User Behavior Analysis for Cyber Threat Detection: A Comparative Study of Machine Learning Algorithms

Over the years, the increasing sophistication of cyber-attacks including insider threats, advanced persistent threats (APTs), and zero-day vulnerabilities have placed critical sectors such as health care, and government at severe risk. User behavioral analysis integration with machine learning (ML) provides an efficient solution for the detection and mitigation of these threats. This study aims to perform a comparative assessment of ML-based methods, analyzing their efficacy in handling insider attacks, APTs, and zero-day vulnerabilities. The detailed study of the state of the art showcases the ability of ML-based user behavioral analysis while highlighting key challenges faced by existing solutions such as data imbalance, high energy consumption, and scalability and latency concerns, among others. Moreover, real-life use cases of user behavioral analysis integration with ML for cyber security are also explored. The article also explores the research directions, providing future professionals and researchers with the road map for cyber security enhancement. To our knowledge, this is the first article to perform a comparative assessment of ML-based behavioral analysis solutions for insider threats, APTs, and zero-day vulnerabilities.