TITLE:
Applying Machine Learning for Real-Time Threat Detection in Adaptive Cybersecurity Systems
AUTHORS:
Godfrey Wandwi, Theodore Habimana
KEYWORDS:
Real-Time Detection, Machine Learning, Adaptive Cybersecurity, Threat Detection, Network Intrusion, Classification Accuracy
JOURNAL NAME:
Open Journal of Applied Sciences,
Vol.16 No.9,
September
21,
2026
ABSTRACT: The dynamic landscape of cybersecurity threats necessitates intelligent and responsive defense mechanisms. As cyberattacks become increasingly sophisticated, real-time detection within adaptive systems becomes crucial. This study proposes a machine learning-driven framework for real-time threat detection, specifically tailored for adaptive cybersecurity environments. The architecture leverages supervised learning algorithms integrated with dynamic feature selection to process and classify evolving network behaviors effectively. Emphasis is placed on minimizing detection latency and improving classification precision across diverse attack vectors. Performance evaluation is conducted using benchmark datasets, including UNSW-NB15, NSL-KDD, and KDD Cup’99, under varying traffic conditions and attack intensities. Experimental findings demonstrate that the applied machine learning framework consistently achieves high detection accuracy and reduced false-positive rates, affirming its reliability for real-time deployment. The framework’s ability to adaptively respond to novel threats while maintaining computational efficiency positions it as a practical solution for next-generation cybersecurity systems.