Untitled

Abstract

In an era marked by widespread computer usage, security emerges as a critical focal point demanding meticulous attention. The spectrum of potential threats encompasses various methods of attacking computer systems, with Denial of Service (DoS) attacks being a prominent concern. This study delves into the enhancement of cybersecurity by implementing a system capable of discerning between DoS attack data and normal data, employing the Support Vector Machine (SVM) algorithm. To optimize the efficacy of the classification system, a strategic feature selection process is imperative. This research advocates for the utilization of the Chi-square method for this purpose, aiming to eliminate irrelevant features and thereby enhance system performance. The Support Vector Machine algorithm, hinging on hyperplanes for classification, gains efficiency through judicious feature selection. The empirical findings of this research unveil that employing Chi-square feature selection significantly elevates the performance of the classification system when dealing with application layer attacks. Remarkably, this enhancement is achieved without compromising the accuracy of the system. Specifically, the classification of DoS application layer attacks using SVM in tandem with Chi-square yielded identical accuracy results compared to using SVM alone. The average accuracy reached an impressive 99.9995%, with a processing time of 6.08 minutes with chi-square selection feature. In contrast, the classification system without feature selection consumed a comparatively longer processing time of 6.85 minutes. This underscores the efficacy of Chi-square feature selection in optimizing the performance of cybersecurity systems, demonstrating a streamlined approach to safeguarding computer networks from malicious threats.