Targeting Insider Threats and Zero-Day Vulnerabilities with Advanced Machine Learning and Behavioral Analytics
Somnath Raghunath Wategaonkar, Alakbarova Tamara Shaki, Abbasova Parvin Ali, Zeynalov Javanshir Ibrahim, L.N. Jayanthi, S. Nalini Jayanthi · 2024
This study presents a revolutionary approach to cybersecurity that uses behavioural analytics and machine learning in tandem to show how it is far better than the status quo. When compared to Isolation Forest and SVM Classifier when used separately, the approach performs far better when implemented as a whole. The integrated solution effectively addresses zero-day vulnerabilities and insider threats, as evidenced by its precision, recall, and F1-score values of 0.93, 0.94, and 0.93, respectively. Improving anomaly detection, decision boundary visualisation shows regions of collaborative consensus. Remarkably fewer false positives and negatives highlight the practicality. This paper paves the way for further research into fine-tuning, dynamic adaptability, ensemble approaches, and the difficulties of large-scale deployment, in addition to improving cybersecurity. This research represents a major step forward in developing agile and trustworthy cybersecurity solutions, which is crucial for dealing with the ever-changing cyber threat scenario.