Generative Adversarial Networks in Cybersecurity : Analysis and Response

Abstract

Cybersecurity is one of the key problems of the twenty-first century, as the digital environment has already become equally important to the real world, and in some situations, perhaps, even surpasses it. Especially when it comes to classified data or information with limited access. The main challenge for cybersecurity is a timely and adequate response to any type of threat since every day there are more and more malicious scenarios for compromising information technology data. For cyber-attack defense algorithms to be effective, it is important to maintain a high level of awareness of the widest range of state-of-the-art threat types and malicious technologies. Machine learning and AI in general are considered as both a helpful tool and a threat. Generative adversarial networks (GANs) and their modifications can pose a serious threat to the entire field and thus should be properly and thoroughly researched, especially in light of the fact that GANs can be used to improve known attack types so that even AI-based detection system cannot identify them. The purpose of this article is a comprehensive analysis and structuring of existing GAN methodologies, with the subsequent development of approaches for an adequate response to potential threats.