Network Intrusion Detection Based on Support Vector Machine

Statistical learning theory （SLT） is introduced to intrusion detection （ID） and an ID method based upon support vector machine （SVM） is presented in this paper. The SVM algorithm is generalized for high dimensional and heterogeneous datasets acquired in ID and a new RBF kernel function is developed based on HVDM distance metric of heterogeneous datasets. Supervised C-SVM and unsupervised One-Class SVM algorithms utilizing kernel function are applied in detecting the intrusions hidden in the network connection records. The testing results on the DARPA data show that the method is effective and efficient.