Machine Learning

How to Code the GAN Training Algorithm and Loss Functions

Last Updated on January 10, 2020

The Generative Adversarial Network, or GAN for short, is an architecture for training a generative model.

The architecture is comprised of two models. The generator that we are interested in, and a discriminator model that is used to assist in the training of the generator. Initially, both of the generator and discriminator models were implemented as Multilayer Perceptrons (MLP), although more recently, the models are implemented as deep convolutional neural networks.

It can be challenging to understand how a GAN is trained and exactly how to understand and implement the loss function for the generator and discriminator models.

In this tutorial, you will discover how to implement the generative adversarial network training algorithm and loss functions.

After completing this tutorial, you will know:

  • How to implement the training algorithm for a generative adversarial network.
  • How the loss function for the discriminator and generator work.
  • How to implement weight updates for the discriminator and generator models in practice.

Kick-start your project with my new book Generative Adversarial Networks with Python, including step-by-step tutorials and the Python source code files for all examples.

Let’s get started.