Generative adversarial networks are one of the recent research areas in deep learning. It is used in various applications in image/text/video generations etc. GANs are widely known for the adversarial process it follows and the two models in its architecture – the generator and the discriminator. Since it gives better results than other generative models it is preferred more. In this paper initially we discuss the basic architecture of GANs, the mathematical concept involved in it and how it outshines other generative models. The various applications of GANs to name a few, implementation of GANs in Radiology, beauty GANs, GANs deployed in cyber security and attention prediction have been investigated and mentioned with a brief description about the same. Further in the end we have discussed the challenges faced by it and its future scope in the fourth coming years.

Keywords : Generative Adversarial Network, Generator, Discriminator, Adversarial Process, Deep Learning.