Generative Adversarial Networks (GANs) - A Gentle Introduction

Author: Umberto Michelucci umberto.michelucci@toelt.ai.

This repository contains the material for a Gentle Introduction to GANs. Here you can find

Main Content

• A Gentle Introduction to GANs (PDF) - A paper with a short introduction on the main idea about GANs
• Under code you can find some examples.

Lecture given at the Google Student Developer Group Munich

Complete Examples

In case you are interested, you can find compelte examples at httsp://adl.toelt.ai. In particular you can find the complete code for a generic GAN applied to MNIST HERE.

Slides

You can find the slides in the PDF file that you can find in the root folder of the repository.

Code Folder

In the code folder the following files are relevant:

• Example of a GAN with MNIST
• There are several h5 files that are used by the notebook. Those are the trained generator and discriminator. You should not need to use them directly.
• GAN_Fashion_MNIST.py this file is a script that will train a GAN on MNIST and save the generator model as generator.h5 (already avaialble in the folder). This script can be used when letting the training run for a few hours on a server.
• Example of a conditional GAN with MNIST

The other folders contain output images and are not directly relevant.