Intrusion Detection Based On Support Vector Machine

With the rapid development and popularization of the compute network, the security problem is get more and more important. As a kind of helpful reinforce for firewall, virus detection and encryption, Intrusion Detection System (IDS) reinforced protection scope on the system and compute network. Support Vector Machine (SVM) was put forward as a kind of machine study arithmetic based on statistic study theory. Because SVM have much advantage over other arithmetic, it was paid more attention to and get success in application of pattern recognition and regression estimation.Nowadays, many researchers applied SVM to IDS for its much advantage. But a lot of problem bring out. The research work on the paper is based on the problems. The main research work and results are followed:1 The multi-dimension data whose individual attribute is different is called as heterogeneous data. The HVDM distance function is brought in and improved by adding a appropriate weight on individual attribute (WHVDM). So the distance measurement problem is well solved.2 A new intrusion detection method is proposed based on WHVDM kernel function and Center Distance Ratio(CDR).Firstly, a new RBF kernel function based on WHVDM is put forward and proved positive and definite in mathematic for the high dimensional and heterogeneous datasets acquired in Intrusion Detection(ID). The computer simulation experiments have showed the kernel function feasible and effective. Secondly, the number of samples that are trained directly has decreased greatly by applying CDR to ID. In the situation that the performance is pretty much the same thing, the time complexity of the method is decreased obviously. The new method of ID is proved feasible and effective by experiment result.