A Comparative Analysis of Machine Learning and Hybrid Approaches for Network Intrusion Detection: Performance Evaluation and Implementation Framework

Fasanmi Ezekiel Olufemi
Department of Computer Science, Delta State University, Abraka, Nigeria

Abstract

Network Intrusion Detection Systems (NIDS) are vital in modern cybersecurity, serving as the first defense against complex threats. This research analyzes signature-based, anomaly-based, and machine learning NIDS across DARPA, NSL-KDD, and CICIDS2017 datasets, measuring accuracy, false positives, and computational cost. Hybrid methods combining signature-based and ML achieved 94.7% accuracy with 2.3% false positives. Deep learning, especially CNNs, excelled at zero-day detection but needed more resources. Findings highlight that integrating AI is essential for evolving threats and encrypted traffic analysis, guiding practitioners in selecting optimal NIDS configurations.

Keywords

Network Intrusion Detection, Cybersecurity, Machine Learning, Anomaly Detection, Signature-based Detection