4.16024327e-02 9.41824019e-01 8.45945835e-01 -1.36763620e+00 -1.83265626e-01 -5.46602666e-01 1.09243155e-01 4.73091722e-01 -1.74067259e-01 -3.28084439e-01 -1.18838239e+00 -9.90886569e-01 -9.86280814e-02 -1.24915826e+00 -1.84835821e-01 -1.73260048e-01 -9.75516260e-01 7.60273784e-02 5.48462212e-01 -1.16311169e+00 9.81698275e-01 -1.89912271e+00 3.89394075e-01 4.16553766e-01 4.08544272e-01 2.19498456e-01 -1.88482642e-01 4.27405447e-01 -1.99629202e-01 -8.87891427e-02 4.87696454e-02 -3.37565511e-01 1.14074741e-02 -1.22808158e-01 2.75559753e-01 4.56873655e-01 -2.94172060e-04 1.36361647e+00 -7.27931023e-01 -5.09553850e-01 3.43207031e-01 7.06120491e-01 -5.08340597e-01 3.38283986e-01 -5.98255638e-03 8.13136518e-01 -3.63242686e-01 4.80891317e-01 9.47477520e-01 -1.95009589e-01 2.49830380e-01 -5.67647934e-01 4.14076805e-01 -2.87724108e-01 -1.32013178e+00 1.65298998e+00 -3.67221199e-02 -4.14123796e-02 -1.08550757e-01 -1.97723284e-01 9.42568779e-01 1.03001915e-01 4.05921727e-01 -6.98745787e-01 3.83461058e-01 -2.44299650e-01 -7.77264386e-02 -1.98794454e-01 2.10405901e-01 2.13731617e-01 -2.53453195e-01 3.31999250e-02 4.80596162e-02 -7.21521899e-02 3.42098027e-01 4.74713147e-01 8.11564028e-01 1.99620843e-01 1.09269291e-01 -1.09121665e-01 7.55802810e-01 -3.60679418e-01 6.48318708e-01 2.21573710e-01 1.38796484e-02 9.11461592e-01 5.87423503e-01 -4.15988445e-01 -9.83579457e-01 -1.08603156e+00 5.60576439e-01 5.59114099e-01 6.21174514e-01 -7.23538458e-01 -9.72654164e-01 -7.09913373e-01 -4.81502190e-02 3.33458513e-01 -7.52281249e-01 -3.36245000e-01 -4.12956506e-01 -4.97675180e-01 2.99071521e-01 5.09659469e-01 9.45597827e-01 -1.10019052e+00 -1.55020699e-01 -3.61626863e-01 -4.67332780e-01 -1.21339154e+00 -1.03888834e+00 -8.78119528e-01 -3.30656022e-01 -1.09345138e+00 -8.71550322e-01 -7.56756365e-01 1.21565938e+00 1.33210495e-01 8.96400809e-01 3.28543574e-01 -4.05649215e-01 4.74314421e-01 -3.96140039e-01 -5.15559241e-02 -4.16764095e-02 -1.03406534e-01 -1.38632774e-01 3.62283826e-01 -2.01320544e-01 -4.11691695e-01 -9.95802402e-01 6.01879537e-01 -6.98753893e-01 6.26552999e-01 6.83515847e-01 7.48767495e-01 5.42145610e-01 2.24915832e-01 1.77750960e-01 -8.30577374e-01 4.83134508e-01 5.10025956e-03 -4.92949009e-01 2.00172424e-01 -2.36619025e-01 -2.73086578e-01 3.25307310e-01 -1.86238840e-01 -1.24898124e+00 3.25269610e-01 -2.19491974e-01 -3.91834289e-01 -3.76449898e-02 9.46978405e-02 -7.50352144e-01 -1.87245354e-01 2.99211323e-01 2.54665554e-01 6.35893717e-02 -2.44784623e-01 5.16266525e-01 6.68016970e-02 6.68736994e-01 -6.46288753e-01 1.39164376e+00 4.50738698e-01 1.22886248e-01 -5.06527364e-01 -5.58647931e-01 -1.75216421e-01 -4.85875964e-01 -6.67508900e-01 9.95181501e-01 -1.09482741e+00 -1.11858261e+00 9.26001847e-01 -1.11027658e+00 -1.67259187e-01 -2.53914565e-01 1.00314520e-01 -4.48103935e-01 4.27666605e-01 -6.76903486e-01 -4.36396927e-01 -5.19034386e-01 -1.05412328e+00 1.25277972e+00 6.68635488e-01 7.54153132e-02 -8.42729688e-01 -1.38011038e-01 8.41053486e-01 5.11388369e-02 8.07287335e-01 4.28974420e-01 -6.95990026e-02 -6.76700354e-01 -3.14914286e-01 -4.72450405e-01 1.71615452e-01 1.25184759e-01 -3.46848726e-01 -4.82173443e-01 -6.09361231e-01 -5.89190662e-01 -3.23871642e-01 5.18665552e-01 2.73839802e-01 8.94313335e-01 -3.14274997e-01 -2.13505954e-01 6.42574489e-01 1.22355878e+00 -5.73195256e-02 9.68055367e-01 -3.13008875e-02 1.33073473e+00 6.32744074e-01 5.36316216e-01 4.67644542e-01 5.88109910e-01 1.00235641e+00 4.05816853e-01 -5.15848994e-01 -5.99973738e-01 -7.43713677e-01 1.08720623e-01 6.51669681e-01 -5.32062590e-01 -2.94071376e-01 -6.97683513e-01 1.58731923e-01 -1.90266287e+00 -7.96814084e-01 -3.91825140e-01 2.07146072e+00 5.34005046e-01 1.10779606e-01 3.74857217e-01 -2.42915243e-01 9.44823802e-01 6.16615713e-02 -4.11173403e-01 2.21246541e-01 -1.05796956e-01 -1.63843691e-01 3.60903919e-01 4.18565959e-01 -9.31269825e-01 9.16566193e-01 4.80571842e+00 7.90663481e-01 -6.49539113e-01 2.21293066e-02 7.93332160e-01 1.75672024e-01 -2.64575779e-01 -1.35331601e-01 -6.88615084e-01 4.66781318e-01 1.01797491e-01 -1.33918524e-02 5.19713521e-01 5.14824271e-01 7.27230757e-02 1.14697494e-01 -5.99449515e-01 1.28334236e+00 2.98512012e-01 -9.51089501e-01 3.67589206e-01 1.48620293e-01 7.10116506e-01 -8.31788599e-01 1.09533221e-01 2.74325222e-01 1.72234058e-01 -7.41117477e-01 8.06382120e-01 6.77685618e-01 9.61215258e-01 -8.37497294e-01 7.30293512e-01 -6.57351762e-02 -1.54744124e+00 3.62018272e-02 -2.59594154e-02 2.79089332e-01 3.28276396e-01 3.85440826e-01 -5.42501032e-01 1.08016908e+00 7.60839045e-01 5.70852697e-01 -1.06407261e+00 7.98702776e-01 -6.37000918e-01 1.02869570e-01 3.15995589e-02 3.75533730e-01 -1.93767607e-01 -4.54570144e-01 7.95649707e-01 7.13835478e-01 1.93188936e-01 1.71219260e-01 2.25612611e-01 9.90109026e-01 -1.74855873e-01 4.74149920e-02 -3.29546213e-01 8.82126167e-02 1.55287266e-01 1.55078483e+00 -7.06085503e-01 -1.51846269e-02 -1.54854506e-01 1.22924435e+00 1.52932763e-01 2.92260498e-01 -1.20589054e+00 -3.21645945e-01 2.91962326e-01 4.94016379e-01 2.85740793e-01 -3.97230759e-02 6.06045760e-02 -1.20090806e+00 3.01113456e-01 -1.04428911e+00 5.53091228e-01 -1.19088137e+00 -1.19150221e+00 6.39248729e-01 2.09093809e-01 -1.17530847e+00 1.41004482e-02 -2.43131816e-01 -6.21885419e-01 7.22397447e-01 -7.32218206e-01 -1.98870671e+00 -9.39524233e-01 8.58350813e-01 1.75046772e-01 -9.97404233e-02 4.12930608e-01 4.57622707e-01 -6.92093313e-01 7.59411991e-01 -6.99954569e-01 4.91083086e-01 5.88755190e-01 -1.09961236e+00 5.60175240e-01 1.11600411e+00 -7.50118494e-02 5.39977014e-01 6.17013395e-01 -1.01246345e+00 -1.53162062e+00 -1.27376592e+00 3.12164128e-01 -2.80652016e-01 7.96943307e-02 -6.02535307e-01 -2.14597404e-01 7.88476586e-01 1.78820819e-01 1.95220392e-02 1.25117049e-01 -2.90824231e-02 -2.42760077e-01 -3.72498006e-01 -1.11995828e+00 8.50364447e-01 1.36522222e+00 -2.43528541e-02 -1.26909055e-02 4.82988864e-01 6.64189458e-01 -9.76779222e-01 -6.55940533e-01 4.84642237e-01 5.45042396e-01 -1.00136852e+00 1.21030581e+00 -3.08419406e-01 3.09823424e-01 -5.44816375e-01 1.18545279e-01 -1.41158044e+00 -5.34638047e-01 -8.95932555e-01 3.82856317e-02 1.57196677e+00 3.53116728e-02 -6.71856165e-01 8.07571828e-01 5.23286939e-01 1.08244807e-01 -6.59641325e-01 -6.17994785e-01 -6.85752451e-01 -5.70652366e-01 1.29405662e-01 8.04948509e-01 5.65488160e-01 -4.36402023e-01 6.26312852e-01 -8.55935693e-01 3.01442444e-01 8.87586713e-01 1.62505754e-03 1.22553241e+00 -8.61940503e-01 -4.76082057e-01 -2.10400969e-02 -8.68359327e-01 -8.65451872e-01 -3.35785806e-01 -5.81581771e-01 -1.51263446e-01 -1.80507410e+00 5.21722317e-01 -2.20380023e-01 1.54883429e-01 5.04532516e-01 -3.51963043e-01 6.01042092e-01 4.81504619e-01 -1.74923837e-01 -6.57843888e-01 7.43916333e-01 1.79986095e+00 -2.12838039e-01 -2.08831772e-01 -3.88571545e-02 -8.42227221e-01 5.27974725e-01 6.83594584e-01 -9.17943493e-02 -5.87998152e-01 -2.81629294e-01 3.27465385e-01 7.18883276e-02 8.33552659e-01 -1.11240876e+00 -2.42617801e-02 6.66262209e-02 6.40666723e-01 -5.15073836e-01 6.44137204e-01 -6.67778552e-01 8.56030464e-01 6.17271841e-01 -8.22112635e-02 1.07505947e-01 1.34732679e-01 6.58645868e-01 6.44838437e-02 4.34689522e-01 8.63946438e-01 -8.47215727e-02 -4.53903079e-01 5.75330138e-01 5.19729972e-01 1.17051952e-01 1.38565540e+00 -1.76664032e-02 -5.23631036e-01 -9.03141081e-01 -8.32488894e-01 4.58316386e-01 4.56047237e-01 5.28470457e-01 7.85379946e-01 -1.73462057e+00 -9.46075439e-01 2.55429685e-01 2.27166370e-01 8.57189521e-02 8.78460169e-01 9.09584880e-01 -6.39140964e-01 1.08632609e-01 -4.46597040e-01 -3.74117732e-01 -1.43964005e+00 4.44748580e-01 3.94809097e-01 -3.37612718e-01 -7.95643508e-01 1.00751460e+00 6.16665542e-01 -4.90057349e-01 3.83231118e-02 1.56267151e-01 -6.91291839e-02 -3.23626399e-01 2.44110778e-01 4.06148583e-01 -2.79843271e-01 -1.05530989e+00 -4.21856552e-01 6.51892066e-01 -2.42268741e-02 5.37588261e-02 1.09616649e+00 -1.98839620e-01 -1.59110781e-02 -2.84938753e-01 7.04106092e-01 2.20414907e-01 -1.26184130e+00 1.28115907e-01 -9.09940839e-01 -7.47392058e-01 -4.78531390e-01 -8.96885991e-01 -1.57502317e+00 5.42260051e-01 6.31371498e-01 -6.82621524e-02 1.21151757e+00 5.97859360e-02 9.97484386e-01 -2.90105104e-01 3.27276796e-01 -7.83405066e-01 3.58408958e-01 -2.35325173e-02 1.39120138e+00 -8.65696728e-01 1.43009156e-01 -1.03324425e+00 -8.93921494e-01 7.08654284e-01 1.10322583e+00 -6.55554533e-02 3.93828869e-01 -2.37036735e-01 -7.87250027e-02 -3.68292272e-01 -5.07130958e-02 -2.92877078e-01 7.31558740e-01 8.40266764e-01 3.20285000e-02 3.25145692e-01 -2.58650303e-01 6.94380879e-01 -3.87977242e-01 -2.02736869e-01 2.62643903e-01 5.29422820e-01 1.82203844e-01 -1.05216157e+00 -5.42085171e-01 3.18112046e-01 5.91409728e-02 2.09023744e-01 -7.08839297e-01 7.08130300e-01 4.64644581e-01 9.18550670e-01 -3.43334675e-01 -8.35936666e-01 6.31010950e-01 -6.54859960e-01 6.15586936e-01 -4.15653199e-01 -3.98276091e-01 1.61483943e-01 2.65595734e-01 -6.11203730e-01 -1.81274116e-01 -4.68522459e-01 -1.01283026e+00 -4.71492976e-01 -1.83334827e-01 -4.08995561e-02 1.65486783e-01 5.75605154e-01 4.36337918e-01 7.78865755e-01 5.13155460e-01 -8.53130519e-01 -7.52648637e-02 -9.26249921e-01 -6.61136210e-01 9.71131861e-01 -2.25966170e-01 -8.88625383e-01 5.53517416e-02 1.76017463e-01]
[11.981534004211426, -0.8177694082260132]
7002be68-9ba8-45c1-9908-0d6c4222c9b1
auxiliary-tasks-in-multi-task-learning
1805.06334
null
http://arxiv.org/abs/1805.06334v2
http://arxiv.org/pdf/1805.06334v2.pdf
Auxiliary Tasks in Multi-task Learning
Multi-task convolutional neural networks (CNNs) have shown impressive results for certain combinations of tasks, such as single-image depth estimation (SIDE) and semantic segmentation. This is achieved by pushing the network towards learning a robust representation that generalizes well to different atomic tasks. We extend this concept by adding auxiliary tasks, which are of minor relevance for the application, to the set of learned tasks. As a kind of additional regularization, they are expected to boost the performance of the ultimately desired main tasks. To study the proposed approach, we picked vision-based road scene understanding (RSU) as an exemplary application. Since multi-task learning requires specialized datasets, particularly when using extensive sets of tasks, we provide a multi-modal dataset for multi-task RSU, called synMT. More than 2.5 $\cdot$ 10^5 synthetic images, annotated with 21 different labels, were acquired from the video game Grand Theft Auto V (GTA V). Our proposed deep multi-task CNN architecture was trained on various combination of tasks using synMT. The experiments confirmed that auxiliary tasks can indeed boost network performance, both in terms of final results and training time.
['Marco Körner', 'Lukas Liebel']
2018-05-16
null
null
null
null
['road-scene-understanding']
['computer-vision']
[ 4.37388271e-01 2.01076537e-01 7.48419017e-02 -5.12593091e-01 -8.30039144e-01 -2.89357394e-01 7.37455964e-01 -2.50896007e-01 -5.69495320e-01 9.27761376e-01 -2.05668107e-01 -9.98300407e-03 -1.51686445e-01 -8.84297550e-01 -1.16156816e+00 -4.99029726e-01 1.13714367e-01 4.68949080e-01 5.75317204e-01 -4.63234067e-01 2.69986272e-01 1.95552990e-01 -1.76196539e+00 4.73297179e-01 9.69434023e-01 1.20629728e+00 6.51330233e-01 5.81713915e-01 -1.06209539e-01 8.61476004e-01 -4.76240605e-01 -5.04824936e-01 3.48371953e-01 -2.40157954e-02 -9.75964129e-01 6.31642491e-02 5.03031909e-01 -1.38704345e-01 1.02546057e-02 7.33881652e-01 3.98806542e-01 3.00333261e-01 6.88136578e-01 -1.27391398e+00 -2.09200203e-01 2.43960619e-01 -5.56024432e-01 -5.23023196e-02 2.78849248e-02 2.77582426e-02 9.95607257e-01 -7.02438831e-01 5.33575654e-01 1.22310376e+00 6.75938606e-01 4.94707584e-01 -8.16530526e-01 -5.05240619e-01 6.03789426e-02 2.22100630e-01 -1.17509985e+00 -2.61254877e-01 6.91001058e-01 -4.76012796e-01 7.99081087e-01 -9.73792002e-02 2.61473656e-01 1.25777471e+00 1.49904639e-01 1.01055169e+00 1.32660723e+00 -1.81480855e-01 1.82039976e-01 1.34276852e-01 1.06260963e-01 7.35629618e-01 1.51812300e-01 -6.89787641e-02 -3.13287318e-01 4.47478801e-01 8.24205816e-01 -2.38872185e-01 -1.10575855e-01 -3.02857876e-01 -1.12753391e+00 7.96142519e-01 4.65545833e-01 2.13491097e-01 -2.05842718e-01 3.89565468e-01 6.99160576e-01 1.77597299e-01 6.04449570e-01 9.03017521e-02 -7.22673655e-01 -3.60069424e-02 -7.21021593e-01 2.85450637e-01 3.82909030e-01 9.35850263e-01 1.11737442e+00 3.55309784e-01 -1.56392798e-01 1.06029022e+00 6.63749932e-04 2.78775990e-01 4.22969252e-01 -9.27530229e-01 5.17795146e-01 5.51734924e-01 1.68060586e-02 -6.66009188e-01 -5.53867817e-01 -5.88653982e-01 -9.44782078e-01 5.01548529e-01 6.57786787e-01 -2.17324123e-01 -1.22249651e+00 1.72154355e+00 9.48111266e-02 8.64193365e-02 3.21641266e-01 9.72598433e-01 1.12048054e+00 5.68084359e-01 1.14198767e-01 1.30196974e-01 1.15977263e+00 -1.23214293e+00 -5.10327339e-01 -5.69308996e-01 7.20111072e-01 -6.29324555e-01 1.22662914e+00 6.51082218e-01 -9.47174668e-01 -9.97748613e-01 -1.01375198e+00 5.17827533e-02 -4.90335256e-01 3.26681525e-01 8.90927672e-01 4.73263383e-01 -1.00582731e+00 5.24166703e-01 -3.47817212e-01 -3.26173007e-01 6.90999329e-01 1.71122804e-01 -4.77865249e-01 -2.92304993e-01 -1.29896009e+00 8.71959388e-01 5.97705722e-01 2.51370609e-01 -1.47648978e+00 -3.43628556e-01 -8.53819430e-01 -1.43574193e-01 6.28536582e-01 -6.60208821e-01 1.03306818e+00 -1.04868591e+00 -1.28652525e+00 1.06677186e+00 1.56988353e-01 -4.48266536e-01 6.50544345e-01 -4.16565299e-01 -1.20222479e-01 3.58837470e-02 3.12942058e-01 8.98995399e-01 1.08668482e+00 -1.36229205e+00 -7.86156595e-01 -3.36031705e-01 4.03550982e-01 2.07339108e-01 -7.98344761e-02 -2.95562953e-01 -3.92341852e-01 -3.82406861e-01 -1.38593182e-01 -8.48959684e-01 -3.10742855e-01 -2.58153856e-01 -4.05945271e-01 -2.06434295e-01 9.40974355e-01 -4.38033342e-01 5.94444573e-01 -1.90689731e+00 3.34172130e-01 -3.01197439e-01 1.27703294e-01 2.98828155e-01 -9.27423164e-02 1.70091420e-01 -4.79392298e-02 -5.55266701e-02 -4.64513898e-01 -7.21408308e-01 -2.55897671e-01 5.27502120e-01 1.15297757e-01 3.05125564e-01 1.75287515e-01 1.11568916e+00 -5.62509477e-01 -4.94310439e-01 4.05139267e-01 2.16885701e-01 -1.55470476e-01 1.23720042e-01 -5.53028464e-01 8.03944349e-01 -5.02217233e-01 5.55735290e-01 6.75271511e-01 -8.12923983e-02 -3.04246455e-01 -3.12528372e-01 3.24666756e-03 -2.50294149e-01 -1.07707381e+00 2.10329032e+00 -8.45306873e-01 6.38405204e-01 1.04107343e-01 -1.50753307e+00 1.12597430e+00 4.91426624e-02 4.42829847e-01 -8.83619785e-01 1.88722894e-01 2.54497766e-01 -2.76665371e-02 -6.87260687e-01 6.07119322e-01 7.10964389e-03 -2.00919092e-01 2.41173163e-01 2.68457204e-01 -4.38252151e-01 3.06678385e-01 1.89754944e-02 7.40732193e-01 4.87687051e-01 1.30322590e-01 -4.13380742e-01 6.62103832e-01 1.74345091e-01 4.49473798e-01 5.97120285e-01 -2.92162985e-01 7.15813279e-01 4.61840361e-01 -6.11822963e-01 -1.16347539e+00 -7.91201830e-01 -1.20408133e-01 1.15557325e+00 2.31289506e-01 7.01489598e-02 -6.52373612e-01 -7.57619798e-01 -2.31684849e-01 4.40568149e-01 -6.84371710e-01 3.50413136e-02 -6.13925517e-01 -9.62888062e-01 6.56895518e-01 5.70707858e-01 9.97755408e-01 -1.28159678e+00 -7.54478037e-01 7.32691512e-02 -1.86813921e-01 -1.64231455e+00 6.98746741e-02 3.76711279e-01 -8.08252275e-01 -1.27698874e+00 -9.69148517e-01 -8.49960446e-01 2.70333588e-01 2.11342111e-01 1.06471813e+00 -2.11843595e-01 -1.65868551e-01 2.95233428e-01 -4.89731729e-01 -3.45057428e-01 -2.57286817e-01 2.06882671e-01 -2.08351061e-01 2.62273014e-01 -1.40904889e-01 -6.41799569e-01 -4.97584075e-01 3.30938607e-01 -1.01231480e+00 3.49926472e-01 8.11758757e-01 8.48082542e-01 4.97953892e-01 -1.14007294e-01 8.54152143e-01 -9.30252194e-01 4.00324911e-01 -2.36252338e-01 -4.78326112e-01 1.29866526e-01 -2.31586754e-01 6.54694065e-02 7.18438923e-01 -1.76761329e-01 -1.36586440e+00 1.28128767e-01 -4.01438892e-01 -4.24586296e-01 -4.00372684e-01 3.30386668e-01 -1.46976247e-01 -2.23850667e-01 7.85969436e-01 3.43729883e-01 -1.34927496e-01 -2.65553623e-01 4.54986811e-01 5.21554530e-01 4.94343460e-01 -6.85177147e-01 5.32165706e-01 5.47813177e-01 2.76639134e-01 -9.49211359e-01 -1.05618691e+00 -2.89994568e-01 -6.87787354e-01 -4.33492780e-01 1.18122447e+00 -1.07902324e+00 -6.88149810e-01 8.48779321e-01 -1.15272331e+00 -7.30135739e-01 -1.08697908e-02 2.97514021e-01 -8.70029688e-01 3.45172524e-01 -3.17835838e-01 -6.86773002e-01 -2.35499009e-01 -1.45917547e+00 1.19644892e+00 1.10314988e-01 2.10741967e-01 -1.22986794e+00 -2.02749342e-01 7.59484291e-01 3.67936850e-01 4.92217958e-01 8.84482026e-01 -5.89474022e-01 -7.29283571e-01 9.70817581e-02 -5.11483788e-01 6.85025036e-01 -1.37355905e-02 -3.20124716e-01 -1.42153502e+00 -9.90154892e-02 -9.25827771e-02 -8.39646339e-01 1.26014423e+00 6.68824434e-01 1.34060693e+00 2.81743437e-01 -1.76944852e-01 7.07040191e-01 1.54713190e+00 6.93070441e-02 8.91458035e-01 4.84798789e-01 1.06833076e+00 7.79087842e-01 8.68164837e-01 2.45544776e-01 3.99007380e-01 7.99230635e-01 9.55214858e-01 -2.47773588e-01 -2.14037612e-01 6.95897713e-02 2.27160394e-01 4.22745138e-01 -3.43625158e-01 -2.05837950e-01 -9.01302099e-01 6.69066429e-01 -1.92741382e+00 -5.40159881e-01 -4.53756809e-01 1.91038024e+00 3.93084854e-01 4.61832017e-01 8.17718264e-03 2.22492382e-01 5.47016680e-01 4.51822311e-01 -4.51599956e-01 -4.83671546e-01 -3.38481933e-01 3.32008570e-01 5.41318119e-01 2.65275806e-01 -1.28070784e+00 1.27437615e+00 5.32699060e+00 1.22722554e+00 -1.05025351e+00 3.39121521e-01 7.59033918e-01 2.85581648e-01 -1.20821223e-01 -3.14009041e-01 -6.85805500e-01 1.58577845e-01 6.37992442e-01 2.22236723e-01 -2.31779777e-02 9.51720536e-01 1.97794765e-01 -4.18675780e-01 -8.52516532e-01 1.01649928e+00 3.73714566e-02 -1.30785418e+00 -2.44476050e-02 -1.51431531e-01 1.01787138e+00 5.39757460e-02 7.31964111e-02 5.25364161e-01 2.34803125e-01 -1.12008762e+00 6.46011770e-01 3.03974867e-01 9.82108593e-01 -6.52717888e-01 8.72594714e-01 4.48256791e-01 -1.25736201e+00 -1.97386339e-01 -2.83647716e-01 -2.30090529e-01 3.47192675e-01 5.60885251e-01 -5.38336575e-01 9.45933938e-01 7.70197213e-01 9.00834203e-01 -4.54911739e-01 6.94083571e-01 -2.26818383e-01 1.42745659e-01 -1.53893843e-01 1.55399486e-01 6.00923061e-01 -3.06221873e-01 3.19749743e-01 1.08212149e+00 1.72017381e-01 -1.95204392e-01 9.40857753e-02 7.32096076e-01 -8.88881646e-03 7.08590448e-02 -9.04982269e-01 3.20340663e-01 -4.30146046e-02 1.51296079e+00 -7.87947834e-01 -3.63407046e-01 -4.21126127e-01 1.02143395e+00 3.88012439e-01 3.81087631e-01 -8.87592733e-01 -1.74291998e-01 6.47568762e-01 -7.43279085e-02 2.05959365e-01 -3.08813661e-01 -4.52514559e-01 -9.58651900e-01 1.80460867e-02 -5.38439810e-01 1.49089113e-01 -8.55026424e-01 -1.01200831e+00 7.60733247e-01 8.27402174e-02 -1.19518507e+00 -1.96945414e-01 -8.00075471e-01 -6.35424554e-01 6.62462234e-01 -1.84999573e+00 -1.60069346e+00 -7.15014100e-01 7.40193903e-01 1.06971169e+00 -2.88425654e-01 5.01987338e-01 4.81577933e-01 -6.02982819e-01 3.72342646e-01 -2.43588626e-01 -1.82721600e-01 6.43331230e-01 -1.14059198e+00 2.09469572e-01 7.58367538e-01 -1.02053605e-01 -1.91180810e-01 4.94518906e-01 -4.42623496e-01 -9.96148825e-01 -1.39236438e+00 2.18620270e-01 -3.88111979e-01 4.62288141e-01 -3.16421837e-01 -7.37526238e-01 7.27351487e-01 4.59188707e-02 2.11482029e-02 5.85213490e-02 -1.56507343e-01 1.22269653e-01 -1.34159282e-01 -9.53545868e-01 3.88463378e-01 1.22368431e+00 -4.48029041e-01 -3.45227271e-01 3.97298455e-01 7.66738415e-01 -4.89176840e-01 -7.95238793e-01 7.73379505e-01 2.94302016e-01 -1.26907265e+00 1.09580100e+00 -4.19895709e-01 9.17038739e-01 -1.05989479e-01 -3.64157021e-01 -1.31723845e+00 2.64252514e-01 -1.21402636e-01 2.14137748e-01 9.20170307e-01 4.11272824e-01 -5.03050983e-01 9.00346160e-01 4.67262901e-02 -8.35997641e-01 -8.82959545e-01 -9.39335465e-01 -6.96638823e-01 1.67109683e-01 -8.35163832e-01 1.96024999e-01 8.36840689e-01 -6.36893392e-01 4.70453173e-01 -7.72734106e-01 -9.04679745e-02 6.85534656e-01 -6.97788671e-02 9.67808127e-01 -1.26032281e+00 -1.12784050e-01 -2.87462890e-01 -3.67345899e-01 -1.12111211e+00 3.11029375e-01 -7.04968393e-01 -4.46283966e-02 -1.70728374e+00 -1.27567247e-01 -6.22038364e-01 4.10948917e-02 4.22995687e-01 -1.04610249e-01 4.01660442e-01 2.13768263e-03 4.99456786e-02 -6.54272437e-01 8.04166794e-01 1.68141413e+00 -1.35962725e-01 -4.59997319e-02 3.55258971e-01 -4.95726645e-01 8.52125585e-01 7.76249528e-01 -6.17987663e-02 -5.83512187e-01 -5.63088894e-01 1.74717501e-01 1.73470452e-01 4.88037765e-01 -1.22420835e+00 -1.63953949e-03 -6.72998652e-02 8.96204114e-02 -5.58311701e-01 7.94016600e-01 -7.41098166e-01 7.93137588e-04 4.41504121e-01 -1.92702338e-01 -2.75512844e-01 3.10374022e-01 3.49925846e-01 -4.92966771e-01 -3.68866980e-01 8.49323452e-01 -4.33270365e-01 -1.48079860e+00 3.83309662e-01 -6.16704673e-02 3.43594961e-02 1.21864307e+00 -5.75184226e-01 -3.22691351e-01 -3.48597288e-01 -9.15235460e-01 3.37648362e-01 1.20382622e-01 4.03491110e-01 6.44263029e-01 -1.00418651e+00 -7.13177025e-01 -1.92397777e-02 2.35817686e-01 5.12396097e-01 5.64713180e-01 7.12688029e-01 -5.13354540e-01 5.22264600e-01 -6.87672436e-01 -8.40779722e-01 -9.23955798e-01 2.49193743e-01 4.27348226e-01 -2.96352327e-01 -5.80216348e-01 8.88929725e-01 5.98437726e-01 -5.58052480e-01 1.85267534e-02 -2.85150439e-01 -5.26616693e-01 2.08506927e-01 8.21552128e-02 3.01626980e-01 1.87364146e-01 -5.15511572e-01 -1.31036192e-01 8.46876025e-01 2.46110559e-02 6.02123179e-02 1.43719506e+00 -2.26757720e-01 2.14405060e-01 3.72077882e-01 1.17752004e+00 -5.80716312e-01 -1.59323049e+00 -2.07699016e-01 -1.18025810e-01 -2.57403731e-01 -2.43992731e-02 -6.39737904e-01 -1.20147324e+00 1.11721969e+00 5.35234272e-01 -5.58017530e-02 1.10483885e+00 -1.29485307e-02 7.53945589e-01 3.41770947e-01 6.90493643e-01 -9.89303112e-01 5.31423628e-01 6.78016186e-01 7.64302552e-01 -1.85200810e+00 -3.37742954e-01 -4.96458352e-01 -9.75574076e-01 1.04819214e+00 9.88781989e-01 -1.51107430e-01 3.65030497e-01 3.40814739e-02 4.94647659e-02 -3.53430808e-01 -4.55996782e-01 -6.32787526e-01 1.97468437e-02 8.11754644e-01 1.73648357e-01 -1.53456971e-01 -1.20112807e-01 4.42419827e-01 -1.65172294e-02 6.41818196e-02 7.31565118e-01 6.48332775e-01 -5.08231997e-01 -8.99571002e-01 -9.89984944e-02 3.64983618e-01 -3.90307099e-01 7.11944178e-02 1.05516717e-01 1.04799902e+00 4.04094726e-01 7.16137648e-01 -9.44982469e-02 -3.73500794e-01 4.04100209e-01 -1.91025436e-01 3.97313982e-01 -5.20992517e-01 -3.62105638e-01 -2.94783920e-01 4.56448704e-01 -5.08367538e-01 -6.95881665e-01 -3.71815324e-01 -9.48300779e-01 -1.42842785e-01 -3.00088283e-02 -1.67342097e-01 7.39715278e-01 1.31867564e+00 -1.30370095e-01 9.93951619e-01 4.91727769e-01 -1.08041513e+00 -5.64583316e-02 -1.21428406e+00 -5.41355193e-01 3.91297460e-01 1.66437775e-01 -1.00246012e+00 -9.15117562e-02 2.10983939e-02]
[9.518941879272461, 1.2689512968063354]
49d6fda3-2333-41cc-ba84-297e7a57b07b
prediction-of-bottleneck-points-for
1911.04676
null
https://arxiv.org/abs/1911.04676v1
https://arxiv.org/pdf/1911.04676v1.pdf
Prediction of Bottleneck Points for Manipulation Planning in Cluttered Environment using a 3D Convolutional Neural Network
Latest research in industrial robotics is aimed at making human robot collaboration possible seamlessly. For this purpose, industrial robots are expected to work on the fly in unstructured and cluttered environments and hence the subject of perception driven motion planning plays a vital role. Sampling based motion planners are proven to be the most effective for such high dimensional planning problems with real time constraints. Unluckily random stochastic samplers suffer from the phenomenon of 'narrow passages' or bottleneck regions which need targeted sampling to improve their convergence rate. Also identifying these bottleneck regions in a diverse set of planning problems is a challenge. In this paper an attempt has been made to address these two problems by designing an intelligent 'bottleneck guided' heuristic for a Rapidly Exploring Random Tree Star (RRT*) planner which is based on relevant context extracted from the planning scenario using a 3D Convolutional Neural Network and it is also proven that the proposed technique generalises to unseen problem instances. This paper benchmarks the technique (bottleneck guided RRT*) against a 10% Goal biased RRT star planner, shows significant improvement in planning time and memory requirement and uses ABB 1410 industrial manipulator as a platform for implantation and validation of the results.
['B. K. Rout', 'Indraneel Patil', 'V. Kalaichelvi']
2019-11-12
null
null
null
null
['industrial-robots']
['robots']
[ 5.59382737e-01 3.76938045e-01 2.10390121e-01 -9.90416482e-02 -5.23161352e-01 -3.75611871e-01 6.54641449e-01 5.32136038e-02 -2.75736749e-01 9.04687166e-01 -1.49067253e-01 -4.15405542e-01 -6.25411570e-01 -6.62597954e-01 -4.14081573e-01 -7.11257935e-01 -3.77262741e-01 1.20356452e+00 4.64486390e-01 -5.10381401e-01 8.17300141e-01 8.06907535e-01 -1.34141290e+00 -3.02715506e-02 3.35211068e-01 5.17428994e-01 1.25263786e+00 8.99459839e-01 1.53995484e-01 5.52395999e-01 -5.41927457e-01 6.65618539e-01 5.73731303e-01 -2.77977735e-01 -1.30908787e+00 1.29896179e-01 -1.28253222e-01 4.69679758e-02 1.73920617e-01 7.96894968e-01 6.45183444e-01 4.94557381e-01 7.25634217e-01 -1.13838530e+00 4.25915658e-01 4.63948935e-01 -3.08332890e-01 1.69047579e-01 4.27862287e-01 4.26657796e-01 5.59034348e-01 -6.73795760e-01 1.04017353e+00 1.32583463e+00 3.31693143e-01 3.47507924e-01 -8.82687032e-01 -4.02958430e-02 -2.12410778e-01 1.44948319e-01 -9.79623377e-01 3.78253451e-03 7.65092313e-01 -2.60690242e-01 1.26584089e+00 -1.62428920e-03 4.47152674e-01 1.03359580e+00 7.40267575e-01 5.74296176e-01 8.44759285e-01 -3.67549330e-01 5.65736115e-01 -4.80726622e-02 -2.89334387e-01 5.43255746e-01 2.07931265e-01 3.46148640e-01 -6.69802576e-02 3.13247412e-01 1.08622003e+00 -3.36869359e-01 -2.07982324e-02 -8.40866923e-01 -1.60686982e+00 8.01484466e-01 5.63042283e-01 3.40821266e-01 -7.74559021e-01 1.17727540e-01 6.12439811e-01 2.25675717e-01 -4.39830959e-01 9.23736334e-01 -5.40706396e-01 -2.35029146e-01 -5.23111045e-01 7.16931045e-01 9.36602950e-01 1.35337746e+00 4.39761251e-01 8.54462311e-02 2.82600168e-02 5.01895607e-01 3.72649014e-01 1.61302928e-02 3.60187113e-01 -1.08276391e+00 6.05565906e-01 4.27478820e-01 4.26729470e-01 -7.85361707e-01 -8.51007640e-01 -2.55376399e-01 -7.35044599e-01 8.90631497e-01 4.98907089e-01 -2.27064833e-01 -1.13091171e+00 1.04128039e+00 5.81165850e-01 -5.32397687e-01 2.65288621e-01 1.03930807e+00 1.15865111e-01 7.24931479e-01 -1.44227922e-01 -1.16982199e-01 1.10733056e+00 -1.17024171e+00 -5.37307918e-01 -1.33934408e-01 5.18721879e-01 -9.12313223e-01 8.17204058e-01 8.29099000e-01 -7.23818481e-01 -8.01418841e-01 -1.24948204e+00 3.83082360e-01 -4.20440733e-01 -1.89198062e-01 4.66064423e-01 3.64540726e-01 -7.47723579e-01 1.00648570e+00 -7.43061841e-01 -7.25279212e-01 4.10633862e-01 5.74333727e-01 -3.20582956e-01 -3.64215106e-01 -7.33758867e-01 1.41149664e+00 8.34629059e-01 3.02492946e-01 -1.27703261e+00 1.26008645e-01 -5.52543044e-01 -4.61504489e-01 6.10400498e-01 -3.61777455e-01 1.44930947e+00 -3.56438160e-01 -1.79454160e+00 3.18936497e-01 2.11716130e-01 -6.02238953e-01 7.27875948e-01 -2.45912135e-01 1.46624088e-01 6.01861700e-02 1.78174913e-01 7.02386558e-01 7.18326330e-01 -1.17842138e+00 -9.37079608e-01 -2.28832945e-01 -5.67306345e-03 4.76441264e-01 7.10716903e-01 -4.49818164e-01 1.86885387e-01 -7.67994821e-02 2.46979401e-01 -1.19028854e+00 -1.01524889e+00 -5.56746542e-01 -4.62794721e-01 -5.25522351e-01 1.10678566e+00 -2.33616441e-01 5.80241919e-01 -1.51311100e+00 1.99018225e-01 1.06194578e-01 -3.10179740e-01 2.42034718e-01 -1.86004713e-02 8.14192176e-01 1.63264215e-01 -4.52776313e-01 -2.11460367e-01 3.23498160e-01 3.20829116e-02 3.33533317e-01 5.72627857e-02 4.19233620e-01 2.46125549e-01 5.07801175e-01 -1.07975459e+00 -4.98011410e-01 8.70351434e-01 2.06521023e-02 -1.69813633e-01 7.33667985e-02 -5.38905978e-01 7.31999338e-01 -7.76485205e-01 6.04014695e-01 4.14491594e-01 2.74778575e-01 1.00324422e-01 2.75802076e-01 -4.91117090e-01 1.77032530e-01 -1.30132067e+00 2.04203701e+00 -5.78323960e-01 6.08578086e-01 1.96694985e-01 -1.14410830e+00 1.27022624e+00 5.30346274e-01 4.42625880e-01 -4.74199474e-01 2.77426273e-01 4.91279155e-01 -9.89165064e-03 -5.99187672e-01 6.66238785e-01 -7.34741688e-02 -6.32319823e-02 9.41377357e-02 -2.68009752e-01 -5.31966507e-01 7.09796399e-02 -4.04236615e-01 1.45136511e+00 7.41239905e-01 5.92714489e-01 -4.20559704e-01 7.18067229e-01 8.66527438e-01 2.18715832e-01 6.27555907e-01 -4.88516033e-01 5.65370023e-01 1.18437529e-01 -8.77765775e-01 -1.13778448e+00 -8.84513855e-01 1.67213440e-01 4.71439749e-01 5.91754735e-01 2.94485033e-01 -4.55330104e-01 -6.33509576e-01 -4.45555687e-01 1.08851600e+00 -1.73489884e-01 2.23704502e-01 -1.02654326e+00 -4.16709810e-01 -2.17556313e-01 2.23695174e-01 4.97041047e-01 -1.79785740e+00 -1.76293874e+00 8.63049805e-01 1.92240521e-01 -1.00312471e+00 1.40308514e-01 7.62716413e-01 -1.10547614e+00 -1.23945546e+00 -6.08704627e-01 -1.05972445e+00 5.37969708e-01 3.23406428e-01 1.00981438e+00 -4.90688443e-01 -5.89244604e-01 9.28513985e-03 -6.11676037e-01 -8.23750794e-01 -6.53945148e-01 4.24674422e-01 -2.10642278e-01 -9.64042306e-01 7.26187900e-02 -5.36130011e-01 -5.05882263e-01 4.86691862e-01 -5.02663910e-01 3.87264118e-02 1.00837541e+00 8.19424689e-01 2.81607807e-01 6.18668318e-01 7.13111758e-01 -6.26340985e-01 7.89865255e-01 -4.39869732e-01 -7.96587527e-01 -2.54764944e-01 -6.22073591e-01 1.53286919e-01 6.90028846e-01 -2.65531272e-01 -8.77959907e-01 5.92481375e-01 -3.82084842e-03 -1.07416071e-01 -6.31363690e-01 1.28379107e-01 -1.51365966e-01 2.23474279e-01 6.78176820e-01 -4.94730957e-02 -4.83314730e-02 -2.33980209e-01 1.65918186e-01 5.56872308e-01 3.48678112e-01 -4.26041901e-01 6.38626456e-01 2.20142826e-01 5.54182291e-01 -9.01325226e-01 -4.09669094e-02 -7.65922904e-01 -7.97665954e-01 -4.72583234e-01 8.96257758e-01 -2.08884269e-01 -5.57748139e-01 2.04482935e-02 -1.47085786e+00 -3.87651324e-01 -4.94522482e-01 4.11011994e-01 -1.26341486e+00 1.64295044e-02 -1.00178927e-01 -1.20149970e+00 -3.06848258e-01 -1.48608017e+00 1.03982532e+00 2.25879118e-01 -4.59704459e-01 -5.88503778e-01 -5.64647838e-02 5.68146892e-02 3.47587943e-01 7.66240895e-01 7.21554399e-01 -6.20133460e-01 -9.58172858e-01 -4.04663116e-01 9.89446789e-02 -1.72851250e-01 2.49651030e-01 -5.98425925e-01 -5.36432803e-01 4.13508620e-03 1.52944058e-01 -1.66297495e-01 1.02965020e-01 5.11646748e-01 3.80740970e-01 -2.21029837e-02 -6.67784512e-01 -1.40615836e-01 1.70668411e+00 7.64253974e-01 5.65645456e-01 8.31392705e-01 3.30067247e-01 7.56148100e-01 1.45795560e+00 3.90747011e-01 -2.82509178e-01 5.95873594e-01 1.06105995e+00 4.51628119e-01 2.80909866e-01 1.31695913e-02 1.39186725e-01 3.63986850e-01 -2.31246620e-01 -2.38188937e-01 -9.89623666e-01 9.47548509e-01 -1.83534443e+00 -8.56181562e-01 -2.62696534e-01 2.13004851e+00 2.01008514e-01 5.94959259e-01 3.41926552e-02 4.76114094e-01 6.78908885e-01 -1.43303126e-01 -4.08230096e-01 -8.99947405e-01 5.24873137e-01 2.12207392e-01 7.24443078e-01 5.54595172e-01 -1.04767239e+00 9.36923206e-01 4.74195766e+00 5.92147231e-01 -6.80637360e-01 -2.81876594e-01 -1.60224177e-02 2.20959902e-01 5.51253259e-01 -1.97119210e-02 -8.02748501e-01 1.27439871e-01 9.69757140e-01 1.81410268e-01 4.39933002e-01 1.19293177e+00 5.08460879e-01 -6.01312876e-01 -1.12169266e+00 5.88810086e-01 -5.30301452e-01 -1.28583634e+00 -2.29236543e-01 2.04011425e-01 5.87392151e-01 -2.71827113e-02 -2.74218887e-01 2.18124524e-01 6.59648478e-01 -1.25308037e+00 6.96089804e-01 2.67817110e-01 -7.09637552e-02 -1.03608656e+00 1.03256214e+00 7.69008398e-01 -1.13951850e+00 -2.79798687e-01 -5.50079107e-01 -8.25171173e-02 5.82079351e-01 2.79630840e-01 -1.90710473e+00 7.97725677e-01 4.56603438e-01 1.66889817e-01 1.35531276e-01 1.18333566e+00 1.02379195e-01 -8.83606151e-02 -3.97591621e-01 -6.27625048e-01 9.09997404e-01 -7.69934133e-02 1.02660537e+00 9.63459909e-01 3.44868213e-01 -4.26933736e-01 3.15578520e-01 7.15212643e-01 8.72127831e-01 -8.97341892e-02 -1.08331990e+00 1.51172355e-01 2.10203394e-01 9.87753987e-01 -1.18475950e+00 1.58741742e-01 2.58388132e-01 9.02914584e-01 -1.20001577e-01 -1.54293299e-01 -5.96143365e-01 -5.89374185e-01 3.47004712e-01 1.42597277e-02 5.36855519e-01 -7.08254576e-01 -2.09204882e-01 1.31222233e-01 -2.61952907e-01 -6.64103687e-01 -5.90684451e-02 -7.83060730e-01 -7.64528632e-01 6.56661451e-01 1.51469290e-01 -1.37845445e+00 -6.11168146e-01 -9.81585681e-01 -3.38461876e-01 8.96007955e-01 -1.30754995e+00 -7.86800385e-01 -3.18887353e-01 2.45563701e-01 1.61690700e+00 -2.12649912e-01 8.62465322e-01 -3.65502685e-01 -7.21819550e-02 -5.18080175e-01 -1.17069639e-01 -4.06446159e-01 1.35694519e-01 -1.36369836e+00 3.10627311e-01 4.03753370e-01 -2.31463507e-01 3.04976821e-01 1.21670079e+00 -8.28691721e-01 -1.61870432e+00 -1.02996886e+00 5.30624807e-01 -3.42679918e-01 1.74687207e-01 2.90175490e-02 -3.08694899e-01 3.36035579e-01 3.92364562e-01 -4.01791900e-01 -3.16505939e-01 -3.68282586e-01 8.09348106e-01 2.76999325e-01 -1.31755018e+00 7.56717563e-01 1.04787457e+00 3.85867685e-01 -6.12175047e-01 6.41697109e-01 4.91384834e-01 -5.22084653e-01 -5.09834528e-01 4.60052013e-01 4.78694588e-01 -1.09809566e+00 8.20502102e-01 -2.71247506e-01 3.00691068e-01 -4.05668259e-01 -5.40326908e-02 -1.14788389e+00 -2.43047792e-02 -1.07378542e+00 3.07378590e-01 7.07238495e-01 3.76105100e-01 -2.54546016e-01 1.17688262e+00 5.10173626e-02 -3.73295009e-01 -8.72607768e-01 -1.03679025e+00 -8.53657424e-01 -2.44716778e-01 -3.35885823e-01 2.91970462e-01 4.44986671e-01 7.48180505e-03 5.59868395e-01 -2.51840428e-02 2.81039983e-01 5.69480658e-01 6.49893377e-03 8.82620871e-01 -1.18177950e+00 2.90243089e-01 -2.31736720e-01 -4.88520443e-01 -8.59462500e-01 -3.95165980e-01 -5.05856812e-01 7.82831073e-01 -2.00898981e+00 -7.47445345e-01 -6.93304300e-01 9.31245014e-02 -1.28009751e-01 5.64043999e-01 -5.06581962e-01 1.68762207e-01 4.69293743e-02 -3.43740880e-01 2.85597503e-01 1.60041833e+00 5.21455370e-02 -4.93823439e-01 4.63974208e-01 9.98553857e-02 5.30687630e-01 1.18331385e+00 -2.71057308e-01 -9.09429312e-01 -8.37026048e-04 1.25091951e-02 4.19345170e-01 1.31721511e-01 -1.52151036e+00 3.90165240e-01 -3.39800835e-01 1.46455750e-01 -1.07328570e+00 3.51704031e-01 -1.29010344e+00 2.33021960e-01 8.59299123e-01 -3.03897768e-01 4.52889621e-01 3.16369496e-02 8.48866880e-01 8.02817494e-02 -7.80612767e-01 7.35890806e-01 -8.85709941e-01 -1.07317758e+00 -1.43945307e-01 -6.14118993e-01 -3.67863148e-01 1.45658040e+00 -8.13559651e-01 2.82753080e-01 -9.34477970e-02 -6.75100088e-01 2.43013620e-01 1.63347945e-01 4.39476430e-01 6.73557818e-01 -8.86466086e-01 -4.69026834e-01 2.12571453e-02 -2.24865586e-01 7.78108656e-01 -6.55703619e-02 7.54873574e-01 -1.10644495e+00 8.21256340e-01 -5.91013491e-01 -7.12327242e-01 -1.02641368e+00 6.37317538e-01 1.10479310e-01 -5.91735482e-01 -1.07739174e+00 6.52086616e-01 -4.19712245e-01 -5.13642848e-01 2.45686725e-01 -5.60708642e-01 -3.40730011e-01 -3.11971098e-01 -3.26919816e-02 6.98870540e-01 -6.22474775e-02 -3.74536067e-01 -1.87138721e-01 3.19490612e-01 1.86625347e-01 -1.88241646e-01 1.42467046e+00 -1.24944590e-01 3.95931035e-01 4.68868136e-01 8.32439423e-01 -4.41916883e-01 -1.36951399e+00 4.95759457e-01 6.67398691e-01 -2.13501841e-01 -1.50090516e-01 -7.66112149e-01 -5.22290826e-01 6.97865605e-01 6.79071188e-01 3.90628368e-01 7.23200977e-01 -3.92705142e-01 7.43159354e-01 5.53056538e-01 1.03748405e+00 -1.27341104e+00 7.25112483e-02 6.64934158e-01 1.06971490e+00 -1.18908060e+00 6.23664670e-02 -5.27841687e-01 -6.29772782e-01 1.48293746e+00 3.58970135e-01 -4.59024638e-01 2.41649419e-01 2.88325012e-01 -4.66155596e-02 -3.49318147e-01 -5.92947125e-01 -1.31580949e-01 -4.64841485e-01 1.07454419e+00 2.64192056e-02 -5.26212379e-02 -2.95910627e-01 -2.61310309e-01 -3.70093912e-01 1.45646185e-01 5.31889677e-01 1.43142903e+00 -8.13031495e-01 -9.38594162e-01 -5.12874424e-01 1.34520620e-01 1.03922685e-04 5.20907044e-01 6.45438284e-02 1.41026247e+00 2.09227055e-01 9.56776202e-01 -2.94327825e-01 -1.97855234e-01 4.02971387e-01 -2.75025647e-02 5.53751051e-01 -6.60914004e-01 -5.94542682e-01 -6.87012821e-02 4.17858303e-01 -7.32712150e-01 -3.86470020e-01 -8.68022859e-01 -1.54908693e+00 3.60543460e-01 -3.02028209e-01 8.71730074e-02 1.16387308e+00 9.48016465e-01 -3.73844542e-02 8.33581686e-01 4.82238561e-01 -1.43602228e+00 -8.37112069e-01 -1.00970685e+00 -4.33032095e-01 -1.79913491e-01 1.49016514e-01 -8.45253348e-01 1.50092645e-02 -1.45317033e-01]
[4.852146625518799, 1.3439222574234009]
6ce84307-21bd-4afd-921d-0a3f12201ed5
detection-and-rectification-of-arbitrary
2103.00785
null
https://arxiv.org/abs/2103.00785v1
https://arxiv.org/pdf/2103.00785v1.pdf
Detection and Rectification of Arbitrary Shaped Scene Texts by using Text Keypoints and Links
Detection and recognition of scene texts of arbitrary shapes remain a grand challenge due to the super-rich text shape variation in text line orientations, lengths, curvatures, etc. This paper presents a mask-guided multi-task network that detects and rectifies scene texts of arbitrary shapes reliably. Three types of keypoints are detected which specify the centre line and so the shape of text instances accurately. In addition, four types of keypoint links are detected of which the horizontal links associate the detected keypoints of each text instance and the vertical links predict a pair of landmark points (for each keypoint) along the upper and lower text boundary, respectively. Scene texts can be located and rectified by linking up the associated landmark points (giving localization polygon boxes) and transforming the polygon boxes via thin plate spline, respectively. Extensive experiments over several public datasets show that the use of text keypoints is tolerant to the variation in text orientations, lengths, and curvatures, and it achieves superior scene text detection and rectification performance as compared with state-of-the-art methods.
['Steven Hoi', 'Shijian Lu', 'Chuhui Xue']
2021-03-01
null
null
null
null
['scene-text-detection']
['computer-vision']
[ 1.49776354e-01 -5.72030842e-01 5.73327765e-02 -2.25518540e-01 -6.87612772e-01 -8.56966436e-01 7.15209126e-01 3.18295658e-01 -6.54650033e-02 1.11701049e-01 1.58093646e-01 9.67165083e-02 -1.10829987e-01 -2.87902743e-01 -6.40674353e-01 -6.04117155e-01 3.56884837e-01 8.56196642e-01 5.20545900e-01 -1.40404940e-01 8.55363429e-01 7.84350991e-01 -1.14469743e+00 4.20996130e-01 8.23514819e-01 7.48010516e-01 6.16490617e-02 7.60019958e-01 -4.93859470e-01 -5.12839071e-02 -6.19384468e-01 -3.67672741e-01 3.88894051e-01 3.25448006e-01 -2.87599891e-01 5.12847662e-01 1.06745923e+00 -2.75756240e-01 -2.87137598e-01 9.90305305e-01 3.45291972e-01 5.63348234e-02 1.04041278e+00 -1.16306412e+00 -7.33272672e-01 3.30141127e-01 -1.36568272e+00 -3.50571088e-02 4.46797550e-01 -3.71746033e-01 9.35613573e-01 -1.37345576e+00 5.54902911e-01 1.24547470e+00 1.08753026e+00 5.94022078e-03 -8.04931641e-01 -4.88413423e-01 2.40630046e-01 -9.39431265e-02 -2.02018571e+00 -3.92078251e-01 7.78271556e-01 -4.27759945e-01 9.70238745e-01 5.01725197e-01 3.09124514e-02 6.66145563e-01 3.64135593e-01 8.71528089e-01 3.22069883e-01 -5.61191559e-01 -2.06844404e-01 2.37487853e-01 7.24605396e-02 8.27182293e-01 2.00916708e-01 -6.74302518e-01 -4.58529443e-01 -1.89686909e-01 9.61613536e-01 3.47894698e-01 -2.25313485e-01 -5.59126198e-01 -1.56058300e+00 4.83285278e-01 8.00438970e-02 5.27735829e-01 5.79853319e-02 -1.38086632e-01 2.86471635e-01 -1.52400032e-01 4.54124659e-01 8.25097635e-02 -4.06964898e-01 2.25264072e-01 -1.06864417e+00 3.98830585e-02 6.64740145e-01 1.50215852e+00 5.94016373e-01 -1.25539480e-02 -1.36972338e-01 9.69300330e-01 3.09334427e-01 8.03916633e-01 3.83659571e-01 -1.05490670e-01 1.13015568e+00 1.02832127e+00 7.34328628e-02 -1.52208400e+00 -6.95695341e-01 1.23444043e-01 -9.85103965e-01 -3.02776188e-01 5.78988016e-01 -1.74405739e-01 -7.19963312e-01 7.19514489e-01 4.33020771e-01 8.50683525e-02 -2.42698833e-01 6.58424675e-01 9.32838023e-01 6.53215647e-01 -5.27042329e-01 5.90154156e-02 1.50257826e+00 -9.07508016e-01 -5.37993670e-01 -2.89802969e-01 6.56962276e-01 -1.38431346e+00 7.70333052e-01 2.35984981e-01 -7.54869819e-01 -5.85056067e-01 -8.24589074e-01 -2.23555923e-01 -5.21249115e-01 8.05103421e-01 1.93067104e-01 4.29141581e-01 -9.78370130e-01 3.10500830e-01 -4.98801976e-01 -3.90071094e-01 1.10709630e-01 3.54249805e-01 -8.42478350e-02 2.08141744e-01 -4.19275612e-01 4.88035709e-01 6.76624663e-03 3.57153445e-01 -7.53511861e-02 -4.79123950e-01 -9.72708941e-01 1.40727097e-02 3.66597146e-01 -2.58436233e-01 7.54495978e-01 -6.23332500e-01 -1.16100335e+00 9.10282671e-01 -4.42973316e-01 1.49202347e-01 1.00160158e+00 -2.69888431e-01 -5.00526845e-01 4.50680330e-02 2.77581275e-01 5.35864234e-01 1.46464908e+00 -1.21436620e+00 -9.89288747e-01 -4.62930232e-01 -7.40855455e-01 5.21146357e-01 -3.87978882e-01 2.15500936e-01 -8.27569425e-01 -7.26773441e-01 5.68428278e-01 -6.64537370e-01 1.00293845e-01 1.97546855e-01 -1.05554819e+00 -6.06463552e-01 1.55700052e+00 -6.08091593e-01 9.69727993e-01 -2.23454785e+00 -1.15880840e-01 3.37546945e-01 2.27592543e-01 -1.87943593e-01 1.41944271e-03 4.21443224e-01 1.26065433e-01 1.08602986e-01 2.21077845e-01 -6.06491446e-01 1.77442357e-01 -3.42753291e-01 -5.65081179e-01 9.08332050e-01 -1.01201564e-01 8.50536823e-01 -3.33264738e-01 -6.58352733e-01 6.48278296e-01 5.29064238e-01 7.20592141e-02 -2.17592627e-01 -1.60516635e-01 -1.28054973e-02 -6.67814612e-01 6.87071145e-01 8.23277652e-01 -2.37978578e-01 -2.32055932e-01 -4.58082467e-01 -3.94793838e-01 -2.53571540e-01 -1.81133974e+00 1.21807337e+00 1.43038854e-02 9.34430420e-01 -9.96577442e-02 -3.76659989e-01 1.44967377e+00 2.84052700e-01 4.17023212e-01 -1.45430058e-01 1.31472051e-01 3.84035916e-03 -6.27858043e-01 -5.28238714e-01 1.08312190e+00 6.76829159e-01 -1.09479181e-01 5.01161098e-01 -5.99831522e-01 -5.15842319e-01 9.71707031e-02 1.14929564e-01 5.97324610e-01 7.47368066e-03 1.56182155e-01 -1.20759100e-01 8.31286609e-01 -8.64006355e-02 1.62251011e-01 7.27880597e-01 -7.54220113e-02 8.54408562e-01 2.94688553e-01 -6.91591680e-01 -1.37399352e+00 -7.89659142e-01 -3.74436259e-01 1.23125768e+00 3.66762966e-01 -2.79913366e-01 -4.98597205e-01 -4.77491766e-01 1.40458480e-01 3.63095045e-01 -3.22911650e-01 5.55233121e-01 -8.95377994e-01 -5.06544530e-01 4.64690924e-01 6.49231851e-01 6.06227219e-01 -8.00301313e-01 -2.82392234e-01 -2.21077666e-01 -1.26065418e-01 -1.45362425e+00 -1.22048330e+00 -6.10810742e-02 -6.56532705e-01 -9.97746766e-01 -9.40536141e-01 -1.28389513e+00 1.12032592e+00 6.23666704e-01 5.23841381e-01 1.68488145e-01 -3.77254307e-01 5.00499070e-01 -2.53554195e-01 -1.00897998e-01 -2.26489082e-01 -1.77574484e-03 -7.64426738e-02 2.70165592e-01 3.75177950e-01 7.30624571e-02 -3.31706434e-01 8.23084354e-01 -5.16903579e-01 3.24417651e-02 2.46592239e-01 3.20212692e-01 6.22796774e-01 2.34489352e-01 1.86729822e-02 -4.43614691e-01 5.54795980e-01 5.18415384e-02 -8.45477581e-01 4.88332868e-01 -1.81170166e-01 -3.02688330e-01 5.89712024e-01 -5.52193403e-01 -7.18366563e-01 5.08895218e-01 3.97434771e-01 -3.58877331e-01 -6.33495212e-01 -8.37852806e-02 1.19561777e-01 -2.93980181e-01 5.28976440e-01 4.10132587e-01 -5.55102527e-01 -3.45311284e-01 3.95893931e-01 9.33987796e-01 5.59240758e-01 -2.44790390e-01 1.09403479e+00 6.54811442e-01 8.86287540e-02 -1.45435131e+00 -4.78763580e-01 -9.32317078e-01 -1.47301877e+00 -1.17374770e-01 8.86534631e-01 -6.84607625e-01 -6.43567801e-01 6.13314092e-01 -1.40749836e+00 1.33550698e-02 2.49821275e-01 4.03818935e-02 -2.90878981e-01 8.90361667e-01 -6.30868614e-01 -6.69181287e-01 -6.38981164e-01 -1.00827956e+00 1.75662673e+00 2.34891072e-01 -1.02938138e-01 -1.15782201e+00 -3.19790214e-01 1.80801705e-01 4.08142898e-03 3.15527245e-02 8.67262900e-01 -8.14583898e-01 -6.41790867e-01 -5.65070033e-01 -5.15487432e-01 -4.59106952e-01 1.73546597e-01 6.91393375e-01 -6.94028437e-01 -2.62073457e-01 -1.86902121e-01 3.66107941e-01 3.96387488e-01 7.99296498e-01 9.42747533e-01 -1.10000476e-01 -7.73417771e-01 7.97211111e-01 1.29004645e+00 5.42905852e-02 2.01795399e-01 4.06538427e-01 1.17826092e+00 3.30390245e-01 3.87087822e-01 5.72600603e-01 3.72189850e-01 7.15835631e-01 1.17954887e-01 -1.99722111e-01 1.80817351e-01 -8.87126401e-02 1.02204621e-01 6.19890869e-01 2.79320776e-01 -4.26863551e-01 -1.02958274e+00 3.59644651e-01 -1.82796466e+00 -5.53670406e-01 -9.19136882e-01 2.02529097e+00 2.95675904e-01 9.99416634e-02 5.63222915e-02 1.63022786e-01 1.33156967e+00 1.57812506e-01 -5.69815218e-01 -2.99428374e-01 -3.39736730e-01 -5.80999076e-01 7.87494600e-01 3.34406883e-01 -1.45935822e+00 1.12016499e+00 6.35175896e+00 8.52339029e-01 -9.59664166e-01 -4.29252088e-01 3.30217510e-01 3.61438751e-01 2.08469376e-01 -4.47501928e-01 -1.46769559e+00 1.10078849e-01 -3.91212851e-02 7.79213235e-02 2.74378449e-01 6.97177470e-01 2.20434785e-01 -6.91589713e-02 -9.92002130e-01 1.21834850e+00 6.19890988e-01 -1.26179373e+00 1.24136418e-01 -2.97966212e-01 1.00106633e+00 2.11881716e-02 1.18213318e-01 -2.17123091e-01 2.95589096e-03 -6.85332477e-01 7.27505565e-01 3.57892543e-01 9.40341234e-01 -5.59833407e-01 4.59619522e-01 5.17638624e-01 -1.66971111e+00 9.43960771e-02 -5.12354732e-01 4.66282994e-01 -2.72095725e-02 2.73697257e-01 -1.27254057e+00 3.45014513e-01 5.40556073e-01 8.04754555e-01 -7.97331572e-01 1.18729150e+00 -2.69890398e-01 -2.73335259e-02 -5.01882672e-01 -4.19513911e-01 8.38981047e-02 -3.63078803e-01 6.21783197e-01 1.50529957e+00 3.29371423e-01 -2.31027886e-01 4.53781962e-01 7.91521907e-01 -5.12498319e-02 4.73661005e-01 -5.27439237e-01 3.21868747e-01 5.95159292e-01 1.54700637e+00 -1.32732451e+00 -2.93420821e-01 -4.08900112e-01 1.15619504e+00 -1.16695449e-01 4.32491809e-01 -6.59828305e-01 -6.91628933e-01 4.50827088e-03 -1.86346769e-02 5.48412800e-01 -4.96756226e-01 -8.17571580e-01 -9.91049886e-01 3.34069818e-01 -5.58859050e-01 2.91110575e-01 -9.88162577e-01 -1.23530912e+00 3.60600114e-01 -3.47888559e-01 -1.28761566e+00 8.60891268e-02 -6.91911757e-01 -9.40129220e-01 7.03396380e-01 -1.14111257e+00 -1.48620772e+00 -5.30215144e-01 7.68500090e-01 1.08172309e+00 -8.49592388e-02 5.52110672e-01 -5.57718128e-02 -7.14127839e-01 8.03559363e-01 6.42851651e-01 5.48778534e-01 8.32549930e-01 -1.19732380e+00 8.79430056e-01 7.26967871e-01 2.99171031e-01 2.52365261e-01 5.95049262e-01 -9.60593164e-01 -1.45866930e+00 -1.12302887e+00 8.84627283e-01 -6.30990565e-01 8.40997219e-01 -6.92390740e-01 -1.07840359e+00 8.21014822e-01 -1.64906234e-01 -2.36664981e-01 5.93464449e-02 -1.05919890e-01 -3.07376653e-01 1.85437009e-01 -1.07352424e+00 8.12679708e-01 6.04560792e-01 -3.78103524e-01 -5.45174122e-01 8.14440846e-01 3.88864130e-01 -7.33488977e-01 -4.74624395e-01 7.40717947e-02 3.68586749e-01 -7.02394962e-01 1.06911147e+00 6.66374117e-02 -1.28813675e-02 -3.79035175e-01 1.50016040e-01 -8.28117907e-01 -2.83856422e-01 -6.68650806e-01 2.80286402e-01 1.24595046e+00 4.29601669e-01 -4.59098101e-01 8.64170909e-01 3.62747908e-01 -1.84434533e-01 -3.41978341e-01 -1.07440674e+00 -4.07142013e-01 -5.95111921e-02 -2.42447674e-01 5.29677570e-01 1.00696135e+00 1.40364036e-01 2.84654051e-01 -3.50129515e-01 4.96571988e-01 4.29121941e-01 3.09159100e-01 9.13566530e-01 -1.29947448e+00 4.12655681e-01 -6.48647428e-01 -3.46325904e-01 -1.50867438e+00 5.26077859e-03 -8.21287513e-01 1.92076087e-01 -1.42115474e+00 1.51079781e-02 -5.12679040e-01 6.18959486e-01 2.69748718e-01 -2.41076704e-02 7.12761283e-02 1.11056998e-01 4.66017097e-01 -5.90251684e-01 1.26087695e-01 1.13023353e+00 -2.76119888e-01 -5.74485540e-01 2.02788755e-01 -1.88070983e-01 1.17655003e+00 6.80652142e-01 -2.54801154e-01 2.27695495e-01 -5.86940885e-01 2.21676171e-01 7.66447512e-03 1.61224335e-01 -7.40468919e-01 7.43245959e-01 -1.25537887e-01 8.82472992e-01 -1.54659951e+00 2.77363837e-01 -9.56894040e-01 -3.80652219e-01 6.76018000e-02 -1.67603284e-01 3.20055068e-01 2.56315053e-01 5.45681059e-01 4.32143629e-01 -4.39127803e-01 5.68829596e-01 2.70139664e-01 -3.77705127e-01 2.20967650e-01 -2.57510513e-01 -2.57531315e-01 1.16145122e+00 -6.43035650e-01 -4.57176328e-01 -1.33603022e-01 -2.04793394e-01 2.91260660e-01 3.94063890e-01 6.40066922e-01 9.00720358e-01 -1.00436246e+00 -7.40621090e-01 4.79281932e-01 -5.77134453e-02 3.06135863e-01 -5.21214120e-03 8.00296068e-01 -4.25474793e-01 6.67731047e-01 3.32888991e-01 -1.18009484e+00 -1.61836529e+00 4.62974519e-01 3.74413729e-01 1.06693983e-01 -1.04485643e+00 6.32362783e-01 8.23149681e-02 -3.82001936e-01 5.36044538e-01 -5.37465632e-01 -3.05528879e-01 -1.88147537e-02 4.53073442e-01 4.75400597e-01 -5.19634485e-02 -1.00300288e+00 -2.23431244e-01 1.53148556e+00 -2.94359177e-01 3.39000762e-01 1.08692384e+00 -4.05940801e-01 -4.34261300e-02 3.34753662e-01 8.51352811e-01 4.22479123e-01 -1.15216601e+00 -4.52476323e-01 1.80711821e-01 -6.06186569e-01 -3.06008667e-01 -5.11723757e-01 -7.20801175e-01 7.50803471e-01 4.00722235e-01 4.52797174e-01 6.08990848e-01 -5.90647459e-02 6.70670390e-01 5.29882312e-01 9.24882218e-02 -1.18301439e+00 1.82330668e-01 7.69246817e-01 9.45707321e-01 -1.10299480e+00 1.38800636e-01 -7.45396137e-01 -5.97869754e-01 1.74016094e+00 5.66224635e-01 -9.99867320e-02 3.70773822e-01 5.09049296e-01 9.07092616e-02 -2.26780862e-01 -2.11291581e-01 2.07282171e-01 6.21575296e-01 5.64019442e-01 2.95560986e-01 -9.53684375e-02 1.94968283e-01 -3.53565775e-02 -8.33041593e-02 -7.40768731e-01 5.33100426e-01 5.96196473e-01 -8.66359591e-01 -3.91946435e-01 -1.14874506e+00 5.68075120e-01 -1.75730601e-01 -1.61992103e-01 -6.06716514e-01 7.94384003e-01 -1.79420456e-01 8.35265279e-01 5.77527106e-01 -1.59680963e-01 4.92016256e-01 -1.64844636e-02 3.41456160e-02 -5.30725956e-01 -5.32227218e-01 6.35627747e-01 -3.29152882e-01 1.36402488e-01 1.13413155e-01 -8.60612631e-01 -1.58394337e+00 -3.47224921e-01 -8.34551752e-01 -2.76200175e-01 8.31125498e-01 1.00605106e+00 2.36146227e-01 1.28035292e-01 8.24764788e-01 -1.05369341e+00 -4.64152098e-01 -9.52008843e-01 -5.76763630e-01 2.54295081e-01 3.34633231e-01 -2.94397473e-01 -6.60956621e-01 3.51349145e-01]
[12.079937934875488, 2.27795672416687]
a9780e88-9f15-4811-9054-3d99dcfca9c3
advancing-the-state-of-the-art-in-open-domain
1812.10757
null
http://arxiv.org/abs/1812.10757v1
http://arxiv.org/pdf/1812.10757v1.pdf
Advancing the State of the Art in Open Domain Dialog Systems through the Alexa Prize
Building open domain conversational systems that allow users to have engaging conversations on topics of their choice is a challenging task. Alexa Prize was launched in 2016 to tackle the problem of achieving natural, sustained, coherent and engaging open-domain dialogs. In the second iteration of the competition in 2018, university teams advanced the state of the art by using context in dialog models, leveraging knowledge graphs for language understanding, handling complex utterances, building statistical and hierarchical dialog managers, and leveraging model-driven signals from user responses. The 2018 competition also included the provision of a suite of tools and models to the competitors including the CoBot (conversational bot) toolkit, topic and dialog act detection models, conversation evaluators, and a sensitive content detection model so that the competing teams could focus on building knowledge-rich, coherent and engaging multi-turn dialog systems. This paper outlines the advances developed by the university teams as well as the Alexa Prize team to achieve the common goal of advancing the science of Conversational AI. We address several key open-ended problems such as conversational speech recognition, open domain natural language understanding, commonsense reasoning, statistical dialog management, and dialog evaluation. These collaborative efforts have driven improved experiences by Alexa users to an average rating of 3.61, the median duration of 2 mins 18 seconds, and average turns to 14.6, increases of 14%, 92%, 54% respectively since the launch of the 2018 competition. For conversational speech recognition, we have improved our relative Word Error Rate by 55% and our relative Entity Error Rate by 34% since the launch of the Alexa Prize. Socialbots improved in quality significantly more rapidly in 2018, in part due to the release of the CoBot toolkit.
['Dilek Hakkani-Tur', 'Kate Bland', 'Ming Cheng', 'Han Song', 'Rohit Prasad', 'Raefer Gabriel', 'Sanju Pancholi', 'Jeff Nunn', 'Gene Hwang', 'Arindam Mandal', 'Anu Venkatesh', 'Yi Pan', 'Sanjeev Kwatra', 'Qing Liu', 'Nate Michel', 'Lauren Stubel', 'Karthik Gopalakrishnan', 'Behnam Hedayatnia', 'Anna Gottardi', 'Qinglang Chen', 'Eric King', 'Chandra Khatri']
2018-12-27
null
null
null
null
['open-domain-dialog']
['natural-language-processing']
[-1.97518587e-01 6.06741428e-01 7.71718696e-02 -5.67533195e-01 -9.96744931e-01 -8.92421722e-01 8.50139856e-01 3.94238308e-02 -2.23314181e-01 8.98066223e-01 9.41828251e-01 -2.15507329e-01 2.06124231e-01 -4.03902769e-01 8.77275392e-02 6.20517693e-02 1.54699013e-01 1.01447070e+00 2.53889579e-02 -8.40257406e-01 3.23770523e-01 -4.80266586e-02 -1.18714154e+00 6.38316214e-01 9.15855825e-01 6.38581753e-01 1.05995059e-01 1.08234084e+00 -6.14129901e-01 1.17553139e+00 -7.62004375e-01 -5.20629942e-01 -5.25040589e-02 -3.03151608e-01 -1.50104177e+00 -1.52203560e-01 1.46244943e-01 -5.16443074e-01 -1.21780768e-01 4.15128648e-01 5.51777840e-01 4.64112461e-01 3.95959109e-01 -1.35673344e+00 -5.21994531e-01 1.08850873e+00 2.48443261e-01 -4.39269887e-03 7.30099022e-01 5.85170090e-01 1.40945828e+00 -6.48186028e-01 6.50386095e-01 1.53567660e+00 5.70542693e-01 9.86850262e-01 -1.09538341e+00 -6.43299460e-01 1.31560981e-01 3.50241810e-02 -6.24804318e-01 -8.76239777e-01 2.74216324e-01 -6.46958590e-01 1.29265320e+00 3.98104519e-01 3.49555939e-01 1.36822581e+00 -4.94451046e-01 7.99181879e-01 1.19626987e+00 -3.44857872e-01 1.04533426e-01 5.97209990e-01 5.57351053e-01 5.10563493e-01 -5.73297918e-01 -2.38715425e-01 -5.23518741e-01 -4.62504715e-01 2.09990740e-01 -4.34258550e-01 5.91999926e-02 3.51374686e-01 -1.09401500e+00 1.20437014e+00 1.11054122e-01 3.63947660e-01 -2.07491025e-01 -3.95487726e-01 7.08454013e-01 4.49726969e-01 5.99001825e-01 1.06753993e+00 -6.20360613e-01 -9.65612531e-01 -2.47255221e-01 4.85020489e-01 1.80224848e+00 9.14611578e-01 4.66982424e-01 -3.14655900e-01 -5.00433207e-01 1.38765919e+00 2.86930799e-01 3.22972417e-01 3.40448737e-01 -1.43271792e+00 4.81160283e-01 7.80242980e-01 2.24656433e-01 -5.29887319e-01 -5.34097075e-01 2.73653835e-01 -1.52059734e-01 -1.77316755e-01 8.68492782e-01 -7.28522003e-01 -2.74025649e-01 1.79019308e+00 3.30193698e-01 -3.89401078e-01 3.80847216e-01 5.67196846e-01 1.14155984e+00 6.97596848e-01 2.38100410e-01 1.14038792e-02 1.53465438e+00 -1.02889884e+00 -7.74358451e-01 -2.88429052e-01 8.04424286e-01 -9.35623348e-01 1.25682509e+00 1.85015976e-01 -1.04004538e+00 -2.58556634e-01 -5.49393892e-01 -1.87637791e-01 -3.77685010e-01 -4.38172162e-01 6.06630325e-01 6.51789010e-01 -9.70283985e-01 1.45933837e-01 -3.62564057e-01 -6.99348867e-01 4.60586436e-02 1.44425631e-01 -9.45488065e-02 2.48387843e-01 -1.47063792e+00 1.30299819e+00 -1.48747712e-01 -3.75016928e-01 -7.40579545e-01 -8.07804108e-01 -6.42359138e-01 1.25353456e-01 3.17497551e-01 -4.10081357e-01 1.86778629e+00 -4.99736428e-01 -2.02115750e+00 9.87960815e-01 -3.86365466e-02 -4.29192752e-01 4.43585634e-01 -3.07469279e-01 -2.02187285e-01 -2.57690430e-01 1.83851793e-01 5.89153409e-01 1.54895391e-02 -8.49427223e-01 -7.12281644e-01 -2.26983532e-01 4.62334603e-01 4.31518108e-01 -3.09578478e-01 2.89954543e-01 7.63251930e-02 6.37302473e-02 -4.27646101e-01 -1.04607821e+00 -1.30842552e-01 -5.53796232e-01 -4.46724147e-01 -7.60923922e-01 7.67424822e-01 -8.48478854e-01 8.86349499e-01 -2.11378527e+00 -2.82353926e-02 -2.55172104e-01 5.06006479e-01 1.99377507e-01 4.90675196e-02 8.25258970e-01 4.14464861e-01 1.28192812e-01 8.89798701e-02 -5.80161452e-01 4.14118171e-01 4.70204744e-03 -4.18755144e-01 -2.58813292e-01 9.84969884e-02 7.19728172e-01 -8.98790300e-01 -2.16865078e-01 2.98297524e-01 2.62951897e-03 -5.40037751e-01 7.51514256e-01 -5.74248731e-01 6.30397141e-01 -2.43609011e-01 2.08578750e-01 1.84474766e-01 -2.00310305e-01 2.87472934e-01 2.82453775e-01 -4.25568491e-01 9.61739421e-01 -9.20516014e-01 1.43813419e+00 -7.93347359e-01 9.45880711e-01 6.23771906e-01 -4.43289369e-01 1.17086518e+00 5.93511939e-01 3.64250541e-01 -2.91130334e-01 1.87127560e-01 1.25191405e-01 1.27124175e-01 -5.65528035e-01 7.62410045e-01 -1.57720551e-01 -3.39899808e-01 8.02183986e-01 1.03071749e-01 -4.64393646e-01 -4.31436338e-02 6.19098067e-01 1.24878788e+00 -4.92514968e-01 8.30675811e-02 -2.07976073e-01 5.01350999e-01 3.30199271e-01 1.45987839e-01 7.44094610e-01 -5.77882171e-01 8.08851942e-02 6.68032408e-01 -2.82349288e-01 -9.64270532e-01 -7.63261735e-01 -6.96346350e-03 1.78832436e+00 -1.65995970e-01 -5.00661016e-01 -8.93044114e-01 -4.81987685e-01 -1.31767660e-01 7.91959643e-01 -1.65843830e-01 7.30723590e-02 -5.17227590e-01 -2.72004694e-01 9.52026963e-01 2.96220481e-01 7.09328175e-01 -1.33321893e+00 -9.36947390e-02 3.75710577e-01 -8.11871529e-01 -1.41185594e+00 -4.97780293e-01 -2.51425710e-02 -2.83148408e-01 -8.77007008e-01 -3.47129971e-01 -7.05784738e-01 -2.07364291e-01 -3.18463668e-02 1.06833899e+00 -1.48998678e-01 -1.80510342e-01 6.05356276e-01 -4.60754216e-01 -4.64535743e-01 -1.05231774e+00 3.68266195e-01 -1.24263978e-02 -2.92273581e-01 4.92898643e-01 -3.00124645e-01 -3.14450651e-01 5.11371195e-01 -2.50989407e-01 3.40335257e-02 5.85874505e-02 9.59158778e-01 -7.29154766e-01 -9.26608205e-01 9.88881469e-01 -8.02054524e-01 1.27956557e+00 -5.88463247e-01 -1.69505402e-01 1.55724004e-01 -5.42769492e-01 -7.98070952e-02 3.34060252e-01 -2.15170979e-01 -1.40524304e+00 -2.44366214e-01 -2.98347682e-01 1.64699659e-01 -3.53402615e-01 9.10822079e-02 1.51598662e-01 1.59815624e-01 9.56241131e-01 -1.22773297e-01 4.10209745e-01 -3.39889914e-01 7.77505755e-01 1.37891448e+00 2.92202532e-01 -6.67668223e-01 2.71746218e-01 3.28234695e-02 -8.93333554e-01 -1.02939904e+00 -7.71392822e-01 -6.80920303e-01 -2.70962089e-01 -2.63123125e-01 1.05966449e+00 -9.21873271e-01 -1.34089577e+00 4.27556336e-01 -1.35354960e+00 -8.08019161e-01 -1.77244678e-01 1.24537483e-01 -4.18460608e-01 2.04905912e-01 -9.64514494e-01 -1.06694245e+00 -5.86069286e-01 -1.05153668e+00 6.81222618e-01 2.95109421e-01 -1.04347980e+00 -1.14134502e+00 2.63734788e-01 1.18488669e+00 8.19170356e-01 -1.71806887e-01 7.82318711e-01 -1.49348271e+00 -2.20699042e-01 6.42327666e-02 -1.93507597e-01 3.85981649e-01 4.33775671e-02 -1.01217084e-01 -1.16383171e+00 1.83360115e-01 -1.63988560e-01 -9.11381483e-01 1.28179207e-01 4.44692150e-02 2.10527122e-01 -4.14492756e-01 -2.59851366e-01 -2.81323135e-01 5.16718030e-01 3.35149109e-01 1.19834945e-01 2.12413762e-02 3.53932202e-01 1.16260433e+00 4.62155253e-01 5.24879992e-01 8.42272937e-01 7.46378183e-01 6.36159107e-02 3.47894102e-01 -5.77017963e-02 -8.83801728e-02 4.57212210e-01 8.03438187e-01 3.26917171e-01 -9.47043970e-02 -1.09808588e+00 5.85153997e-01 -1.90826631e+00 -9.47490036e-01 -2.59096146e-01 1.79623103e+00 1.09458935e+00 1.09687328e-01 4.29673284e-01 -5.35561264e-01 7.19528615e-01 2.21007347e-01 -3.86533946e-01 -9.53310728e-01 1.40206307e-01 -3.72959450e-02 -1.09350063e-01 1.20924735e+00 -7.57605910e-01 1.31039870e+00 6.36852264e+00 5.41392267e-01 -8.15459490e-01 2.76741743e-01 6.64940298e-01 8.36170986e-02 -7.19405636e-02 1.98587686e-01 -1.11264396e+00 3.34482998e-01 1.27135146e+00 -4.15205598e-01 7.16556847e-01 1.03668141e+00 1.56765252e-01 -6.74778447e-02 -1.15668166e+00 7.19767392e-01 -1.23082861e-01 -1.46868682e+00 -5.85488796e-01 -9.84288566e-03 5.39519429e-01 3.12034577e-01 -3.78071994e-01 9.81501877e-01 1.12265635e+00 -1.19636250e+00 1.42794907e-01 2.96030134e-01 3.12015235e-01 -3.34992677e-01 5.68502605e-01 5.29194355e-01 -7.80574977e-01 -2.78558552e-01 1.22690640e-01 -5.94876885e-01 3.63546193e-01 -1.11615546e-02 -1.67793071e+00 -5.38801886e-02 3.91772985e-01 3.12718034e-01 1.16685487e-01 3.09836984e-01 4.19600634e-03 6.24297142e-01 -4.17461455e-01 -6.45470381e-01 3.05452257e-01 1.71163697e-02 8.33421826e-01 1.38306427e+00 -5.75400174e-01 3.94593239e-01 6.10002339e-01 9.74300206e-01 -1.64537862e-01 1.34225741e-01 -5.58111846e-01 -8.39499682e-02 9.67041731e-01 1.43895328e+00 -1.64630949e-01 -3.82202983e-01 -1.94058791e-01 5.96405447e-01 3.90635371e-01 -1.03229610e-02 -5.27508974e-01 -3.62332612e-01 1.00002611e+00 -2.61855513e-01 -2.48181477e-01 -1.54198080e-01 -5.49935937e-01 -8.00909936e-01 -2.97585726e-01 -1.28057432e+00 3.81421059e-01 -4.56466943e-01 -1.58061683e+00 9.03179228e-01 -6.29185792e-03 -3.84099633e-01 -7.10529506e-01 -4.07815248e-01 -8.81332576e-01 1.09649897e+00 -7.30168879e-01 -8.96544993e-01 -4.35572326e-01 4.56936032e-01 1.09942806e+00 -3.31760406e-01 1.23127401e+00 1.13140389e-01 -4.14620191e-01 4.18864191e-01 -1.45291641e-01 3.04779977e-01 1.01588750e+00 -1.34333396e+00 5.55176377e-01 -6.38364926e-02 -2.19253078e-01 6.96848214e-01 8.59371424e-01 -4.98334050e-01 -1.07259727e+00 -5.21731377e-01 1.02243447e+00 -1.08816636e+00 9.89791274e-01 -7.26673782e-01 -7.19137907e-01 7.12731242e-01 5.23355544e-01 -8.12241018e-01 8.05127203e-01 7.27022886e-01 -3.55946958e-01 3.15437973e-01 -1.21893942e+00 7.11937845e-01 7.78402448e-01 -8.59174550e-01 -8.37231278e-01 7.01105356e-01 1.04732120e+00 -5.43103933e-01 -9.10597563e-01 -3.00121307e-02 6.50627136e-01 -7.92427897e-01 5.86417913e-01 -7.83579528e-01 4.15216684e-01 5.78461409e-01 -1.22063033e-01 -1.33374548e+00 1.15558036e-01 -1.18002045e+00 1.70234829e-01 1.54379261e+00 6.50907218e-01 -5.77389777e-01 6.59138024e-01 1.30695987e+00 -2.86815047e-01 -5.29042721e-01 -7.23267376e-01 -3.82412821e-01 1.73148498e-01 -4.18740362e-01 1.32583067e-01 9.50590968e-01 8.74360025e-01 1.06722891e+00 -3.35534334e-01 -3.25758487e-01 2.11932473e-02 -1.49312571e-01 1.22680414e+00 -1.31149232e+00 -1.60666749e-01 -5.57704031e-01 1.05189905e-01 -1.35120618e+00 2.48540655e-01 -7.28726804e-01 2.24955052e-01 -1.48230648e+00 1.07216187e-01 -5.59547007e-01 5.39077938e-01 4.90438282e-01 -1.25292554e-01 -3.01308215e-01 3.92172068e-01 2.03402951e-01 -5.66763103e-01 4.55837101e-01 9.64118481e-01 -1.39280921e-02 -6.60765827e-01 1.27724394e-01 -1.05596745e+00 5.74550033e-01 7.64015853e-01 -7.86264390e-02 -2.28855461e-01 -1.64247565e-02 -2.08482817e-01 1.81811422e-01 1.53593719e-01 -5.67380548e-01 3.57266128e-01 -9.81749594e-02 -4.04339045e-01 -1.00933023e-01 8.91819000e-01 -2.32934967e-01 -3.63606304e-01 3.02217096e-01 -8.73672903e-01 -3.00674915e-01 4.20071572e-01 1.86198607e-01 -3.38510722e-02 -9.91932452e-02 6.63432598e-01 -3.10706496e-01 -5.63006043e-01 -2.44039133e-01 -6.97039366e-01 5.66439092e-01 9.54830348e-01 4.09701355e-02 -7.88305938e-01 -1.15601254e+00 -9.82906759e-01 6.23146832e-01 -1.32821426e-01 7.13056982e-01 6.69160262e-02 -6.53991997e-01 -8.56105983e-01 -2.09977031e-01 1.40000097e-02 -2.52212048e-01 2.77533084e-01 6.81014240e-01 -1.41750202e-01 6.68956935e-01 -1.45767927e-01 -3.73787493e-01 -1.58501303e+00 -2.94347674e-01 2.40716249e-01 -5.35175085e-01 -4.51318145e-01 1.08570743e+00 -1.18017495e-02 -1.06377184e+00 4.97385263e-01 1.04844328e-02 -3.11191022e-01 2.01940686e-01 6.51099324e-01 5.81028640e-01 -1.58305898e-01 -2.63209850e-01 -2.44693518e-01 -2.28901282e-01 -3.23991001e-01 -6.45876884e-01 1.09346282e+00 -2.33244240e-01 -9.31697488e-02 6.99171841e-01 9.64828312e-01 6.41959384e-02 -1.03565967e+00 -3.83367538e-01 1.68233797e-01 -1.55353518e-02 -4.58529055e-01 -1.38828123e+00 -1.13205105e-01 6.97613597e-01 2.39306942e-01 7.99398184e-01 2.68168628e-01 3.82362843e-01 9.53768134e-01 5.97326756e-01 1.35593995e-01 -1.03963828e+00 3.55738848e-01 1.40577257e+00 1.06954360e+00 -1.41294003e+00 -5.30928433e-01 -3.89853984e-01 -1.13332140e+00 1.03058767e+00 9.94289637e-01 1.89930856e-01 3.53684068e-01 2.12502837e-01 4.51070070e-01 -4.17913735e-01 -1.13670433e+00 -3.25753167e-02 -5.71645461e-02 5.42160273e-01 7.51252890e-01 2.32366711e-01 -6.45036176e-02 6.56174302e-01 -5.48206687e-01 -4.00284678e-01 4.78039563e-01 6.27664089e-01 -6.52296841e-01 -1.12486506e+00 -2.79343754e-01 2.67885089e-01 -1.47231370e-01 -3.44408929e-01 -1.00657082e+00 5.89258492e-01 -4.66223091e-01 1.65180004e+00 4.77099009e-02 -4.77488011e-01 4.30646837e-01 6.56910181e-01 -2.12037265e-01 -9.98699248e-01 -1.16398215e+00 -4.81061578e-01 9.97030318e-01 -3.48624259e-01 -2.19445918e-02 -6.96047843e-01 -1.23816931e+00 -7.68900692e-01 -4.52252209e-01 5.52858949e-01 7.72740126e-01 1.12963486e+00 5.91078401e-01 2.02262446e-01 7.47916281e-01 -5.23080468e-01 -9.73514915e-01 -1.57411766e+00 2.55824509e-03 1.83097646e-01 1.41071588e-01 -3.87250096e-01 -4.66665119e-01 -1.92563847e-01]
[12.669913291931152, 7.965521812438965]