Applications of Generative Adversarial Networks in Anomaly Detection: A Systematic Literature Review

Anomaly detection has become an indispensable tool for modern society, applied in a wide range of applications, from detecting fraudulent transactions to malignant brain tumors. Over time, many anomaly techniques have been introduced. However, general, they all suffer the same problem: lack data that represents anomalous behaviour. As behaviour is usually costly (or dangerous) system, it difficult gather enough such This, turn, makes develop and evaluate techniques. Recently, generative adversarial networks (GANs) attracted much attention research, due their unique ability generate new data. In this paper, we present systematic review literature area, covering 128 papers. The goal paper analyze relation between types GANs, identify most common application domains GAN-assisted GAN-based detection, assemble information on datasets performance metrics used assess them. Our study helps researchers practitioners find suitable technique application. addition, research roadmap future studies area. summary, GANs are address problem insufficient amount behaviour, either through augmentation or representation learning. commonly GAN architectures DCGANs, standard cGANs. primary include medicine, surveillance intrusion detection.