A generative adversarial network (GAN) is a class of machine learning frameworks where two neural networks are pitted against each other in the form of a zero-sum game. Where a generator network tries to generate an image, while the discriminator network tries to check its authenticity.

I trained a generator and discriminator to create images of clothes, using the Fashion MNIST dataset. Both the generator and discriminator were compiled using binary cross-entropy loss. The training started with a learning rate of 0.0002 and a batch-size of 64.

The model converged pretty quickly, and the training was terminated when the loss stopped decreasing. You can view sample output in the adjacent image.