Intrusion Detection System Technique Based on BP-SVM

Due to the fact that the detection of intrusion is inefficient and lacks intelligence in current intrusion detection system, this paper integrates BP neural network and support vector machine (SVM) based on the theory of neural network integration, applying fuzzy clustering technology to cluster data, choosing data from the cluster centre to train ensemble individuals, then selecting and integrating those individuals of significant diversity. The theoretical analysis and experimental results show that this ensemble method is efficient for detection rates and unknown attacks.