GAN by Example using Keras on Tensorflow Backend – Towards Data Science

Generative Adversarial Networks (GAN) is one of the most promising recent developments in Deep Learning. GAN, introduced by Ian Goodfellow in 2014, attacks the problem of unsupervised learning by training two deep networks, called Generator and Discriminator, that compete and cooperate with each other. In the course of training, both networks eventually learn how to perform their tasks. GAN is almost always explained like the case of a counterfeiter (Generative) and the police (Discriminator). Initially, the counterfeiter will show the police a fake money.