Comparative Analysis of Deep Convolutional Generative Adversarial Network and Conditional Generative Adversarial Network using Hand Written Digits

Abstract

Generative adversarial network framework has recently emerged as a promising generative modeling approach. It is composed or made up of a generative network and a discriminative network. There have been various types of adversarial networks present today. Among these types of networks, the most popular network is Deep Convolutional Generative Adversarial Network (DCGAN) for performing on the convolutional networks without using multilayer perceptrons. The multilayer layer perceptrons have the hidden layers because of this we have to more bind to extract the data with the parameters, because of this we also study the Conditional Generative Adversarial Network (CGAN) to add an extra label to the generator and the discriminator. We study the comparative analysis between these two popular networks to highlight the main differences and similarities using the handwritten image datasets.