Abstract: Detecting network intrusions with high accuracy and precision is vital for safeguarding systems and preventing cyber threats. Traditional intrusion detection systems (IDS) often struggle with issues such as adaptability, efficiency, and precision. This paper presents an advanced approach that integrates machine learning techniques, specifically Support Vector Machines (SVM) and Random Forest, with historical attack data to improve IDS performance. Additionally, a Case-Based Reasoning (CBR) system is incorporated to compare new incidents with similar historical cases, offering contextual insights that enhance detection accuracy. The goal is to achieve a detection accuracy of more than 97%, minimizing false positives and improving the overall reliability of the IDS. Experimental results show that the integration of machine learning models such as SVM and Random Forest with the UNSW-NB15 dataset leads to significantly improved detection rates and strengthens cybersecurity defenses. This method provides a robust, scalable solution for responding to evolving cyber threats.
Keywords: Intrusion Detection, Machine Learning, Support Vector Machines (SVM), Random Forest, Network Security, UNSW-NB15, Malicious Network Activities Detection
|
DOI:
10.17148/IJARCCE.2025.14272
