Elevating Intrusion Detection Precision with Multi-Classification Algorithm Analysis
J Suriya Prakash, Talluri Rashmika, N. Thangadurai, U. Prakash, S Kiran · 2024
This paper examines the use of machine learning methods to detect network security threats. By analysing network traffic data and using a range of machine learning techniques. This paper focuses on improving IDS performance with ML techniques, utilizing the Kyoto20151207 dataset as a benchmark. It employs fifteen machine learning algorithms over a wide range of techniques to identify network breaches. This study aims to identify the most effective approach for intrusion detection tasks. Using comprehensive testing and assessment, several algorithm-performance couples are ranked according to correctness and accuracy. The study’s goal is to give insight on the efficacy of various strategies, allowing practitioners to choose the best approaches for intrusion detection tasks. Furthermore, the accuracies of these methods are compared to those reported in other works, providing insight into the overall performance context. Finally, the study seeks to help practitioners make educated judgments when selecting relevant categorization algorithms for similar scenarios in IDS deployment. The key findings of our research are we implemented 15 different machine learning algorithms and identified the best accuracy in CatBoostClassifier algorithm as $99.738 \%$. This paper’s source code may be seen at the following website https://github.com/rashmika357/ideal-palm- tree