Generative Adversarial Networks: An Overview

Generative adversarial networks (GANs) provide a way to learn deep\nrepresentations without extensively annotated training data. They achieve this\nthrough deriving backpropagation signals through a competitive process\ninvolving a pair of networks. The representations that can be learned by GANs\nmay be used in a variety of applications, including image synthesis, semantic\nimage editing, style transfer, image super-resolution and classification. The\naim of this review paper is to provide an overview of GANs for the signal\nprocessing community, drawing on familiar analogies and concepts where\npossible. In addition to identifying different methods for training and\nconstructing GANs, we also point to remaining challenges in their theory and\napplication.\n