Datasets:

ds3lab

/

ac-sgd-arxiv21

like 0

Follow

DS3Lab 4

A prototype version of the Q & U bolometric interferometer for cosmology (QUBIC) underwent a campaign of testing in the laboratory at Astroparticle Physics and Cosmology laboratory in Paris (APC). The detection chain is currently made of 256 NbSi transition edge sensors (TES) cooled to 320 mK. The readout system is a 128:1 time domain multiplexing scheme based on 128 SQUIDs cooled at 1 K that are controlled and amplified by an SiGe application specific integrated circuit at 40 K. We report the performance of this readout chain and the characterization of the TES. The readout system has been functionally tested and characterized in the lab and in QUBIC. The low noise amplifier demonstrated a white noise level of 0.3 nV.Hz^-0.5. Characterizations of the QUBIC detectors and readout electronics includes the measurement of I-V curves, time constant and the noise equivalent power. The QUBIC TES bolometer array has approximately 80% detectors within operational parameters. It demonstrated a thermal decoupling compatible with a phonon noise of about 5.10^-17 W.Hz^-0.5 at 410 mK critical temperature. While still limited by microphonics from the pulse tubes and noise aliasing from readout system, the instrument noise equivalent power is about 2.10^-16 W.Hz^-0.5, enough for the demonstration of bolometric interferometry.

We investigate the viability of producing galaxy mock catalogues with COmoving Lagrangian Acceleration (COLA) simulations in Modified Gravity (MG) models employing the Halo Occupation Distribution (HOD) formalism. In this work, we focus on two theories of MG: $f(R)$ gravity with the chameleon mechanism, and a braneworld model (nDGP) that incorporates the Vainshtein mechanism. We use a suite of full $N$-body simulations in MG as a benchmark to test the accuracy of COLA simulations. At the level of Dark Matter (DM), we show that COLA accurately reproduces the matter power spectrum up to $k \sim 1 h {\rm Mpc}^{-1}$, while it is less accurate in reproducing the velocity field. To produce halo catalogues, we find that the ROCKSTAR halo-finder does not perform well with COLA simulations. On the other hand, using a simple Friends-of-Friends (FoF) finder and an empirical mass conversion from FoF to spherical over-density masses, we are able to produce halo catalogues in COLA that are in good agreement with those in $N$-body simulations. To consider the effects of the MG fifth force on the halo profile, we derive simple fitting formulae for the concentration-mass and the velocity dispersion-mass relations that we calibrate using ROCKSTAR halo catalogues in $N$-body simulations. We then use these results to extend the HOD formalism to modified gravity simulations in COLA. We use an HOD model with five parameters that we tune to obtain galaxy catalogues in redshift space. We find that despite the great freedom of the HOD model, MG leaves characteristic imprints in the redshift space power spectrum multipoles and these features are well captured by the COLA galaxy catalogues.

Combinatorial design theory studies set systems with certain balance and symmetry properties and has applications to computer science and elsewhere. This paper presents a modular approach to formalising designs for the first time using Isabelle and assesses the usability of a locale-centric approach to formalisations of mathematical structures. We demonstrate how locales can be used to specify numerous types of designs and their hierarchy. The resulting library, which is concise and adaptable, includes formal definitions and proofs for many key properties, operations, and theorems on the construction and existence of designs.

Motivated by the study of high energy Steklov eigenfunctions, we examine the semi-classical Robin Laplacian. In the two dimensional situation, we determine an effective operator describing the asymptotic distribution of the negative eigenvalues, and we prove that the corresponding eigenfunctions decay away from the boundary, for all dimensions.

Given a finite abelian group $G$ and a natural number $t$, there are two natural substructures of the Cartesian power $G^t$; namely, $S^t$ where $S$ is a subset of $G$, and $x+H$ a coset of a subgroup $H$ of $G^t$. A natural question is whether two such different structures have non-empty intersection. This turns out to be an NP-complete problem. If we fix $G$ and $S$, then the problem is in $P$ if $S$ is a coset in $G$ or if $S$ is empty, and NP-complete otherwise; if we restrict to intersecting powers of $S$ with subgroups, the problem is in $P$ if $\bigcap_{n\in\mathbb{Z} \mid nS \subset S} nS$ is a coset or empty, and NP-complete otherwise. These theorems have applications in the article [Spe21], where they are used as a stepping stone between a purely combinatorial and a purely algebraic problem.

Cycles, which can be found in many different kinds of networks, make the problems more intractable, especially when dealing with dynamical processes on networks. On the contrary, tree networks in which no cycle exists, are simplifications and usually allow for analyticity. There lacks a quantity, however, to tell the ratio of cycles which determines the extent of network being close to tree networks. Therefore we introduce the term Cycle Nodes Ratio (CNR) to describe the ratio of number of nodes belonging to cycles to the number of total nodes, and provide an algorithm to calculate CNR. CNR is studied in both network models and real networks. The CNR remains unchanged in different sized Erd\"os R\'enyi (ER) networks with the same average degree, and increases with the average degree, which yields a critical turning point. The approximate analytical solutions of CNR in ER networks are given, which fits the simulations well. Furthermore, the difference between CNR and two-core ratio (TCR) is analyzed. The critical phenomenon is explored by analysing the giant component of networks. We compare the CNR in network models and real networks, and find the latter is generally smaller. Combining the coarse-graining method can distinguish the CNR structure of networks with high average degree. The CNR is also applied to four different kinds of transportation networks and fungal networks, which give rise to different zones of effect. It is interesting to see that CNR is very useful in network recognition of machine learning.

We develop a theory of T-duality for transitive Courant algebroids. We show that T-duality between transitive Courant algebroids E\rightarrow M and \tilde{E}\rightarrow \tilde{M} induces a map between the spaces of sections of the corresponding canonical weighted spinor bundles \mathbb{S}_{E} and \mathbb{S}_{\tilde{E}} intertwining the canonical Dirac generating operators. The map is shown to induce an isomorphism between the spaces of invariant spinors, compatible with an isomorphism between the spaces of invariant sections of the Courant algebroids. The notion of invariance is defined after lifting the vertical parallelisms of the underlying torus bundles M\rightarrow B and \tilde{M} \rightarrow B to the Courant algebroids and their spinor bundles. We prove a general existence result for T-duals under assumptions generalizing the cohomological integrality conditions for T-duality in the exact case. Specializing our construction, we find that the T-dual of an exact or a heterotic Courant algebroid is again exact or heterotic, respectively.

Reggeon field theory (RFT), originally developed in the context of high energy diffraction scattering, has a much wider applicability, describing, for example, the universal critical behavior of stochastic population models as well as probabilistic geometric problems such as directed percolation. In 1975 Suranyi and others developed cut RFT, which can incorporate the cutting rules of Abramovskii, Gribov and Kancheli for how each diagram contributes to inclusive cross-sections. In this note we describe the corresponding probabilistic interpretations of cut RFT: as a population model of two genotypes, which can reproduce both asexually and sexually; and as a kind of bicolor directed percolation problem. In both cases the AGK rules correspond to simple limiting cases of these problems.

Utilizing the Atacama Large Millimeter/submillimeter Array (ALMA), we present CS line maps in five rotational lines ($J_{\rm u}=7, 5, 4, 3, 2$) toward the circumnuclear disk (CND) and streamers of the Galactic Center. Our primary goal is to resolve the compact structures within the CND and the streamers, in order to understand the stability conditions of molecular cores in the vicinity of the supermassive black hole (SMBH) Sgr A*. Our data provide the first homogeneous high-resolution ($1.3'' = 0.05$ pc) observations aiming at resolving density and temperature structures. The CS clouds have sizes of $0.05-0.2$ pc with a broad range of velocity dispersion ($\sigma_{\rm FWHM}=5-40$ km s$^{-1}$). The CS clouds are a mixture of warm ($T_{\rm k}\ge 50-500$ K, n$_{\rm H_2}$=$10^{3-5}$ cm$^{-3}$) and cold gas ($T_{\rm k}\le 50$ K, n$_{\rm H_2}$=$10^{6-8}$ cm$^{-3}$). A stability analysis based on the unmagnetized virial theorem including tidal force shows that $84^{+16}_{-37}$ % of the total gas mass is tidally stable, which accounts for the majority of gas mass. Turbulence dominates the internal energy and thereby sets the threshold densities $10-100$ times higher than the tidal limit at distance $\ge 1.5$ pc to Sgr A*, and therefore, inhibits the clouds from collapsing to form stars near the SMBH. However, within the central $1.5$ pc, the tidal force overrides turbulence and the threshold densities for a gravitational collapse quickly grow to $\ge 10^{8}$ cm$^{-3}$.

Cross-modal recipe retrieval has recently gained substantial attention due to the importance of food in people's lives, as well as the availability of vast amounts of digital cooking recipes and food images to train machine learning models. In this work, we revisit existing approaches for cross-modal recipe retrieval and propose a simplified end-to-end model based on well established and high performing encoders for text and images. We introduce a hierarchical recipe Transformer which attentively encodes individual recipe components (titles, ingredients and instructions). Further, we propose a self-supervised loss function computed on top of pairs of individual recipe components, which is able to leverage semantic relationships within recipes, and enables training using both image-recipe and recipe-only samples. We conduct a thorough analysis and ablation studies to validate our design choices. As a result, our proposed method achieves state-of-the-art performance in the cross-modal recipe retrieval task on the Recipe1M dataset. We make code and models publicly available.

Quantization is one of the core components in lossy image compression. For neural image compression, end-to-end optimization requires differentiable approximations of quantization, which can generally be grouped into three categories: additive uniform noise, straight-through estimator and soft-to-hard annealing. Training with additive uniform noise approximates the quantization error variationally but suffers from the train-test mismatch. The other two methods do not encounter this mismatch but, as shown in this paper, hurt the rate-distortion performance since the latent representation ability is weakened. We thus propose a novel soft-then-hard quantization strategy for neural image compression that first learns an expressive latent space softly, then closes the train-test mismatch with hard quantization. In addition, beyond the fixed integer quantization, we apply scaled additive uniform noise to adaptively control the quantization granularity by deriving a new variational upper bound on actual rate. Experiments demonstrate that our proposed methods are easy to adopt, stable to train, and highly effective especially on complex compression models.

In this paper, we introduce PASSAT, a practical system to boost the security assurance delivered by the current cloud architecture without requiring any changes or cooperation from the cloud service providers. PASSAT is an application transparent to the cloud servers that allows users to securely and efficiently store and access their files stored on public cloud storage based on a single master password. Using a fast and light-weight XOR secret sharing scheme, PASSAT secret-shares users' files and distributes them among n publicly available cloud platforms. To access the files, PASSAT communicates with any k out of n cloud platforms to receive the shares and runs a secret-sharing reconstruction algorithm to recover the files. An attacker (insider or outsider) who compromises or colludes with less than k platforms cannot learn the user's files or modify the files stealthily. To authenticate the user to multiple cloud platforms, PASSAT crucially stores the authentication credentials, specific to each platform on a password manager, protected under the user's master password. Upon requesting access to files, the user enters the password to unlock the vault and fetches the authentication tokens using which PASSAT can interact with cloud storage. Our instantiation of PASSAT based on (2, 3)-XOR secret sharing of Kurihara et al., implemented with three popular storage providers, namely, Google Drive, Box, and Dropbox, confirms that our approach can efficiently enhance the confidentiality, integrity, and availability of the stored files with no changes on the servers.

We describe a time lens to expand the dynamic range of photon Doppler velocimetry (PDV) systems. The principle and preliminary design of a time-lens PDV (TL-PDV) are explained and shown to be feasible through simulations. In a PDV system, an interferometer is used for measuring frequency shifts due to the Doppler effect from the target motion. However, the sampling rate of the electronics could limit the velocity range of a PDV system. A four-wave-mixing (FWM) time lens applies a quadratic temporal phase to an optical signal within a nonlinear FWM medium (such as an integrated photonic waveguide or highly nonlinear optical fiber). By spectrally isolating the mixing product, termed the idler, and with appropriate lengths of dispersion prior and after to this FWM time lens, a temporally magnified version of the input signal is generated. Therefore, the frequency shifts of PDV can be "slowed down" with the magnification factor $M$ of the time lens. $M=1$ corresponds to a regular PDV without a TL. $M=10$ has been shown to be feasible for a TL-PDV. Use of this effect for PDV can expand the velocity measurement range and allow the use of lower bandwidth electronics. TL-PDV will open up new avenues for various dynamic materials experiments.

3D perception using sensors under vehicle industrial standard is the rigid demand in autonomous driving. MEMS LiDAR emerges with irresistible trend due to its lower cost, more robust, and meeting the mass-production standards. However, it suffers small field of view (FoV), slowing down the step of its population. In this paper, we propose LEAD, i.e., LiDAR Extender for Autonomous Driving, to extend the MEMS LiDAR by coupled image w.r.t both FoV and range. We propose a multi-stage propagation strategy based on depth distributions and uncertainty map, which shows effective propagation ability. Moreover, our depth outpainting/propagation network follows a teacher-student training fashion, which transfers depth estimation ability to depth completion network without any scale error passed. To validate the LiDAR extension quality, we utilize a high-precise laser scanner to generate a ground-truth dataset. Quantitative and qualitative evaluations show that our scheme outperforms SOTAs with a large margin. We believe the proposed LEAD along with the dataset would benefit the community w.r.t depth researches.

This paper proposes VARA-TTS, a non-autoregressive (non-AR) text-to-speech (TTS) model using a very deep Variational Autoencoder (VDVAE) with Residual Attention mechanism, which refines the textual-to-acoustic alignment layer-wisely. Hierarchical latent variables with different temporal resolutions from the VDVAE are used as queries for residual attention module. By leveraging the coarse global alignment from previous attention layer as an extra input, the following attention layer can produce a refined version of alignment. This amortizes the burden of learning the textual-to-acoustic alignment among multiple attention layers and outperforms the use of only a single attention layer in robustness. An utterance-level speaking speed factor is computed by a jointly-trained speaking speed predictor, which takes the mean-pooled latent variables of the coarsest layer as input, to determine number of acoustic frames at inference. Experimental results show that VARA-TTS achieves slightly inferior speech quality to an AR counterpart Tacotron 2 but an order-of-magnitude speed-up at inference; and outperforms an analogous non-AR model, BVAE-TTS, in terms of speech quality.

This work introduces a methodology for studying synchronization in adaptive networks with heterogeneous plasticity (adaptation) rules. As a paradigmatic model, we consider a network of adaptively coupled phase oscillators with distance-dependent adaptations. For this system, we extend the master stability function approach to adaptive networks with heterogeneous adaptation. Our method allows for separating the contributions of network structure, local node dynamics, and heterogeneous adaptation in determining synchronization. Utilizing our proposed methodology, we explain mechanisms leading to synchronization or desynchronization by enhanced long-range connections in nonlocally coupled ring networks and networks with Gaussian distance-dependent coupling weights equipped with a biologically motivated plasticity rule.

We consider the problem of budget allocation for competitive influence maximization over social networks. In this problem, multiple competing parties (players) want to distribute their limited advertising resources over a set of social individuals to maximize their long-run cumulative payoffs. It is assumed that the individuals are connected via a social network and update their opinions based on the classical DeGroot model. The players must decide the budget distribution among the individuals at a finite number of campaign times to maximize their overall payoff given as a function of individuals' opinions. We show that i) the optimal investment strategy for the case of a single-player can be found in polynomial time by solving a concave program, and ii) the open-loop equilibrium strategies for the multiplayer dynamic game can be computed efficiently by following natural regret minimization dynamics. Our results extend the earlier work on the static version of the problem to a dynamic multistage game.

There is mounting evidence that ultra-energetic neutrinos of astrophysical origin may be associated with blazars. Here we investigate a unique sample of 47 blazars, $\sim 20$ of which could be new neutrino sources. In particular, we focus on 17 objects of yet unknown redshift, for which we present optical spectroscopy secured at the Gran Telescopio Canarias and the ESO Very Large Telescope. We find all sources but one (a quasar) to be BL Lac objects. For nine targets we are able to determine the redshift (0.09~$<$~z~$<$~1.6), while for the others we set a lower limit on it, based on either the robust detection of intervening absorption systems or on an estimation derived from the absence of spectral signatures of the host galaxy. In some spectra we detect forbidden and semi-forbidden emission lines with luminosities in the range $10^{40} - 10^{41}$ erg s$^{-1}$. We also report on the spectroscopy of seven blazars possibly associated with energetic neutrinos that partially meet the criteria of our sample and are discussed in the Appendix. These results represent the starting point of our investigation into the real nature of these objects and their likelihood of being neutrino emitters.

Given C$^*$-algebras $A$ and $B$ and a $^*$-homomorphism $\phi:A\rightarrow B$, we adopt the portrait of the relative $K$-theory $K_*(\phi)$ due to Karoubi using Banach categories and Banach functors. We show that the elements of the relative groups may be represented in a simple form. We prove the existence of two six-term exact sequences, and we use these sequences to deduce the fact that the relative theory is isomorphic, in a natural way, to the $K$-theory of the mapping cone of $\phi$ as an excision result.

We experimentally demonstrate, for the first time, noise diagnostics by repeated quantum measurements. Specifically, we establish the ability of a single photon, subjected to random polarisation noise, to diagnose non-Markovian temporal correlations of such a noise process. In the frequency domain, these noise correlations correspond to colored noise spectra, as opposed to the ones related to Markovian, white noise. Both the noise spectrum and its corresponding temporal correlations are diagnosed by probing the photon by means of frequent, (partially-)selective polarisation measurements. Our main result is the experimental demonstration that noise with positive temporal correlations corresponds to our single photon undergoing a dynamical regime enabled by the quantum Zeno effect (QZE), while noise characterized by negative (anti-) correlations corresponds to regimes associated with the anti-Zeno effect (AZE). This demonstration opens the way to a new kind of noise spectroscopy based on QZE and AZE in photon (or other single-particle) state probing.

Known force terms arising in the Ehrenfest dynamics of quantum electrons and classical nuclei, due to a moving basis set for the former, can be understood in terms of the curvature of the manifold hosting the quantum states of the electronic subsystem. Namely, the velocity-dependent terms appearing in the Ehrenfest forces on the nuclei acquire a geometrical meaning in terms of the intrinsic curvature of the manifold, while Pulay terms relate to its extrinsic curvature.

The production of dileptons with an invariant mass in the range 1 GeV < M < 5 GeV provides unique insight into the approach to thermal equilibrium in ultrarelativistic nucleus-nucleus collisions. In this mass range, they are produced through the annihilation of quark-antiquark pairs in the early stages of the collision. They are sensitive to the anisotropy of the quark momentum distribution, and also to the quark abundance, which is expected to be underpopulated relative to thermal equilibrium. We take into account both effects based on recent theoretical developments in QCD kinetic theory, and study how the dilepton mass spectrum depends on the shear viscosity to entropy ratio that controls the equilibration time. We evaluate the background from the Drell-Yan process and argue that future detector developments can suppress the additional background from semileptonic decays of heavy flavors.

In extreme learning machines (ELM) the hidden-layer coefficients are randomly set and fixed, while the output-layer coefficients of the neural network are computed by a least squares method. The randomly-assigned coefficients in ELM are known to influence its performance and accuracy significantly. In this paper we present a modified batch intrinsic plasticity (modBIP) method for pre-training the random coefficients in the ELM neural networks. The current method is devised based on the same principle as the batch intrinsic plasticity (BIP) method, namely, by enhancing the information transmission in every node of the neural network. It differs from BIP in two prominent aspects. First, modBIP does not involve the activation function in its algorithm, and it can be applied with any activation function in the neural network. In contrast, BIP employs the inverse of the activation function in its construction, and requires the activation function to be invertible (or monotonic). The modBIP method can work with the often-used non-monotonic activation functions (e.g. Gaussian, swish, Gaussian error linear unit, and radial-basis type functions), with which BIP breaks down. Second, modBIP generates target samples on random intervals with a minimum size, which leads to highly accurate computation results when combined with ELM. The combined ELM/modBIP method is markedly more accurate than ELM/BIP in numerical simulations. Ample numerical experiments are presented with shallow and deep neural networks for function approximation and boundary/initial value problems with partial differential equations. They demonstrate that the combined ELM/modBIP method produces highly accurate simulation results, and that its accuracy is insensitive to the random-coefficient initializations in the neural network. This is in sharp contrast with the ELM results without pre-training of the random coefficients.

A key task in design work is grasping the client's implicit tastes. Designers often do this based on a set of examples from the client. However, recognizing a common pattern among many intertwining variables such as color, texture, and layout and synthesizing them into a composite preference can be challenging. In this paper, we leverage the pattern recognition capability of computational models to aid in this task. We offer a set of principles for computationally learning personal style. The principles are manifested in PseudoClient, a deep learning framework that learns a computational model for personal graphic design style from only a handful of examples. In several experiments, we found that PseudoClient achieves a 79.40% accuracy with only five positive and negative examples, outperforming several alternative methods. Finally, we discuss how PseudoClient can be utilized as a building block to support the development of future design applications.

Let $\Omega $ be an open subset of $\mathbb{R}^{N}$, and let $p,\, q:\Omega \rightarrow \left[ 1,\infty \right] $ be measurable functions. We give a necessary and sufficient condition for the embedding of the variable exponent space $L^{p(\cdot )}\left( \Omega \right) $ in $L^{q(\cdot )}\left( \Omega \right) $ to be almost compact. This leads to a condition on $\Omega, \, p$ and $q$ sufficient to ensure that the Sobolev space $W^{1,p(\cdot )}\left( \Omega \right) $ based on $L^{p(\cdot )}\left( \Omega \right) $ is compactly embedded in $L^{q(\cdot )}\left( \Omega \right) ;$ compact embedding results of this type already in the literature are included as special cases.

As widely recognized, vortex represents flow rotation. Vortex should have a local rotation axis as its direction and angular speed as its strength. Vorticity vector has been considered the rotation axis, and vorticity magnitude the rotational strength for a long time in classical fluid kinematics. However, this concept cannot stand in viscous flow. This study demonstrates by rigorous mathematical proof that the vorticity vector is not the fluid rotation axis, and vorticity is not the rotation strength. On the other hand, the Liutex vector is mathematically proved as the fluid rotation axis, and the Liutex magnitude is twice the fluid angular speed.

Most countries have started vaccinating people against COVID-19. However, due to limited production capacities and logistical challenges it will take months/years until herd immunity is achieved. Therefore, vaccination and social distancing have to be coordinated. In this paper, we provide some insight on this topic using optimization-based control on an age-differentiated compartmental model. For real-life decision making, we investigate the impact of the planning horizon on the optimal vaccination/social distancing strategy. We find that in order to reduce social distancing in the long run, without overburdening the healthcare system, it is essential to vaccinate the people with the highest contact rates first. That is also the case if the objective is to minimize fatalities provided that the social distancing measures are sufficiently strict. However, for short-term planning it is optimal to focus on the high-risk group.

We present a model of exclusive $\phi$-meson lepto-production $ep \to e'p'\phi$ near threshold which features the strangeness gravitational form factors of the proton. We argue that the shape of the differential cross section $d\sigma/dt$ is a sensitive probe of the strangeness D-term of the proton.

With the promise of reliability in cloud, more enterprises are migrating to cloud. The process of continuous integration/deployment (CICD) in cloud connects developers who need to deliver value faster and more transparently with site reliability engineers (SREs) who need to manage applications reliably. SREs feed back development issues to developers, and developers commit fixes and trigger CICD to redeploy. The release cycle is more continuous than ever, thus the code to production is faster and more automated. To provide this higher level agility, the cloud platforms become more complex in the face of flexibility with deeper layers of virtualization. However, reliability does not come for free with all these complexities. Software engineers and SREs need to deal with wider information spectrum from virtualized layers. Therefore, providing correlated information with true positive evidences is critical to identify the root cause of issues quickly in order to reduce mean time to recover (MTTR), performance metrics for SREs. Similarity, knowledge, or statistics driven approaches have been effective, but with increasing data volume and types, an individual approach is limited to correlate semantic relations of different data sources. In this paper, we introduce FIXME to enhance software reliability with hybrid diagnosis approaches for enterprises. Our evaluation results show using hybrid diagnosis approach is about 17% better in precision. The results are helpful for both practitioners and researchers to develop hybrid diagnosis in the highly dynamic cloud environment.

In multistage manufacturing systems, modeling multiple quality indices based on the process sensing variables is important. However, the classic modeling technique predicts each quality variable one at a time, which fails to consider the correlation within or between stages. We propose a deep multistage multi-task learning framework to jointly predict all output sensing variables in a unified end-to-end learning framework according to the sequential system architecture in the MMS. Our numerical studies and real case study have shown that the new model has a superior performance compared to many benchmark methods as well as great interpretability through developed variable selection techniques.

Phase noise of the frequency synthesizer is one of the main limitations to the short-term stability of microwave atomic clocks. In this work, we demonstrated a low-noise, simple-architecture microwave frequency synthesizer for a coherent population trapping (CPT) clock. The synthesizer is mainly composed of a 100 MHz oven controlled crystal oscillator (OCXO), a microwave comb generator and a direct digital synthesizer (DDS). The absolute phase noises of 3.417 GHz signal are measured to be -55 dBc/Hz, -81 dBc/Hz, -111 dBc/Hz and -134 dBc/Hz, respectively, for 1 Hz, 10 Hz, 100 Hz and 1 kHz offset frequencies, which shows only 1 dB deterioration at the second harmonic of the modulation frequency of the atomic clock. The estimated frequency stability of intermodulation effect is 4.7*10^{-14} at 1s averaging time, which is about half order of magnitude lower than that of the state-of-the-art CPT Rb clock. Our work offers an alternative microwave synthesizer for high-performance CPT Rb atomic clock.

The Schr\"odinger equation is solved numerically for charmonium using the discrete variable representation (DVR) method. The Hamiltonian matrix is constructed and diagonalized to obtain the eigenvalues and eigenfunctions. Using these eigenvalues and eigenfunctions, spectra and various decay widths are calculated. The obtained results are in good agreement with other numerical methods and with experiments.

We are motivated by the problem of providing strong generalization guarantees in the context of meta-learning. Existing generalization bounds are either challenging to evaluate or provide vacuous guarantees in even relatively simple settings. We derive a probably approximately correct (PAC) bound for gradient-based meta-learning using two different generalization frameworks in order to deal with the qualitatively different challenges of generalization at the "base" and "meta" levels. We employ bounds for uniformly stable algorithms at the base level and bounds from the PAC-Bayes framework at the meta level. The result of this approach is a novel PAC bound that is tighter when the base learner adapts quickly, which is precisely the goal of meta-learning. We show that our bound provides a tighter guarantee than other bounds on a toy non-convex problem on the unit sphere and a text-based classification example. We also present a practical regularization scheme motivated by the bound in settings where the bound is loose and demonstrate improved performance over baseline techniques.

Designing intelligent microrobots that can autonomously navigate and perform instructed routines in blood vessels, a complex and crowded environment with obstacles including dense cells, different flow patterns and diverse vascular geometries, can offer enormous possibilities in biomedical applications. Here we report a hierarchical control scheme that enables a microrobot to efficiently navigate and execute customizable routines in blood vessels. The control scheme consists of two highly decoupled components: a high-level controller setting short-ranged dynamic targets to guide the microrobot to follow a preset path and a low-level deep reinforcement learning (DRL) controller responsible for maneuvering microrobots towards these dynamic guiding targets. The proposed DRL controller utilizes three-dimensional (3D) convolutional neural networks and is capable of learning control policy directly from a coarse raw 3D sensory input. In blood vessels with rich configurations of red blood cells and vessel geometry, the control scheme enables efficient navigation and faithful execution of instructed routines. The control scheme is also robust to adversarial perturbations including blood flows. This study provides a proof-of-principle for designing data-driven control systems for autonomous navigation in vascular networks; it illustrates the great potential of artificial intelligence for broad biomedical applications such as target drug delivery, blood clots clear, precision surgery, disease diagnosis, and more.

Communication overhead is the key challenge for distributed training. Gradient compression is a widely used approach to reduce communication traffic. When combining with parallel communication mechanism method like pipeline, gradient compression technique can greatly alleviate the impact of communication overhead. However, there exists two problems of gradient compression technique to be solved. Firstly, gradient compression brings in extra computation cost, which will delay the next training iteration. Secondly, gradient compression usually leads to the decrease of convergence accuracy.

Recently, much attention has been paid to the societal impact of AI, especially concerns regarding its fairness. A growing body of research has identified unfair AI systems and proposed methods to debias them, yet many challenges remain. Representation learning for Heterogeneous Information Networks (HINs), a fundamental building block used in complex network mining, has socially consequential applications such as automated career counseling, but there have been few attempts to ensure that it will not encode or amplify harmful biases, e.g. sexism in the job market. To address this gap, in this paper we propose a comprehensive set of de-biasing methods for fair HINs representation learning, including sampling-based, projection-based, and graph neural networks (GNNs)-based techniques. We systematically study the behavior of these algorithms, especially their capability in balancing the trade-off between fairness and prediction accuracy. We evaluate the performance of the proposed methods in an automated career counseling application where we mitigate gender bias in career recommendation. Based on the evaluation results on two datasets, we identify the most effective fair HINs representation learning techniques under different conditions.

This research recasts the network attack dataset from UNSW-NB15 as an intrusion detection problem in image space. Using one-hot-encodings, the resulting grayscale thumbnails provide a quarter-million examples for deep learning algorithms. Applying the MobileNetV2's convolutional neural network architecture, the work demonstrates a 97% accuracy in distinguishing normal and attack traffic. Further class refinements to 9 individual attack families (exploits, worms, shellcodes) show an overall 56% accuracy. Using feature importance rank, a random forest solution on subsets show the most important source-destination factors and the least important ones as mainly obscure protocols. The dataset is available on Kaggle.

Tamil is a Dravidian language that is commonly used and spoken in the southern part of Asia. In the era of social media, memes have been a fun moment in the day-to-day life of people. Here, we try to analyze the true meaning of Tamil memes by categorizing them as troll and non-troll. We propose an ingenious model comprising of a transformer-transformer architecture that tries to attain state-of-the-art by using attention as its main component. The dataset consists of troll and non-troll images with their captions as text. The task is a binary classification task. The objective of the model is to pay more attention to the extracted features and to ignore the noise in both images and text.

Self-organized spatial patterns of vegetation are frequent in water-limited regions and have been suggested as important indicators of ecosystem health. However, the mechanisms underlying their emergence remain unclear. Some theories hypothesize that patterns could result from a scale-dependent feedback (SDF), whereby interactions favoring plant growth dominate at short distances and growth-inhibitory interactions dominate in the long range. However, we know little about how net plant-to-plant interactions may change sign with inter-individual distance, and in the absence of strong empirical support, the relevance of this SDF for vegetation pattern formation remains disputed. These theories predict a sequential change in pattern shape from gapped to labyrinthine to spotted spatial patterns as precipitation declines. Nonetheless, alternative theories show that the same sequence of patterns could emerge even if net interactions between plants were always inhibitory (purely competitive feedbacks, PCF). Although these alternative hypotheses lead to visually indistinguishable patterns they predict very different desertification dynamics following the spotted pattern. Moreover, vegetation interaction with other ecosystem components can introduce additional spatio-temporal scales that reshape both the patterns and the desertification dynamics. Therefore, to make reliable ecological predictions for a focal ecosystem, it is crucial that models accurately capture the mechanisms at play in the system of interest. Here, we review existing theories for vegetation self-organization and their conflicting predictions about desertification dynamics. We further discuss possible ways for reconciling these predictions and potential empirical tests via manipulative experiments to improve our understanding of how vegetation self-organizes and better predict the fate of the ecosystems where they form.

Deep learning techniques have achieved great success in remote sensing image change detection. Most of them are supervised techniques, which usually require large amounts of training data and are limited to a particular application. Self-supervised methods as an unsupervised approach are popularly used to solve this problem and are widely used in unsupervised binary change detection tasks. However, the existing self-supervised methods in change detection are based on pre-tasks or at patch-level, which may be sub-optimal for pixel-wise change detection tasks. Therefore, in this work, a pixel-wise contrastive approach is proposed to overcome this limitation. This is achieved by using contrastive loss in pixel-level features on an unlabeled multi-view setting. In this approach, a Siamese ResUnet is trained to obtain pixel-wise representations and to align features from shifted positive pairs. Meanwhile, vector quantization is used to augment the learned features in two branches. The final binary change map is obtained by subtracting features of one branch from features of the other branch and using the Rosin thresholding method. To overcome the effects of regular seasonal changes in binary change maps, we also used an uncertainty method to enhance the temporal robustness of the proposed approach. Two homogeneous (OSCD and MUDS) datasets and one heterogeneous (California Flood) dataset are used to evaluate the performance of the proposed approach. Results demonstrate improvements in both efficiency and accuracy over the patch-wise multi-view contrastive method.

Convolutional Neural Networks (CNN) have been rigorously studied for Hyperspectral Image Classification (HSIC) and are known to be effective in exploiting joint spatial-spectral information with the expense of lower generalization performance and learning speed due to the hard labels and non-uniform distribution over labels. Several regularization techniques have been used to overcome the aforesaid issues. However, sometimes models learn to predict the samples extremely confidently which is not good from a generalization point of view. Therefore, this paper proposed an idea to enhance the generalization performance of a hybrid CNN for HSIC using soft labels that are a weighted average of the hard labels and uniform distribution over ground labels. The proposed method helps to prevent CNN from becoming over-confident. We empirically show that in improving generalization performance, label smoothing also improves model calibration which significantly improves beam-search. Several publicly available Hyperspectral datasets are used to validate the experimental evaluation which reveals improved generalization performance, statistical significance, and computational complexity as compared to the state-of-the-art models. The code will be made available at https://github.com/mahmad00.

In recent years, the interest growth in the Blockchains (BC) and Internet-of-Things (IoT) integration -- termed as BIoT -- for more trust via decentralization has led to great potentials in various use cases such as health care, supply chain tracking, and smart cities. A key element of BIoT ecosystems is the data transactions (TX) that include the data collected by IoT devices. BIoT applications face many challenges to comply with the European General Data Protection Regulation (GDPR) i.e., enabling users to hold on to their rights for deleting or modifying their data stored on publicly accessible and immutable BCs. In this regard, this paper identifies the requirements of BCs for being GDPR compliant in BIoT use cases. Accordingly, an on-chain solution is proposed that allows fine-grained modification (update and erasure) operations on TXs' data fields within a BC. The proposed solution is based on a cryptographic primitive called Chameleon Hashing. The novelty of this approach is manifold. BC users have the authority to update their data, which are addressed at the TX level with no side-effects on the block or chain. By performing and storing the data updates, all on-chain, traceability and verifiability of the BC are preserved. Moreover, the compatibility with TX aggregation mechanisms that allow the compression of the BC size is maintained.

Mobile health applications (mHealth apps for short) are being increasingly adopted in the healthcare sector, enabling stakeholders such as governments, health units, medics, and patients, to utilize health services in a pervasive manner. Despite having several known benefits, mHealth apps entail significant security and privacy challenges that can lead to data breaches with serious social, legal, and financial consequences. This research presents an empirical investigation about security awareness of end-users of mHealth apps that are available on major mobile platforms, including Android and iOS. We collaborated with two mHealth providers in Saudi Arabia to survey 101 end-users, investigating their security awareness about (i) existing and desired security features, (ii) security related issues, and (iii) methods to improve security knowledge. Findings indicate that majority of the end-users are aware of the existing security features provided by the apps (e.g., restricted app permissions); however, they desire usable security (e.g., biometric authentication) and are concerned about privacy of their health information (e.g., data anonymization). End-users suggested that protocols such as session timeout or Two-factor authentication (2FA) positively impact security but compromise usability of the app. Security-awareness via social media, peer guidance, or training from app providers can increase end-users trust in mHealth apps. This research investigates human-centric knowledge based on empirical evidence and provides a set of guidelines to develop secure and usable mHealth apps.

We study theoretically subradiant states in the array of atoms coupled to photons propagating in a one-dimensional waveguide focusing on the strongly interacting many-body regime with large excitation fill factor $f$. We introduce a generalized many-body entropy of entanglement based on exact numerical diagonalization followed by a high-order singular value decomposition. This approach has allowed us to visualize and understand the structure of a many-body quantum state. We reveal the breakdown of fermionized subradiant states with increase of $f$ with emergence of short-ranged dimerized antiferromagnetic correlations at the critical point $f=1/2$ and the complete disappearance of subradiant states at $f>1/2$.

A graph $G$ is a prime distance graph (respectively, a 2-odd graph) if its vertices can be labeled with distinct integers such that for any two adjacent vertices, the difference of their labels is prime (either 2 or odd). We prove that trees, cycles, and bipartite graphs are prime distance graphs, and that Dutch windmill graphs and paper mill graphs are prime distance graphs if and only if the Twin Prime Conjecture and dePolignac's Conjecture are true, respectively. We give a characterization of 2-odd graphs in terms of edge colorings, and we use this characterization to determine which circulant graphs of the form $Circ(n, \{1,k\})$ are 2-odd and to prove results on circulant prime distance graphs.

We consider the stochastic shortest path planning problem in MDPs, i.e., the problem of designing policies that ensure reaching a goal state from a given initial state with minimum accrued cost. In order to account for rare but important realizations of the system, we consider a nested dynamic coherent risk total cost functional rather than the conventional risk-neutral total expected cost. Under some assumptions, we show that optimal, stationary, Markovian policies exist and can be found via a special Bellman's equation. We propose a computational technique based on difference convex programs (DCPs) to find the associated value functions and therefore the risk-averse policies. A rover navigation MDP is used to illustrate the proposed methodology with conditional-value-at-risk (CVaR) and entropic-value-at-risk (EVaR) coherent risk measures.

Using a martingale concentration inequality, concentration bounds `from time $n_0$ on' are derived for stochastic approximation algorithms with contractive maps and both martingale difference and Markov noises. These are applied to reinforcement learning algorithms, in particular to asynchronous Q-learning and TD(0).

The properties of quasar-host galaxies might be determined by the growth and feedback of their supermassive (SMBH, $10^{8-10}$ M$_{\odot}$) black holes. We investigate such connection with a suite of cosmological simulations of massive (halo mass $\approx 10^{12}$ M$_{\odot}$) galaxies at $z\simeq 6$ which include a detailed sub-grid multiphase gas and accretion model. BH seeds of initial mass $10^5$ M$_{\odot}$ grow mostly by gas accretion, and become SMBH by $z=6$ setting on the observed $M_{\rm BH} - M_{\star}$ relation without the need for a boost factor. Although quasar feedback crucially controls the SMBH growth, its impact on the properties of the host galaxy at $z=6$ is negligible. In our model, quasar activity can both quench (via gas heating) or enhance (by ISM over-pressurization) star formation. However, we find that the star formation history is insensitive to such modulation as it is largely dominated, at least at $z>6$, by cold gas accretion from the environment that cannot be hindered by the quasar energy deposition. Although quasar-driven outflows can achieve velocities $> 1000~\rm km~s^{-1}$, only $\approx 4$% of the outflowing gas mass can actually escape from the host galaxy. These findings are only loosely constrained by available data, but can guide observational campaigns searching for signatures of quasar feedback in early galaxies.

The existence of $1$-factorizations of an infinite complete equipartite graph $K_m[n]$ (with $m$ parts of size $n$) admitting a vertex-regular automorphism group $G$ is known only when $n=1$ and $m$ is countable (that is, for countable complete graphs) and, in addition, $G$ is a finitely generated abelian group $G$ of order $m$. In this paper, we show that a vertex-regular $1$-factorization of $K_m[n]$ under the group $G$ exists if and only if $G$ has a subgroup $H$ of order $n$ whose index in $G$ is $m$. Furthermore, we provide a sufficient condition for an infinite Cayley graph to have a regular $1$-factorization. Finally, we construct 1-factorizations that contain a given subfactorization, both having a vertex-regular automorphism group.

This paper aims at solving an optimal control problem governed by an anisotropic Allen-Cahn equation numerically. Therefore we first prove the Fr\'echet differentiability of an in time discretized parabolic control problem under certain assumptions on the involved quasilinearity and formulate the first order necessary conditions. As a next step, since the anisotropies are in general not smooth enough, the convergence behavior of the optimal controls are studied for a sequence of (smooth) approximations of the former quasilinear term. In addition the simultaneous limit in the approximation and the time step size is considered. For a class covering a large variety of anisotropies we introduce a certain regularization and show the previously formulated requirements. Finally, a trust region Newton solver is applied to various anisotropies and configurations, and numerical evidence for mesh independent behavior and convergence with respect to regularization is presented.

We answer in negative to the following question of Boris Mitjagin: Is it true that a product of two nuclear operators in Banach spaces can be factored through a trace class operator in a Hilbert space?

We propose a new architecture for diacritics restoration based on contextualized embeddings, namely BERT, and we evaluate it on 12 languages with diacritics. Furthermore, we conduct a detailed error analysis on Czech, a morphologically rich language with a high level of diacritization. Notably, we manually annotate all mispredictions, showing that roughly 44% of them are actually not errors, but either plausible variants (19%), or the system corrections of erroneous data (25%). Finally, we categorize the real errors in detail. We release the code at https://github.com/ufal/bert-diacritics-restoration.

Knowledge distillation transfers knowledge from the teacher network to the student one, with the goal of greatly improving the performance of the student network. Previous methods mostly focus on proposing feature transformation and loss functions between the same level's features to improve the effectiveness. We differently study the factor of connection path cross levels between teacher and student networks, and reveal its great importance. For the first time in knowledge distillation, cross-stage connection paths are proposed. Our new review mechanism is effective and structurally simple. Our finally designed nested and compact framework requires negligible computation overhead, and outperforms other methods on a variety of tasks. We apply our method to classification, object detection, and instance segmentation tasks. All of them witness significant student network performance improvement. Code is available at https://github.com/Jia-Research-Lab/ReviewKD

Cell-free massive MIMO systems consist of many distributed access points with simple components that jointly serve the users. In millimeter wave bands, only a limited set of predetermined beams can be supported. In a network that consolidates these technologies, downlink analog beam selection stands as a challenging task for the network sum-rate maximization. Low-cost digital filters can improve the network sum-rate further. In this work, we propose low-cost joint designs of analog beam selection and digital filters. The proposed joint designs achieve significantly higher sum-rates than the disjoint design benchmark. Supervised machine learning (ML) algorithms can efficiently approximate the input-output mapping functions of the beam selection decisions of the joint designs with low computational complexities. Since the training of ML algorithms is performed off-line, we propose a well-constructed joint design that combines multiple initializations, iterations, and selection features, as well as beam conflict control, i.e., the same beam cannot be used for multiple users. The numerical results indicate that ML algorithms can retain 99-100% of the original sum-rate results achieved by the proposed well-constructed designs.

Classifying the sub-categories of an object from the same super-category (e.g., bird) in a fine-grained visual classification (FGVC) task highly relies on mining multiple discriminative features. Existing approaches mainly tackle this problem by introducing attention mechanisms to locate the discriminative parts or feature encoding approaches to extract the highly parameterized features in a weakly-supervised fashion. In this work, we propose a lightweight yet effective regularization method named Channel DropBlock (CDB), in combination with two alternative correlation metrics, to address this problem. The key idea is to randomly mask out a group of correlated channels during training to destruct features from co-adaptations and thus enhance feature representations. Extensive experiments on three benchmark FGVC datasets show that CDB effectively improves the performance.

We show that for a model complete strongly minimal theory whose pregeometry is flat, the recursive spectrum (SRM($T$)) is either of the form $[0,\alpha)$ for $\alpha\in \omega+2$ or $[0,n]\cup\{\omega\}$ for $n\in \omega$, or $\{\omega\}$, or contained in $\{0,1,2\}$. Combined with previous results, this leaves precisely 4 sets for which it is not yet determined whether each is the spectrum of a model complete strongly minimal theory with a flat pregeometry.

Data-driven methods for battery lifetime prediction are attracting increasing attention for applications in which the degradation mechanisms are poorly understood and suitable training sets are available. However, while advanced machine learning and deep learning methods promise high performance with minimal data preprocessing, simpler linear models with engineered features often achieve comparable performance, especially for small training sets, while also providing physical and statistical interpretability. In this work, we use a previously published dataset to develop simple, accurate, and interpretable data-driven models for battery lifetime prediction. We first present the "capacity matrix" concept as a compact representation of battery electrochemical cycling data, along with a series of feature representations. We then create a number of univariate and multivariate models, many of which achieve comparable performance to the highest-performing models previously published for this dataset. These models also provide insights into the degradation of these cells. Our approaches can be used both to quickly train models for a new dataset and to benchmark the performance of more advanced machine learning methods.

Willems' fundamental lemma asserts that all trajectories of a linear time-invariant system can be obtained from a finite number of measured ones, assuming that controllability and a persistency of excitation condition hold. We show that these two conditions can be relaxed. First, we prove that the controllability condition can be replaced by a condition on the controllable subspace, unobservable subspace, and a certain subspace associated with the measured trajectories. Second, we prove that the persistency of excitation requirement can be relaxed if the degree of a certain minimal polynomial is tightly bounded. Our results show that data-driven predictive control using online data is equivalent to model predictive control, even for uncontrollable systems. Moreover, our results significantly reduce the amount of data needed in identifying homogeneous multi-agent systems.

This article describe globular weak $(n,\infty)$-transformations ($n\in\mathbb{N}$) in the sense of Grothendieck, i.e for each $n\in\mathbb{N}$ we build a coherator $\Theta^{\infty}_{\mathbb{M}^n}$ which sets models are globular weak $(n,\infty)$-transformations. A natural globular filtration emerges from these coherators.

In this article, logical concepts are defined using the internal syntactic and semantic structure of language. For a first-order language, it has been shown that its logical constants are connectives and a certain type of quantifiers for which the universal and existential quantifiers form a functionally complete set of quantifiers. Neither equality nor cardinal quantifiers belong to the logical constants of a first-order language.

To safely deploy autonomous vehicles, onboard perception systems must work reliably at high accuracy across a diverse set of environments and geographies. One of the most common techniques to improve the efficacy of such systems in new domains involves collecting large labeled datasets, but such datasets can be extremely costly to obtain, especially if each new deployment geography requires additional data with expensive 3D bounding box annotations. We demonstrate that pseudo-labeling for 3D object detection is an effective way to exploit less expensive and more widely available unlabeled data, and can lead to performance gains across various architectures, data augmentation strategies, and sizes of the labeled dataset. Overall, we show that better teacher models lead to better student models, and that we can distill expensive teachers into efficient, simple students. Specifically, we demonstrate that pseudo-label-trained student models can outperform supervised models trained on 3-10 times the amount of labeled examples. Using PointPillars [24], a two-year-old architecture, as our student model, we are able to achieve state of the art accuracy simply by leveraging large quantities of pseudo-labeled data. Lastly, we show that these student models generalize better than supervised models to a new domain in which we only have unlabeled data, making pseudo-label training an effective form of unsupervised domain adaptation.

In this work, we propose a Bayesian statistical model to simultaneously characterize two or more social networks defined over a common set of actors. The key feature of the model is a hierarchical prior distribution that allows us to represent the entire system jointly, achieving a compromise between dependent and independent networks. Among others things, such a specification easily allows us to visualize multilayer network data in a low-dimensional Euclidean space, generate a weighted network that reflects the consensus affinity between actors, establish a measure of correlation between networks, assess cognitive judgements that subjects form about the relationships among actors, and perform clustering tasks at different social instances. Our model's capabilities are illustrated using several real-world data sets, taking into account different types of actors, sizes, and relations.

We define partial quasi-morphisms on the group of Hamiltonian diffeomorphisms of the cotangent bundle using the spectral invariants in Lagrangian Floer homology with conormal boundary conditions, where the product compatible with the PSS isomorphism and the homological intersection product is lacking.

In this paper, we propose a data-driven method to discover multiscale chemical reactions governed by the law of mass action. First, we use a single matrix to represent the stoichiometric coefficients for both the reactants and products in a system without catalysis reactions. The negative entries in the matrix denote the stoichiometric coefficients for the reactants and the positive ones for the products. Second, we find that the conventional optimization methods usually get stuck in the local minima and could not find the true solution in learning the multiscale chemical reactions. To overcome this difficulty, we propose a partial-parameters-freezing (PPF) technique to progressively determine the network parameters by using the fact that the stoichiometric coefficients are integers. With such a technique, the dimension of the searching space is gradually reduced in the training process and the global mimina can be eventually obtained. Several numerical experiments including the classical Michaelis-Menten kinetics, the hydrogen oxidation reactions, and the simplified GRI-3.0 mechanism verify the good performance of our algorithm in learning the multiscale chemical reactions. The code is available at \url{https://github.com/JuntaoHuang/multiscale-chemical-reaction}.

Approximate nearest-neighbor search is a fundamental algorithmic problem that continues to inspire study due its essential role in numerous contexts. In contrast to most prior work, which has focused on point sets, we consider nearest-neighbor queries against a set of line segments in $\mathbb{R}^d$, for constant dimension $d$. Given a set $S$ of $n$ disjoint line segments in $\mathbb{R}^d$ and an error parameter $\varepsilon > 0$, the objective is to build a data structure such that for any query point $q$, it is possible to return a line segment whose Euclidean distance from $q$ is at most $(1+\varepsilon)$ times the distance from $q$ to its nearest line segment. We present a data structure for this problem with storage $O((n^2/\varepsilon^{d}) \log (\Delta/\varepsilon))$ and query time $O(\log (\max(n,\Delta)/\varepsilon))$, where $\Delta$ is the spread of the set of segments $S$. Our approach is based on a covering of space by anisotropic elements, which align themselves according to the orientations of nearby segments.

We construct a large class of projective threefolds with one node (aka non-degenerate quadratic singularity) such that their small resolutions are not projective.

Data heterogeneity has been identified as one of the key features in federated learning but often overlooked in the lens of robustness to adversarial attacks. This paper focuses on characterizing and understanding its impact on backdooring attacks in federated learning through comprehensive experiments using synthetic and the LEAF benchmarks. The initial impression driven by our experimental results suggests that data heterogeneity is the dominant factor in the effectiveness of attacks and it may be a redemption for defending against backdooring as it makes the attack less efficient, more challenging to design effective attack strategies, and the attack result also becomes less predictable. However, with further investigations, we found data heterogeneity is more of a curse than a redemption as the attack effectiveness can be significantly boosted by simply adjusting the client-side backdooring timing. More importantly,data heterogeneity may result in overfitting at the local training of benign clients, which can be utilized by attackers to disguise themselves and fool skewed-feature based defenses. In addition, effective attack strategies can be made by adjusting attack data distribution. Finally, we discuss the potential directions of defending the curses brought by data heterogeneity. The results and lessons learned from our extensive experiments and analysis offer new insights for designing robust federated learning methods and systems

Unsupervised domain adaptive classifcation intends to improve the classifcation performance on unlabeled target domain. To alleviate the adverse effect of domain shift, many approaches align the source and target domains in the feature space. However, a feature is usually taken as a whole for alignment without explicitly making domain alignment proactively serve the classifcation task, leading to sub-optimal solution. In this paper, we propose an effective Task-oriented Alignment (ToAlign) for unsupervised domain adaptation (UDA). We study what features should be aligned across domains and propose to make the domain alignment proactively serve classifcation by performing feature decomposition and alignment under the guidance of the prior knowledge induced from the classifcation task itself. Particularly, we explicitly decompose a feature in the source domain into a task-related/discriminative feature that should be aligned, and a task-irrelevant feature that should be avoided/ignored, based on the classifcation meta-knowledge. Extensive experimental results on various benchmarks (e.g., Offce-Home, Visda-2017, and DomainNet) under different domain adaptation settings demonstrate the effectiveness of ToAlign which helps achieve the state-of-the-art performance. The code is publicly available at https://github.com/microsoft/UDA

The design space for inertial confinement fusion (ICF) experiments is vast and experiments are extremely expensive. Researchers rely heavily on computer simulations to explore the design space in search of high-performing implosions. However, ICF multiphysics codes must make simplifying assumptions, and thus deviate from experimental measurements for complex implosions. For more effective design and investigation, simulations require input from past experimental data to better predict future performance. In this work, we describe a cognitive simulation method for combining simulation and experimental data into a common, predictive model. This method leverages a machine learning technique called transfer learning, the process of taking a model trained to solve one task, and partially retraining it on a sparse dataset to solve a different, but related task. In the context of ICF design, neural network models trained on large simulation databases and partially retrained on experimental data, producing models that are far more accurate than simulations alone. We demonstrate improved model performance for a range of ICF experiments at the National Ignition Facility, and predict the outcome of recent experiments with less than ten percent error for several key observables. We discuss how the methods might be used to carry out a data-driven experimental campaign to optimize performance, illustrating the key product -- models that become increasingly accurate as data is acquired.

This paper develops simple feed-forward neural networks that achieve the universal approximation property for all continuous functions with a fixed finite number of neurons. These neural networks are simple because they are designed with a simple and computable continuous activation function $\sigma$ leveraging a triangular-wave function and a softsign function. We prove that $\sigma$-activated networks with width $36d(2d+1)$ and depth $11$ can approximate any continuous function on a $d$-dimensioanl hypercube within an arbitrarily small error. Hence, for supervised learning and its related regression problems, the hypothesis space generated by these networks with a size not smaller than $36d(2d+1)\times 11$ is dense in the space of continuous functions. Furthermore, classification functions arising from image and signal classification are in the hypothesis space generated by $\sigma$-activated networks with width $36d(2d+1)$ and depth $12$, when there exist pairwise disjoint closed bounded subsets of $\mathbb{R}^d$ such that the samples of the same class are located in the same subset.

Adversarial algorithms have shown to be effective against neural networks for a variety of tasks. Some adversarial algorithms perturb all the pixels in the image minimally for the image classification task in image classification. In contrast, some algorithms perturb few pixels strongly. However, very little information is available regarding why these adversarial samples so diverse from each other exist. Recently, Vargas et al. showed that the existence of these adversarial samples might be due to conflicting saliency within the neural network. We test this hypothesis of conflicting saliency by analysing the Saliency Maps (SM) and Gradient-weighted Class Activation Maps (Grad-CAM) of original and few different types of adversarial samples. We also analyse how different adversarial samples distort the attention of the neural network compared to original samples. We show that in the case of Pixel Attack, perturbed pixels either calls the network attention to themselves or divert the attention from them. Simultaneously, the Projected Gradient Descent Attack perturbs pixels so that intermediate layers inside the neural network lose attention for the correct class. We also show that both attacks affect the saliency map and activation maps differently. Thus, shedding light on why some defences successful against some attacks remain vulnerable against other attacks. We hope that this analysis will improve understanding of the existence and the effect of adversarial samples and enable the community to develop more robust neural networks.

Understanding and mitigating loss channels due to two-level systems (TLS) is one of the main cornerstones in the quest of realizing long photon lifetimes in superconducting quantum circuits. Typically, the TLS to which a circuit couples are modeled as a large bath without any coherence. Here we demonstrate that the coherence of TLS has to be considered to accurately describe the ring-down dynamics of a coaxial quarter-wave resonator with an internal quality factor of $0.5\times10^9$ at the single-photon level. The transient analysis reveals effective non-Markovian dynamics of the combined TLS and cavity system, which we can accurately fit by introducing a comprehensive TLS model. The fit returns an average coherence time of around $T_2^*\approx0.3\,\mathrm{\mu s}$ for a total of $N\approx10^{9}$ TLS with power-law distributed coupling strengths. Despite the shortly coherent TLS excitations, we observe long-term effects on the cavity decay due to coherent elastic scattering between the resonator field and the TLS. Moreover, this model provides an accurate prediction of the internal quality factor's temperature dependence.

Weakly nonlinear internal wave-wave interaction is a key mechanism that cascades energy from large to small scales, leading to ocean turbulence and mixing. Oceans typically have a non-uniform density stratification profile; moreover, submarine topography leads to a spatially varying ocean depth ($h$). Under these conditions and assuming mild-slope bathymetry, we employ multiple-scale analysis to derive the wave amplitude equations for triadic- and self-interactions. The waves are assumed to have a slowly (rapidly) varying amplitude (phase) in space and time. For uniform stratifications, the horizontal wavenumber ($k$) condition for waves ($1$,$2$,$3$), given by ${k}_{(1,a)}+{k}_{(2,b)}+{k}_{(3,c)}=0$, is unaffected as $h$ is varied, where $(a,b,c)$ denote the modenumber. Moreover, the nonlinear coupling coefficients (NLC) are proportional to $1/h^2$, implying that triadic waves grow faster while travelling up a seamount. For non-uniform stratifications, triads that do not satisfy the condition $a=b=c$ may not satisfy the horizontal wavenumber condition as $h$ is varied, and unlike uniform stratification, the NLC may not decrease (increase) monotonically with increasing (decreasing) $h$. NLC, and hence wave growth rates for both triads and self-interactions, can also vary rapidly with $h$. The most unstable daughter wave combination of a triad with a mode-1 parent wave can also change for relatively small changes in $h$. We also investigate higher-order self-interactions in the presence of a monochromatic, small amplitude bathymetry; here the bathymetry behaves as a zero frequency wave. We derive the amplitude evolution equations and show that higher-order self-interactions might be a viable mechanism of energy cascade.

We give a $0.5368$-competitive algorithm for edge-weighted online bipartite matching. Prior to our work, the best competitive ratio was $0.5086$ due to Fahrbach, Huang, Tao, and Zadimoghaddam (FOCS 2020). They achieved their breakthrough result by developing a subroutine called \emph{online correlated selection} (OCS) which takes as input a sequence of pairs and selects one item from each pair. Importantly, the selections the OCS makes are negatively correlated. We achieve our result by defining \emph{multiway} OCSes which receive arbitrarily many elements at each step, rather than just two. In addition to better competitive ratios, our formulation allows for a simpler reduction from edge-weighted online bipartite matching to OCSes. While Fahrbach et al. used a factor-revealing linear program to optimize the competitive ratio, our analysis directly connects the competitive ratio to the parameters of the multiway OCS. Finally, we show that the formulation of Farhbach et al. can achieve a competitive ratio of at most $0.5239$, confirming that multiway OCSes are strictly more powerful.

Let $\Omega\subset \mathbb{C}^n$ be a smooth bounded pseudoconvex domain and $A^2 (\Omega)$ denote its Bergman space. Let $P:L^2(\Omega)\longrightarrow A^2(\Omega)$ be the Bergman projection. For a measurable $\varphi:\Omega\longrightarrow \Omega$, the projected composition operator is defined by $(K_\varphi f)(z) = P(f \circ \varphi)(z), z \in\Omega, f\in A^2 (\Omega).$ In 1994, Rochberg studied boundedness of $K_\varphi$ on the Hardy space of the unit disk and obtained different necessary or sufficient conditions for boundedness of $K_\varphi$. In this paper we are interested in projected composition operators on Bergman spaces on pseudoconvex domains. We study boundedness of this operator under the smoothness assumptions on the symbol $\varphi$ on $\overline\Omega$.

A new mechanism for the internal heating of ultra-short-period planets is proposed based on the gravitational perturbation by a non-axisymmetric quadrupole moment of their host stars. Such a quadrupole is due to the magnetic flux tubes in the stellar convection zone, unevenly distributed in longitude and persisting for many stellar rotations as observed in young late-type stars. The rotation period of the host star evolves from its shortest value on the zero-age main sequence to longer periods due to the loss of angular momentum through a magnetized wind. If the stellar rotation period comes close to twice the orbital period of the planet, the quadrupole leads to a spin-orbit resonance that excites oscillations of the star-planet separation. As a consequence, a strong tidal dissipation is produced inside the planet. We illustrate the operation of the mechanism by modeling the evolution of the stellar rotation and of the innermost planetary orbit in the cases of CoRoT-7, Kepler-78, and K2-141 whose present orbital periods range between 0.28 and 0.85 days. If the spin-orbit resonance occurs, the maximum power dissipated inside the planets ranges between $10^{18}$ and $10^{19}$ W, while the total dissipated energy is of the order of $10^{30}-10^{32}$ J over a time interval as short as $(1-4.5) \times 10^{4}$ yr. Such a huge heating over a so short time interval produces a complete melting of the planetary interiors and may shut off their hydromagnetic dynamos. These may initiate a successive phase of intense internal heating owing to unipolar magnetic star-planet interactions and affect the composition and the escape of their atmospheres, producing effects that could be observable during the entire lifetime of the planets [abridged abstract].

This paper presents a fully convolutional scene graph generation (FCSGG) model that detects objects and relations simultaneously. Most of the scene graph generation frameworks use a pre-trained two-stage object detector, like Faster R-CNN, and build scene graphs using bounding box features. Such pipeline usually has a large number of parameters and low inference speed. Unlike these approaches, FCSGG is a conceptually elegant and efficient bottom-up approach that encodes objects as bounding box center points, and relationships as 2D vector fields which are named as Relation Affinity Fields (RAFs). RAFs encode both semantic and spatial features, and explicitly represent the relationship between a pair of objects by the integral on a sub-region that points from subject to object. FCSGG only utilizes visual features and still generates strong results for scene graph generation. Comprehensive experiments on the Visual Genome dataset demonstrate the efficacy, efficiency, and generalizability of the proposed method. FCSGG achieves highly competitive results on recall and zero-shot recall with significantly reduced inference time.

With extreme weather events becoming more common, the risk posed by surface water flooding is ever increasing. In this work we propose a model, and associated Bayesian inference scheme, for generating probabilistic (high-resolution short-term) forecasts of localised precipitation. The parametrisation of our underlying hierarchical dynamic spatio-temporal model is motivated by a forward-time, centred-space finite difference solution to a collection of stochastic partial differential equations, where the main driving forces are advection and diffusion. Observations from both weather radar and ground based rain gauges provide information from which we can learn about the likely values of the (latent) precipitation field in addition to other unknown model parameters. Working in the Bayesian paradigm provides a coherent framework for capturing uncertainty both in the underlying model parameters and also in our forecasts. Further, appealing to simulation based (MCMC) sampling yields a straightforward solution to handling zeros, treated as censored observations, via data augmentation. Both the underlying state and the observations are of moderately large dimension ($\mathcal{O}(10^4)$ and $\mathcal{O}(10^3)$ respectively) and this renders standard inference approaches computationally infeasible. Our solution is to embed the ensemble Kalman smoother within a Gibbs sampling scheme to facilitate approximate Bayesian inference in reasonable time. Both the methodology and the effectiveness of our posterior sampling scheme are demonstrated via simulation studies and also by a case study of real data from the Urban Observatory project based in Newcastle upon Tyne, UK.

Continuous variable measurement-based quantum computation on cluster states has in recent years shown great potential for scalable, universal, and fault-tolerant quantum computation when combined with the Gottesman-Kitaev-Preskill (GKP) code and quantum error correction. However, no complete fault-tolerant architecture exists that includes everything from cluster state generation with finite squeezing to gate implementations with realistic noise and error correction. In this work, we propose a simple architecture for the preparation of a cluster state in three dimensions in which gates by gate teleportation can be efficiently implemented. To accommodate scalability, we propose architectures that allow for both spatial and temporal multiplexing, with the temporal encoded version requiring as little as two squeezed light sources. Due to its three-dimensional structure, the architecture supports topological qubit error correction, while GKP error correction is efficiently realized within the architecture by teleportation. To validate fault-tolerance, the architecture is simulated using surface-GKP codes, including noise from GKP-states as well as gate noise caused by finite squeezing in the cluster state. We find a fault-tolerant squeezing threshold of 12.7 dB with room for further improvement.

In high energy exclusive processes involving leptons, QED corrections can be sensitive to infrared scales like the lepton mass and the soft photon energy cut, resulting in large logarithms that need to be resummed to all order in $\alpha$. When considering the ratio of the exclusive processes between two different lepton flavors, the ratio $R$ can be expressed in terms of factorized functions in the decoupled leptonic sectors. While some of the functional terms cancel, there remain the large logarithms due to the lepton mass difference and the energy cut. This factorization process can be universally applied to the exclusive processes such as $Z\to l^+l^-$ and $B^-\to l^-\bar{\nu}_l$, where the resummed result in the ratio gives significant deviations from the naive expectation from lepton universality.

A hyperbolic group acts by homeomorphisms on its Gromov boundary. We show that if this boundary is a topological n-sphere the action is topologically stable in the dynamical sense: any nearby action is semi-conjugate to the standard boundary action.

In this paper we consider the problem of pointwise determining the fibres of the flat unitary subbundle of a PVHS of weight one. Starting from the associated Higgs field, and assuming the base has dimension $1$, we construct a family of (smooth but possibly non-holomorphic) morphisms of vector bundles with the property that the intersection of their kernels at a general point is the fibre of the flat subbundle. We explore the first one of these morphisms in the case of a geometric PVHS arising from a family of smooth projective curves, showing that it acts as the cup-product with some sort of "second-order Kodaira-Spencer class" which we introduce, and check in the case of a family of smooth plane curves that this additional condition is non-trivial.

Multi-segment reconstruction (MSR) is the problem of estimating a signal given noisy partial observations. Here each observation corresponds to a randomly located segment of the signal. While previous works address this problem using template or moment-matching, in this paper we address MSR from an unsupervised adversarial learning standpoint, named MSR-GAN. We formulate MSR as a distribution matching problem where the goal is to recover the signal and the probability distribution of the segments such that the distribution of the generated measurements following a known forward model is close to the real observations. This is achieved once a min-max optimization involving a generator-discriminator pair is solved. MSR-GAN is mainly inspired by CryoGAN [1]. However, in MSR-GAN we no longer assume the probability distribution of the latent variables, i.e. segment locations, is given and seek to recover it alongside the unknown signal. For this purpose, we show that the loss at the generator side originally is non-differentiable with respect to the segment distribution. Thus, we propose to approximate it using Gumbel-Softmax reparametrization trick. Our proposed solution is generalizable to a wide range of inverse problems. Our simulation results and comparison with various baselines verify the potential of our approach in different settings.

We compare the radial profiles of the specific star formation rate (sSFR) in a sample of 169 star-forming galaxies in close pairs with those of mass-matched control galaxies in the SDSS-IV MaNGA survey. We find that the sSFR is centrally enhanced (within one effective radius) in interacting galaxies by ~0.3 dex and that there is a weak sSFR suppression in the outskirts of the galaxies of ~0.1 dex. We stack the differences profiles for galaxies in five stellar mass bins between log(M/Mstar) = 9.0-11.5 and find that the sSFR enhancement has no dependence on the stellar mass. The same result is obtained when the comparison galaxies are matched to each paired galaxy in both stellar mass and redshift. In addition, we find that that the sSFR enhancement is elevated in pairs with nearly equal masses and closer projected separations, in agreement with previous work based on single-fiber spectroscopy. We also find that the sSFR offsets in the outskirts of the paired galaxies are dependent on whether the galaxy is the more massive or less massive companion in the pair. The more massive companion experiences zero to a positive sSFR enhancement while the less massive companion experiences sSFR suppression in their outskirts. Our results illustrate the complex tidal effects on star formation in closely paired galaxies.

Solving Partially Observable Markov Decision Processes (POMDPs) is hard. Learning optimal controllers for POMDPs when the model is unknown is harder. Online learning of optimal controllers for unknown POMDPs, which requires efficient learning using regret-minimizing algorithms that effectively tradeoff exploration and exploitation, is even harder, and no solution exists currently. In this paper, we consider infinite-horizon average-cost POMDPs with unknown transition model, though a known observation model. We propose a natural posterior sampling-based reinforcement learning algorithm (PSRL-POMDP) and show that it achieves a regret bound of $O(\log T)$, where $T$ is the time horizon, when the parameter set is finite. In the general case (continuous parameter set), we show that the algorithm achieves $O (T^{2/3})$ regret under two technical assumptions. To the best of our knowledge, this is the first online RL algorithm for POMDPs and has sub-linear regret.

Let $D$ be a $k$-regular bipartite tournament on $n$ vertices. We show that, for every $p$ with $2 \le p \le n/2-2$, $D$ has a cycle $C$ of length $2p$ such that $D \setminus C$ is hamiltonian unless $D$ is isomorphic to the special digraph $F_{4k}$. This statement was conjectured by Manoussakis, Song and Zhang [K. Zhang, Y. Manoussakis, and Z. Song. Complementary cycles containing a fixed arc in diregular bipartite tournaments. Discrete Mathematics, 133(1-3):325--328,1994]. In the same paper, the conjecture was proved for $p=2$ and more recently Bai, Li and He gave a proof for $p=3$ [Y. Bai, H. Li, and W. He. Complementary cycles in regular bipartite tournaments. Discrete Mathematics, 333:14--27, 2014].

Missing time-series data is a prevalent practical problem. Imputation methods in time-series data often are applied to the full panel data with the purpose of training a model for a downstream out-of-sample task. For example, in finance, imputation of missing returns may be applied prior to training a portfolio optimization model. Unfortunately, this practice may result in a look-ahead-bias in the future performance on the downstream task. There is an inherent trade-off between the look-ahead-bias of using the full data set for imputation and the larger variance in the imputation from using only the training data. By connecting layers of information revealed in time, we propose a Bayesian posterior consensus distribution which optimally controls the variance and look-ahead-bias trade-off in the imputation. We demonstrate the benefit of our methodology both in synthetic and real financial data.

Recent studies have witnessed that self-supervised methods based on view synthesis obtain clear progress on multi-view stereo (MVS). However, existing methods rely on the assumption that the corresponding points among different views share the same color, which may not always be true in practice. This may lead to unreliable self-supervised signal and harm the final reconstruction performance. To address the issue, we propose a framework integrated with more reliable supervision guided by semantic co-segmentation and data-augmentation. Specially, we excavate mutual semantic from multi-view images to guide the semantic consistency. And we devise effective data-augmentation mechanism which ensures the transformation robustness by treating the prediction of regular samples as pseudo ground truth to regularize the prediction of augmented samples. Experimental results on DTU dataset show that our proposed methods achieve the state-of-the-art performance among unsupervised methods, and even compete on par with supervised methods. Furthermore, extensive experiments on Tanks&Temples dataset demonstrate the effective generalization ability of the proposed method.

Nearly a decade ago, Azrieli and Shmaya introduced the class of $\lambda$-Lipschitz games in which every player's payoff function is $\lambda$-Lipschitz with respect to the actions of the other players. They showed that such games admit $\epsilon$-approximate pure Nash equilibria for certain settings of $\epsilon$ and $\lambda$. They left open, however, the question of how hard it is to find such an equilibrium. In this work, we develop a query-efficient reduction from more general games to Lipschitz games. We use this reduction to show a query lower bound for any randomized algorithm finding $\epsilon$-approximate pure Nash equilibria of $n$-player, binary-action, $\lambda$-Lipschitz games that is exponential in $\frac{n\lambda}{\epsilon}$. In addition, we introduce ``Multi-Lipschitz games,'' a generalization involving player-specific Lipschitz values, and provide a reduction from finding equilibria of these games to finding equilibria of Lipschitz games, showing that the value of interest is the sum of the individual Lipschitz parameters. Finally, we provide an exponential lower bound on the deterministic query complexity of finding $\epsilon$-approximate correlated equilibria of $n$-player, $m$-action, $\lambda$-Lipschitz games for strong values of $\epsilon$, motivating the consideration of explicitly randomized algorithms in the above results. Our proof is arguably simpler than those previously used to show similar results.

The WIMP proposed here yields the observed abundance of dark matter, and is consistent with the current limits from direct detection, indirect detection, and collider experiments, if its mass is $\sim 72$ GeV/$c^2$. It is also consistent with analyses of the gamma rays observed by Fermi-LAT from the Galactic center (and other sources), and of the antiprotons observed by AMS-02, in which the excesses are attributed to dark matter annihilation. These successes are shared by the inert doublet model (IDM), but the phenomenology is very different: The dark matter candidate of the IDM has first-order gauge couplings to other new particles, whereas the present candidate does not. In addition to indirect detection through annihilation products, it appears that the present particle can be observed in the most sensitive direct-detection and collider experiments currently being planned.

It has been shown that the performance of neural machine translation (NMT) drops starkly in low-resource conditions, often requiring large amounts of auxiliary data to achieve competitive results. An effective method of generating auxiliary data is back-translation of target language sentences. In this work, we present a case study of Tigrinya where we investigate several back-translation methods to generate synthetic source sentences. We find that in low-resource conditions, back-translation by pivoting through a higher-resource language related to the target language proves most effective resulting in substantial improvements over baselines.

This paper proposes architectures that facilitate the extrapolation of emotional expressions in deep neural network (DNN)-based text-to-speech (TTS). In this study, the meaning of "extrapolate emotional expressions" is to borrow emotional expressions from others, and the collection of emotional speech uttered by target speakers is unnecessary. Although a DNN has potential power to construct DNN-based TTS with emotional expressions and some DNN-based TTS systems have demonstrated satisfactory performances in the expression of the diversity of human speech, it is necessary and troublesome to collect emotional speech uttered by target speakers. To solve this issue, we propose architectures to separately train the speaker feature and the emotional feature and to synthesize speech with any combined quality of speakers and emotions. The architectures are parallel model (PM), serial model (SM), auxiliary input model (AIM), and hybrid models (PM&AIM and SM&AIM). These models are trained through emotional speech uttered by few speakers and neutral speech uttered by many speakers. Objective evaluations demonstrate that the performances in the open-emotion test provide insufficient information. They make a comparison with those in the closed-emotion test, but each speaker has their own manner of expressing emotion. However, subjective evaluation results indicate that the proposed models could convey emotional information to some extent. Notably, the PM can correctly convey sad and joyful emotions at a rate of >60%.

The baroclinic annular mode (BAM) is a leading-order mode of the eddy-kinetic energy in the Southern Hemisphere exhibiting. oscillatory behavior at intra-seasonal time scales. The oscillation mechanism has been linked to transient eddy-mean flow interactions that remain poorly understood. Here we demonstrate that the finite memory effect in eddy-heat flux dependence on the large-scale flow can explain the origin of the BAM's oscillatory behavior. We represent the eddy memory effect by a delayed integral kernel that leads to a generalized Langevin equation for the planetary-scale heat equation. Using a mathematical framework for the interactions between planetary and synoptic-scale motions, we derive a reduced dynamical model of the BAM - a stochastically-forced oscillator with a period proportional to the geometric mean between the eddy-memory time scale and the diffusive eddy equilibration timescale. Our model provides a formal justification for the previously proposed phenomenological model of the BAM and could be used to explicitly diagnose the memory kernel and improve our understanding of transient eddy-mean flow interactions in the atmosphere.

Spelling irregularities, known now as spelling mistakes, have been found for several centuries. As humans, we are able to understand most of the misspelled words based on their location in the sentence, perceived pronunciation, and context. Unlike humans, computer systems do not possess the convenient auto complete functionality of which human brains are capable. While many programs provide spelling correction functionality, many systems do not take context into account. Moreover, Artificial Intelligence systems function in the way they are trained on. With many current Natural Language Processing (NLP) systems trained on grammatically correct text data, many are vulnerable against adversarial examples, yet correctly spelled text processing is crucial for learning. In this paper, we investigate how spelling errors can be corrected in context, with a pre-trained language model BERT. We present two experiments, based on BERT and the edit distance algorithm, for ranking and selecting candidate corrections. The results of our experiments demonstrated that when combined properly, contextual word embeddings of BERT and edit distance are capable of effectively correcting spelling errors.

In game theory, mechanism design is concerned with the design of incentives so that a desired outcome of the game can be achieved. In this paper, we study the design of incentives so that a desirable equilibrium is obtained, for instance, an equilibrium satisfying a given temporal logic property -- a problem that we call equilibrium design. We base our study on a framework where system specifications are represented as temporal logic formulae, games as quantitative concurrent game structures, and players' goals as mean-payoff objectives. In particular, we consider system specifications given by LTL and GR(1) formulae, and show that implementing a mechanism to ensure that a given temporal logic property is satisfied on some/every Nash equilibrium of the game, whenever such a mechanism exists, can be done in PSPACE for LTL properties and in NP/$\Sigma^{P}_{2}$ for GR(1) specifications. We also study the complexity of various related decision and optimisation problems, such as optimality and uniqueness of solutions, and show that the complexities of all such problems lie within the polynomial hierarchy. As an application, equilibrium design can be used as an alternative solution to the rational synthesis and verification problems for concurrent games with mean-payoff objectives whenever no solution exists, or as a technique to repair, whenever possible, concurrent games with undesirable rational outcomes (Nash equilibria) in an optimal way.

Fonts have had trends throughout their history, not only in when they were invented but also in their usage and popularity. In this paper, we attempt to specifically find the trends in font usage using robust regression on a large collection of text images. We utilize movie posters as the source of fonts for this task because movie posters can represent time periods by using their release date. In addition, movie posters are documents that are carefully designed and represent a wide range of fonts. To understand the relationship between the fonts of movie posters and time, we use a regression Convolutional Neural Network (CNN) to estimate the release year of a movie using an isolated title text image. Due to the difficulty of the task, we propose to use of a hybrid training regimen that uses a combination of Mean Squared Error (MSE) and Tukey's biweight loss. Furthermore, we perform a thorough analysis on the trends of fonts through time.

Using characteristics to treat advection terms in time-dependent PDEs leads to a class of schemes, e.g., semi-Lagrangian and Lagrange-Galerkin schemes, which preserve stability under large Courant numbers, and may therefore be appealing in many practical situations. Unfortunately, the need of locating the feet of characteristics may cause a serious drop of efficiency in the case of unstructured space grids, and thus prevent the use of large time-step schemes on complex geometries. In this paper, we perform an in-depth analysis of the main recipes available for characteristic location, and propose a technique to improve the efficiency of this phase, using additional information related to the advecting vector field. This results in a clear improvement of execution times in the unstructured case, thus extending the range of applicability of large time-step schemes.

In this paper, we construct a sequence $(c_k)_{k\in\mathbb{Z}_{\geq 1}}$ of symplectic capacities based on the Chiu-Tamarkin complex $C_{T,\ell}$, a $\mathbb{Z}/\ell\mathbb{Z}$-equivariant invariant coming from the microlocal theory of sheaves. We compute $(c_k)_{k\in\mathbb{Z}_{\geq 1}}$ for convex toric domains, which are the same as the Gutt-Hutchings capacities. On the other hand, our method works for the contact embedding problem. We define a sequence of "contact capacities" $([c]_k)_{k\in\mathbb{Z}_{\geq 1}}$ on the prequantized contact manifold $\mathbb{R}^{2d}\times S^1$, which could derive some embedding obstructions of prequantized convex toric domains.

We prove that all homology 3-spheres are $J_4$-equivalent, i.e. that any homology 3-sphere can be obtained from one another by twisting one of its Heegaard splittings by an element of the mapping class group acting trivially on the fourth nilpotent quotient of the fundamental group of the gluing surface. We do so by exhibiting an element of $J_4$, the fourth term of the Johnson filtration of the mapping class group, on which (the core of) the Casson invariant takes the value $1$. In particular, this provides an explicit example of an element of $J_4$ that is not a commutator of length $2$ in the Torelli group.

This paper presents an approach to deal with safety of dynamical systems in presence of multiple non-convex unsafe sets. While optimal control and model predictive control strategies can be employed in these scenarios, they suffer from high computational complexity in case of general nonlinear systems. Leveraging control barrier functions, on the other hand, results in computationally efficient control algorithms. Nevertheless, when safety guarantees have to be enforced alongside stability objectives, undesired asymptotically stable equilibrium points have been shown to arise. We propose a computationally efficient optimization-based approach which allows us to ensure safety of dynamical systems without introducing undesired equilibria even in presence of multiple non-convex unsafe sets. The developed control algorithm is showcased in simulation and in a real robot navigation application.