Future of generative adversarial networks (GAN) for anomaly detection in network security: A review

Abstract

Anomaly detection is crucial in various applications, particularly cybersecurity and network intrusion. However, a common challenge across anomaly detection techniques is the scarcity of data that accurately represents abnormal behavior, as such behavior is often detrimental to systems and, consequently, rare. This data limitation hampers the development and evaluation of effective anomaly detection methods. In recent years, Generative Adversarial Networks (GANs) have garnered significant attention in anomaly detection research due to their unique capacity to generate new data. This study conducts a systematic review of the literature to delve into the utilization of GANs for network anomaly detection, with a specific emphasis on representation learning rather than merely data augmentation. Our study also seeks to assess the efficacy of GANs in network anomaly detection by examining their key characteristics. By offering valuable insights, our research can aid researchers and practitioners in understanding the evolving landscape of network anomaly detection and the practical implementation of GANs while addressing the challenges in developing robust GAN-based anomaly detection systems.