Survey on Applying GAN for Anomaly Detection

Abstract

In the current days, most prominent research in machine learning was focused on the generative models. Generative Adversarial Networks (GANs) is one of the generative models used to model the complex high dimensional distribution of real-world data. GANs have two structures, generator to create new data instances resembling our training data, and discriminator to distinguish real data from the data created by the generator. As predicting abnormal data is one of the most important problems across a range of domains. Our Literature survey was conducted on applications of GAN in the field of anomaly detection and a simple experiment is conducted illustrating the usage of GAN in anomaly detection using MNIST dataset.