Generalizing GANs: A Turing Perspective

They place two neural networks--a model and a discriminator--in a competitive setting. The discriminator's objective is to correctly label samples from either the model or the training data. The model's objective is to deceive the discriminator, in other words, to produce