An example of a semi-supervised learning approach is "self-training", in which a model is trained on labeled data, and then used to make predictions on the unlabeled data.