Ensemble-Based Machine Learning Algorithm for Intelligent Network Security Threat Detection

Abstract

The rapid development of cyber threats in the cybersecurity field necessitates advanced strategies for prompt identification and reduction. Conventional approaches frequently struggle to adapt to the complexity of contemporary attacks, emphasizing the requirement for creative approaches utilizing machine learning. This paper creates and assesses the “IntelliGuard Threat Detector” algorithm, developed to independently identify and classify a variety of network security risks employing the CICIDS 2017 dataset. By utilizing advanced machine learning methods, the algorithm aims to enhance accuracy and effectiveness in locating abnormal behaviors suggestive of possible security violations. Present methods for network security usually depend on personal intervention and pre-established guidelines, which may not sufficiently handle the ever-changing nature of cyber threats. The “IntelliGuard Threat Detector” algorithm incorporates robust scaler normalization, Composite Rank Ensemble (CORE) feature selection, and a TrioBoost classifier model to boost predictive accuracy and robustness. The proposed IntelliGuard Threat Detector algorithm attains 94% accuracy, 92% precision, 95% recall, 94% F1-score, and 93% geometric mean, surpassing conventional techniques by up to 6% in accuracy, 8% in precision, 5% in recall, 7% in F1-score, and 7% in geometric mean, respectively. This algorithm provides a proactive and scalable approach for network security threat discovery, signifying a noteworthy development in the area of cybersecurity.