Detecting Zero-Day Attacks with AI

Harnessing AI for Zero-Day Attack Detection: Advancing Cybersecurity with Hybrid Models and Anomaly Detection

Abstract

The escalating threat of cyberattacks has heightened the need for advanced intrusion detection systems, especially against elusive zero-day attacks. Zero-day attacks exploit undiscovered vulnerabilities, leaving systems vulnerable before patches are available. This article reviews and synthesizes cutting-edge AI-based methodologies for detecting zero-day attacks and explores the associated challenges, drawing from a systematic literature review (SLR) by the authors’ team. Additionally, the article highlights the integration of anomaly detection techniques, such as autoencoders, with machine learning models to enhance detection performance for previously unseen data