How do generative adversarial networks (GANs) work?
In: Technology
A GAN consists of two neural networks: one that tries to make a realistic fake (something that looks like its target image but isn’t genuine) and another network that tries to detect genuine examples and fakes. As they are trained, one network tries to get better at making fakes and the other network gets better at detecting fakes. This results in better fakes, or in more precise terms, results like style transfer, in which you train them on works by a certain painter and then start with a different photo and try to make it match that painter’s style.
Latest Answers