Improved Semi-supervised Learning with GANs using Manifold Invariances.

Semi-supervised learning methods using Generative adversarial networks (GANs) have shown promising empirical success recently. Most of these methods use a shared discriminator/classifier which discriminates real examples from fake and also predicts the class label. Motivated by the ability of GANs to capture the data manifold well, we propose to estimate the tangent space to the data manifold using GANs and use it to inject invariances into the classifier. In the process, we propose improvements over existing methods for learning the inverse mapping (i.e., the encoder) cite{donahue2016adversarial} which greatly improve in terms of semantic similarity of reconstructed sample to the input sample. We experiment with SVHN and CIFAR-10 for semi-supervised learning, obtaining significant improvements over baselines, particularly in the cases when the number of labeled examples is low. We also provide insights into how fake examples influence the semi-supervised learning procedure.