An Intrusion Detection Approach Using SVM and Multiple Kernel Method
In this paper, the authors present an intrusion detection method based on multiple kernel support vector machine. This method can calculate the weights of kernel functions and Lagrange multipliers simultaneously through semi-infinite linear programming, improve detection accuracy by using the RBF kernel function with different kernel parameter values and avoid the kernel parameter settings in advance. They also study sparse and non-sparse kernel mixtures by allowing for l1-norm constraint and l2-norm constraint. The experimental results show that MK-SVM method has better detection accuracy than SK-SVM method, l2-MK- SVM method is superior to l1-MK-SVM method.