Generative and Autoencoder Models for Large-Scale Mutivariate Unsupervised Anomaly Detection

Anomaly detection is a major problem that has been well studied in various fields of research and fields of application. In this paper, we present several methods that can be built on existing deep learning solutions for unsupervised anomaly detection, so that outliers can be separated from normal data in an efficient manner. We focus on approaches that use generative adversarial networks (GAN) and autoencoders for anomaly detection. By using these deep anomaly detection techniques, we can overcome the problem that we need to have a large-scale anomaly data in the learning phase of a detection system. So, we compared various methods of machine based and deep learning anomaly detection with its application in various fields. This article used seven available datasets. We report the results on anomaly detection datasets, using performance metrics, and discuss their performance on finding clustered and low density anomalies.