Behavior Analytics for Proactive Cybersecurity: Early Detection, Threat Hunting, and Incident Forensics Using Machine Learning

作者：Nada Hamzaoui, Nisrine Ibadah, K. Minaoui · 发表于：International Symposium on Signal, Image, Video and Communications · 年份：2026 · DOI：10.1109/isivc69944.2026.11574194 · 研究领域：Computer Science

Cyber threats have become increasingly complex in recent years, mainly due to the use of automation, polymorphic techniques, and artificial intelligence by attackers. This evolution has weakened the effectiveness of traditional signaturebased security mechanisms, which are no longer sufficient to detect emerging and previously unseen attacks. Proactive defense strategies, including early detection and threat hunting, are now essential now in modern cybersecurity environments. In this context, this paper explores the use of artificial intelligence for proactive threat detection based on behavioral analysis, considering several cybersecurity scenarios and different learning paradigms. In our study, comparative analysis of several machine learning techniques, starting with supervised approaches which, consist of detecting network intrusions via labeled traffic data, to expand the scope of the study in order to identify Phishing URLs with low resources settings, we used unsupervised methods. The use of deep learning models has been investigated for malware detection and classification via image representations of malware samples. The results exhibit that there is no single learning paradigm considered optimal in all cybersecurity threat scenarios, from which we can conclude that the importance is in selecting detection methods according to the characteristics of the data and the operational context.