Efficient Intrusion Detection Scheme based on SVM

Abstract

The network intrusion detection problem is the focus of current academic research. In this paper, we propose to use Support Vector Machine (SVM) model to identify and detect the network intrusion problem, and simultaneously introduce a new optimization search method, referred to as Improved Harmony Search (IHS) algorithm, to determine the parameters of the SVM model for better classification accuracy. Taking the general mechanism network system of a growing city in China between 2006 and 2012 as the sample, this study divides the mechanism into normal network system and crisis network system according to the harm extent of network intrusion. We consider a crisis network system coupled with two to three normal network systems as paired samples. Experimental results show that SVMs based on IHS have a high prediction accuracy which can perform prediction and classification of network intrusion detection and assist in guarding against network intrusion.