Paper Title
Comparison of Various Generative Adversarial Network Architectures

Abstract
Generative Adversarial Networks (GAN) is a much-explored subfield of machine learning for generating synthetic data using deep generative models, although they have only recently been developed in 2014. As a result, GANs have been used in a wide variety of fields, most notably in machine vision, where they are often used to create or modify artificial images. Therefore, it makes sense that researchers in the field of networking (to which deep learning methods have been applied extensively) would be interested in GAN-based approaches due to their relative simplicity.Many deep learning models have been created as a result of tremendous deep learning success in the field of artificial intelligence. A deep learning model called Generative Adversarial Networks (GAN) was developed using zero-sum game theory and is now a popular field of study. The goal of model variation is to provide more accurate and realistic data by using unsupervised learning to capture the data distribution. Due to their great application potential in areas such as image and image processing, video and speech processing, etc., GANs are now the subject of extensive research. Since their inception, several variants of GANs have been proposed to address their limitations and enhance their capabilities. In this review paper, we provide an in-depth analysis of various GAN variants and their applications. Keywords - GANs, ProGANs, StyleGANs, Pix2Pix, SRGAN, ESRGAN, CGAN, CycleGAN