Research on Intrusion Detection System Based on KPCA and SVM
Jun Gu · Jisuanji fangzhen · 2010
Current IDS is an important part of network security.Current IDS has poor generalization ability when given less priority knowledge.In this paper,the KPCA and SVM are adopted to implement intrusion detection.The Kernel Principal Component Analysis method can not only solve the linear correlation of the input but also compress the data.The kernel parameters of KPCA are optimized by grid algorithm and the parameters of support vector machine model are selected by cross validation method.Compared with traditional algorithms,this method can achieve higher detection rate and better generalization,and decrease the time of performance.In the end of the paper,the experiment on KDD CUP99 data set shows the effectiveness and excellent performance of the method.