Hands-On Generative Adversarial Networks with Keras

Comparing discriminative and generative models

Learning the conditional distribution is easier, because you do not have to make assumptions about the marginal distribution of x or y.

We will use the following diagram to illustrate the differences between discriminative and generative models. We can see two plots with 13 points of a two-dimensional dataset; let's call the blue class labels , and the yellow class labels :

When training a discriminative ...