Overview of GANs¶
Generative Adversarial Networks (GANs) are neural networks which use two components, a generator and a discriminator, to create a mapping from a latent space to a data space.
The generator (G) is responsible for creating the data, and the discriminator is responsible for determining a given input is created by the generator or is actual training data. By using this approach, it is possible for the generator portion of the GAN to create the mapping from the latent space to the data space in an unsupervised fashion. However, during training, the discriminator is provided a signal which indicates whether or not the given input is from G or from the training data.