Machine Learning for Developers

Discriminative and generative models

To understand the concept of adversarial networks, we will first define the two models that interact in a typical GAN setup:

• Generator: This is tasked with taking in a sample from a standard random distribution (for example, a sample from an n-dimensional Gaussian) and producing a point that looks sort of like it could come from the same distribution as XX. It could be said that the generator wants to fool the discriminator to output 1. Mathematically, it learns a function that maps the input data (x) to some desired output class label (y). In probabilistic terms, it learns the conditional distribution P(y|x) of the input data. It discriminates between two (or more) different classes of data—for example, ...