Comparative Evaluation of Bio-Inspired Feature Selection Methods in Intrusion Detection

The escalating prevalence of cyber-activities under-scores the compelling need for advanced protective mechanisms, with Intrusion Detection Systems (IDS) emerging as a pivotal line of defense. Within the operational domain of IDS, the judicious selection of features plays a critical role in enhancing system accuracy and efficiency. This paper initiates an investigation into bio-inspired algorithms to enhance feature selection within the field of IDS. Where six outstanding algorithms—Whale Optimization Algorithm (WOA), Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), Firefly Algorithm (FFA), Manta Ray Foraging Optimization (MRFO), and Bat Algorithm (BAT)—undergo an accurate evaluation using three distinct datasets: CIC-IDS2017, UNSW-NB15, and NSL-KDD. The ensuing findings provide a nuanced understanding of the unique strengths and limitations exhibited by each algorithm, thus offering a clear directive for researchers and practitioners engaged in the domain of IDS feature selection. Moreover, the paper introduces LightGBM, a powerful gradient-boosting framework, as a classification model to assess the efficacy of feature selection. This research is motivated by the pressing need to enhance the performance of IDS systems in the face of evolving cyber threats. However, the outcomes of this research underscore the remarkable potential of bio-inspired algorithms in reshaping the existing paradigms of intrusion detection, setting the stage for future investigations in this dynamic and evolving field.