The increasing prevalence of network assaults presents a well-recognized challenge that can jeopardize critical information's availability, confidentiality, and integrity for individuals and organizations alike. In this paper, we introduce an intrusion detection methodology employing supervised machine learning. Our approach is straightforward yet effective, adaptable to various machine learning techniques. We tested several established machine learning methods to assess the efficacy of our intrusion detection system (IDS). Our empirical findings indicate that Support Vector Machines (SVM) and K-Nearest Neighbour (KNN) techniques outperform others. Consequently, we proceeded to develop an IDS utilizing SVM and KNN algorithms to classify online network data as either normal or indicative of an attack. Furthermore, we identified 12 crucial features of network data essential for detecting network attacks, employing information gain as our feature selection criterion. Our system can discern normal network activities from primary attack types (Probe and Denial of Service (DoS)) with a detection rate exceeding 98% within a 2-second timeframe. Additionally, we devised a novel post-processing method to mitigate the false-alarm rate and enhance the reliability and precision of the intrusion detection system.
Article Details
Unique Paper ID: 165107
Publication Volume & Issue: Volume 11, Issue 1
Page(s): 71 - 77
Article Preview & Download
Share This Article
Join our RMS
Conference Alert
NCSEM 2024
National Conference on Sustainable Engineering and Management - 2024