Author(s): Ahmed Jaber Joudah and Huda Yass Khudhair

Abstract: The number of devices using the Internet, as well as the tasks performed through it, increases every day. So do the attacks against the availability, integrity and confidentiality of the information they handle. Intrusion Detection Systems (IDS) are one of the most effective security mechanisms to protect network systems against computer attacks, whether or not there is prior knowledge of them. The application of Artificial Intelligence and, more specifically, of automatic learning in this type of program stands out. Most of these IDS analyse network traffic and its normal behaviour so that they can activate an alarm when part of said traffic differs from the usual. In this way, some types of attacks can be detected even if they have never been carried out before. The study investigates ensemble learning for detecting network security attacks, intending to improve machine learning algorithm performance by combining their results. This study evaluates the efficacy of different machine learning algorithms in intrusion detection utilizing the NSL KDD dataset, employing Recursive Feature Elimination (RFE) for feature selection. Results indicate that Random Forest (RF) and Gradient Boosting Machine (GBM) are the foremost performers, while Decision Tree (DT) also demonstrates commendable balanced accuracy and precision. Although Support Vector Machine (SVM) has demonstrated favourable outcomes in previous studies, the research indicates that multiple classifiers ought to be evaluated for forthcoming intrusion detection system (IDS) implementations.

DOI: 10.33545/26633582.2025.v7.i2a.201

Pages: 45-56 | Views: 298 | Downloads: 98

Download Full Article: Click Here