GANs – GENERATIVE ADVERSARIAL NETWORKS

definitions ans applications

Abstract

The use of Generative Adversarial Networks - GAN - by discovery and learning standard in datasets wich will bring inovation and change in several aplications. It's a models that uses two neural networks, one generator and one decrimantor, to create data that resembles the original data. They learn the distribution of probabilities of the original data and compete with each other to generate quality and diverse data. Their use with the same type of data allows eficate results in the most diverse fields, be it in art, in entertainment, in data analysis or in medicine, their evolution grows every day. This research aims to analyse the Generative Adversarial Network (GAN), its concept and challenges in its application. The research used went the bibliografic, exploring many academic cientific works. Their results see possibilities of development of more various tools and application in areas diverse, making this tool a promissor instrument of new generative technologies.