Intrusion detection method based on independent component analysis and support vector machine

A novel intrusion detection method was presented, in which the independent component analysis approach was used to acquire the high order statistic information of intrusion action mode and mapped the input mode space into the corresponding independent component space. Then the generalized maximal margin hyperplane was constructed in the independent component space using the powerful feature of the support vector machine (SVM) for small samples and high dimension data generalization. Numerical simulation shows that the proposed method can reduce the dimension of the feature space, and has higher correct classification rate, especially, when the sigma of Gauss kernel is set to 1 to 3, the rate of false negative is just one ninth of the SVM's. It means that the intrusion detection method can effectively get the essential features of intrusion action and possess the higher ability to identify new intrusion activities.