Learning to forge MNIST images with DCGANs

In this recipe, we will use a simple GAN that uses CNNs for learning how to forge MNIST images and create new ones that are not part of the original dataset. The idea is that CNNs used together with GANs will improve the ability to deal with image datasets. Please note that the previous recipe was using GANs with fully connected networks, while here we focus on CNNs.

Getting ready

This recipe is based on the code available at https://github.com/TengdaHan/GAN-TensorFlow.

How to do it...

We proceed with the recipe as follows:

1. Clone...