Cite this Article

Anshul Sharma, S. Indra Priyadharshani, 2025. "Enhancing the Performance of Intrusion Detection Systems using Hybrid Bio-Inspired optimization Algorithms", International Journal of Emerging Information Technology (IJEIT) 1(1): 1-16.

Keywords

Intrusion detection systems, feature selection, filter-based feature selection, Particle Swarm Optimizer, Grey Wolf Optimizer, Hybrid GWOPSO.

Abstract

This research presents a novel hybrid feature selection method, integrating filter-based feature selection using mutual information and ANOVA F-value with the Grey Wolf Optimizer (GWO) and Particle Swarm Optimization (PSO), to enhance the accuracy and efficiency of classification tasks in network intrusion detection systems. Leveraging the UNSW-NB15 dataset, we evaluate four machine learning models—Random Forest, Decision Tree, Logistic Regression, and XGBoost—both before and after feature selection. The optimized feature sets show substantial improvements in performance metrics across models. Random Forest achieved a test accuracy of 95.28%, XGBoost reached 94.84%, Decision Tree recorded 94.01%, and Logistic Regression improved to 90.19% after feature selection. These findings underscore the efficacy of combining filter-based methods with hybrid optimization techniques for feature selection, effectively boosting model accuracy and computational efficiency. This study paves the way for further exploration of hybrid approaches and alternative classifiers to enhance intrusion detection systems.

Introduction

The rapid advancement of network technologies and the increasing sophistication of cyber threats highlight the critical need for effective Intrusion Detection Systems (IDS). These systems are essential for monitoring network traffic to identify and mitigate malicious activities, thus providing a safeguard against a broad spectrum of security threats. Despite significant progress in IDS methodologies, many existing systems still encounter challenges such as high false positive rates, overfitting, and computational inefficiencies. These issues can compromise the effectiveness of IDS, particularly in dynamic network environments characterized by growing data volumes and complexities.

Feature selection is a vital component in enhancing IDS performance, as it directly affects the accuracy and speed of anomaly detection. Traditional feature selection techniques often rely on single optimization methods, which may not adequately balance exploration and exploitation. This imbalance can result in suboptimal performance in identifying genuine threats while minimizing false alarms. Therefore, innovative solutions are required to improve feature selection processes, leading to more accurate and efficient IDS.

Recent research has explored various optimization algorithms for feature selection in IDS, including Genetic Algorithms (GA), Ant Colony Optimization (ACO), Grey Wolf Optimization (GWO), and Particle Swarm Optimization (PSO). Bio-inspired optimization techniques, particularly GWO and PSO, have gained traction due to their ability to mimic natural processes to solve complex optimization problems. GWO is noted for its strength in exploitation, while PSO excels in exploration, making their integration particularly promising.