An Intrusion Detection Algorithm Model Based on Extension Clustering Support Vector Machine

Intrusion detection technology is a key research direction in information technology. For intrusion detection method based support vector machine (SVM), there is a big obstacle that the amount of audit data for modeling is very large even for a small network scale, so it´s impractical to directly train SVM using original training datasets. Selecting important features from input dataset leads to a simplification of the problem, however a defect caused is the lack of sparseness. All training data will become the support vectors of SVM, which causes the low intrusion detection speed. We propose a novel SVM intrusion detection algorithm model using the method of extension clustering which is utilized to obtain a subset including support vectors. Through this approximation, the training dataset is downsized and consequently the number of support vectors of ultimate SVM model is reduced, which will greatly help to improve the response time of intrusion detection. Comparing to others, the arithmetic model is simple implement and better performance. So it is worth applying and popularizing.