Intrusion detection using machine learning and feature selection based on enhancing chernobyl disaster optimizer algorithm

Artificial intelligence (AI) is a powerful technology with the potential to revolutionize cybersecurity by enhancing malware detection and strengthening defenses against cyberattacks. In this study, we explore the application of a binary version of the Chernobyl Disaster Optimizer (CDO) algorithm to address the feature selection problem in intrusion detection system (IDS) classification. The key innovation of this work lies in transforming the original continuous CDO algorithm into a binary optimization technique specifically tailored for feature selection tasks. By converting the search space from continuous to binary, the binary CDO algorithm better handles the inherent binary nature of the feature selection problem, leading to more efficient and effective feature subset identification. The experimental results demonstrate the superior performance of the binary CDO algorithm, coupled with a wrapper method (Random Forest) and a sigmoid transfer function with balancing, compared to other feature selection techniques. This approach outperformed other methods across various performance metrics, including accuracy, precision, recall, and F1-score. The Random Forest model, enhanced by the binary CDO-based feature selection, emerged as the top performer, achieving the highest accuracy of 93.83% and an exceptional F1-score of 94.48%.