INTRUSION DETECTION USING MACHINE LEARNING

Abstract

Cybersecurity is a critical concern in today's interconnected world, with networks facing constant threats. Intrusion Detection Systems (IDS) are essential tools for safeguarding networks by identifying and responding to unauthorized access and malicious activities. However, traditional IDS often struggle to adapt to the evolving threat landscape. This thesis proposes a novel approach to enhance IDS capabilities by leveraging advanced machine learning techniques. The primary objective is to develop a robust and intelligent IDS capable of accurately detecting and classifying a wide range of network intrusion patterns. We investigated and compared multiple machine learning algorithms on the UNSW-NB15 dataset, which comprises both normal and malicious network traffic. Our experimental results demonstrate the effectiveness of our proposed machine learning-based IDS, with Decision Tree, Random Forest, XGBoost, and an ensemble model achieving impressive accuracy rates of 96.84%, 97.78%, 97.45%, and 97.52%, respectively, in multi-label classification on the UNSW-NB15 dataset.