Real-Time Anomaly Detection for Zero-Day Exploits

Zero-day exploits are cyber attacks that take advantage of vulnerabilities that are previously unknown. Lack of prior signatures or patches makes them a critical security threat. This paper is going to explore the approach based on anomalies for real-time detection of such zero-day exploits. The approach tries to flag any deviations from normal behavior to recognise potential attacks. This paper will try to explore the challenges and limitations (including model poisoning, regulatory constraints, adversarial evasion, and operational issues) and observe some zero-day exploit detection in real-world scenarios. The paper will also outline the future directions, federated learning for collaborative defense, adaptive threat modeling, integration with cyber threat intelligence (CTI), and self-healing systems.