Disentangled Representation Learning with Information Maximizing Autoencoder

Kazi Nazmul Haque, Siddique Latif, Rajib Rana

Keywords: Disentangled Representation Learning, Data Augmentation, Generative Adversarial Nets, Unsupervised Learning

TL;DR: Learn disentangle representation in an unsupervised manner.

Abstract: Learning disentangled representation from any unlabelled data is a non-trivial problem. In this paper we propose Information Maximising Autoencoder (InfoAE) where the encoder learns powerful disentangled representation through maximizing the mutual information between the representation and given information in an unsupervised fashion. We have evaluated our model on MNIST dataset and achieved approximately 98.9 % test accuracy while using complete unsupervised training.