A Survey on the Progression and Performance of Generative Adversarial Networks

Generative Adversarial Networks (GANs) are a class of deep neural networks that provide a unique way of modeling and generating data in an unsupervised manner. The literature shows that GANs are currently an important research area, being used in a variety of applications. Our survey paper gives an overview of GANs and how they have progressed to become more efficient and powerful. We discuss the efficiency of GAN-variant models, demonstrate how GANs are used in different applications, elaborate on the shortcomings of GANs, review the various ways that researchers have addressed these issues and list several important metrics that are most used in this field. The objective of this paper is to provide a summary of the progression and performance of GANs and the current research that is being conducted to improve them.