Optimizing Real-Time Network Intrusion Detection Using a Refined Data Filtering Method

This paper proposes a proactive cybersecurity approach to address escalating cyber threats and overcome the limitations of conventional security measures. Our work emphasizes real-time monitoring using advanced machine learning techniques. We introduce an anomaly-driven data filtering strategy that enhances a real-time monitoring system for improved intrusion detection performance. Models are trained on the border gateway protocol anomaly dataset, Slammer, using three distinct approaches: a regular method (no filtering), a random selection method, and an anomaly-driven data filtering method. Experimental results demonstrate that the data filtering approach achieves the best detection performance.