Image Synthesis with Generative Adversarial Networks (GAN)

Abstract

The emergence of artificial Intelligence has paved the way for numerous developments in the domain of machine vision. One of the many frameworks and algorithms which have set a benchmark for the generation of data from learned parameters is Generative Adversarial Networks. In this chapter, Generative Adversarial Networks (GANs) and similar algorithms, such as Variation Auto-Encoders (VAEs), are used to generate handwritten digits from noise. Furthermore, the training data has been visualized to gain a proper understanding of the data our model is trying to learn.