Review and Prospect of Research on Generative Adversarial Networks

Abstract

Since it was proposed in 2014, generative adversarial networks (GAN) has been highly concerned and widely studied by industrial circles and artificial intelligence researchers. It provides a new idea for the construction of the generative model. This paper reviews the research progress of GAN and prospects its development trend. Section 2 describes the basic idea and model structure of GAN. Section 3 introduces several typical derivative models of GAN. Section 4 lists GAN's applications in many fields, such as image, vision, voice, language, and etc. Section 5 makes a forward look and thinks on the development trend of GAN, and discusses the application of GAN in the field of communication countermeasures. Finally, section 6 summarizes this paper.