INTRUSION DETECTION SYSTEMS USING ADAPTIVE REGRESSION SPLINES

Title: INTRUSION DETECTION SYSTEMS USING ADAPTIVE REGRESSION SPLINES

Research Question: Can Adaptive Regression Splines (MARS), Neural Networks (ANNs), and Support Vector Machines (SVMs) be used to effectively classify network traffic patterns based on a 5-class taxonomy?

Methodology: The researchers used data from the MIT Lincoln Lab's DARPA intrusion detection system evaluations as their benchmark. They compared the performance of MARS, ANNs, and SVMs in classifying network traffic patterns into five classes: normal, probe, denial of service, user to super-user, and remote to local.

Results: The study found that MARS, ANNs, and SVMs performed well in classifying network traffic patterns. However, MARS showed better training and testing times, scalability, and classification accuracy compared to the other two methods.

Implications: The results suggest that MARS can be a cost-effective and efficient tool for building intrusion detection systems. The study also highlights the potential of data mining techniques in automating the intrusion detection process and reducing human intervention.

Link to Article: https://arxiv.org/abs/0405016v1 Authors: arXiv ID: 0405016v1