Welcome to the Generative Adversarial Networks Wiki!Generative Adversarial Networks or “GANs”, are one of the newest innovations in the field of neural networks. This is a form of machine learning was invented by Ian Goodfellow and works by setting two neural networks, the Discriminator and Generator, to solve a problem. An easy way to explain GANs are in terms of images and forgery. Firstly, there are 2 main components in the architecture of a GAN, the Discriminator and the Generator. The discriminators job is to identify if an image is real or fake by outputting a 1 or a 0 and it wants to maintain as low of an error rate as possible. The generator produces a fake image from a random input and then feeds it to the discriminator in an attempt to trick it into believing that it is real. The generators goal is to force the discriminator into having a high error rate.
Community content is available under CC-BY-SA unless otherwise noted.