GANs is a method for training generative models that was proposed in 2014 by Goodfellow et al. There are two neural networks in a GAN. One is called the generator which takes a random noise as input and outputs an image. The other is called the discriminator which is a binary classifier to classify whether an input image is a real image or a fake image generated by the generator. The goal of the generator is to generate images that can fool the discriminator into thinking the images are real.
GANs are notoriously finicky with hyperparameters, and also require many training epochs. In order to make this project approachable without a GPU, we will be working on the MNIST dataset, which is 60,000 training and 10,000 test images. Each picture contains a centered image of white digit on black background (0 through 9). This was one of the first datasets used to train convolutional neural networks and it is fairly easy -- a standard CNN model can easily exceed 99% accuracy.
To simplify our code here, we will use the PyTorch MNIST wrapper, which downloads and loads the MNIST dataset. See the documentation https://github.com/pytorch/vision/blob/master/torchvision/datasets/mnist.py for more information about the interface. The default parameters will take 5,000 of the training examples and place them into a validation dataset. The data will be saved into a folder called MNIST_data.