Advancing Intrusion Detection Precision Through Analysis of Diverse Classification Algorithms
J Suriya Prakash, Snehitha Narasani, N. Thangadurai, U. Prakash, S Kiran · 2024
This research commences a thorough examination of the possibilities for leveraging Machine Learning (ML) algorithms to support Intrusion Detection Systems (IDS). We make use of the Kyoto dataset, a standard for intrusion detection studies that includes a wide variety of network traffic patterns related to both benign and malevolent activity. We carefully review a range of machine learning (ML) methods, such as decision trees, random forests, logistic regression, K-Nearest Neighbors (KNN), and Support Vector Machines (SVM) with different kernel functions. The distinct advantages and disadvantages of each algorithm in identifying network anomalies are clarified by this thorough examination. In addition to examining the detection capabilities, we delve into the performance metrics of these algorithms, including accuracy, precision, recall, and F1-score, providing a comprehensive assessment of their effectiveness. We also investigate the computational efficiency of these models by analyzing their training times and the impact of different data splits on their performance. Specifically, we evaluate batch sizes using $80: 20,70: 30$, and 60:40 training-to-test ratios to understand their influence on the training dynamics and the overall efficacy of the IDS. Furthermore, we explore the resilience of these algorithms against various forms of intrusions, such as data alteration attempts, unauthorized access attempts, and denial-ofservice (DoS) attacks. By investigating these state-of-the-art developments and promoting a broader comprehension of IDS approaches, this research ultimately contributes to the strengthening of cybersecurity defenses over time. This guarantees the confidentiality, integrity, and availability of their vital data assets while enabling enterprises to adjust and stay resilient against the constantly shifting threat landscape within complex IT infrastructures. The source code of our paper is available at the following linkhttps://github.com/Snehitha-Narasani/IDS-using-ML-algorithms