Enhancing Cybersecurity Intrusion Detection Using AI-Based Genetic Algorithm Feature Selection

Intrusion detection refers to a systematic process of identifying suspicious behavior that might harm or affect the usage of certain networks. Recently, there has been considerable interest in the use of Machine Learning (ML) and Deep Learning (DL) techniques for intrusion detection. Since ML techniques primarily rely on the feature space used, researchers have attempted to perform feature selection to identify the most accurate subset of features. For this purpose, the literature has shown interest in using Genetic Algorithms (GAs) as a feature selection technique. Although it demonstrated strong performance in finding solutions, GA still has some drawbacks in identifying the optimal feature subset. The main drawback is premature convergence, which causes the GA to converge to a local optimum. This study proposed an enhancement of GA by utilizing a ranking group selection mechanism. The UNSW-NB15 dataset was used, and multiple preparation and preprocessing steps were performed to transform the data into an appropriate form. Subsequently, both the original and enhanced GA were applied to conduct feature selection. For classification, a decision tree (DT) algorithm was employed. With 100 generations, the proposed enhanced GA outperformed the original approach, achieving a classification accuracy of 99.89%. This demonstrates the effectiveness of the enhancement over GA, which enables efficient acquisition of the global solution.