InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets

Paper: arXiv:1606.03657 by Xi Chen, Yan Duan, Rein Houthooft, John Schulman, Ilya Sutskever, Pieter Abbeel

Implementing the InfoGAN architecture in Keras for learning about detangled representation of salient features of the MNIST Hand Recognition Dataset in a completely unsupervised manner. The model was trained over 4000 epochs.