A clustering-based classifier selection method for network intrusion detection
Aizhong Mi, Linpeng Hai · 2010
This paper applies pattern recognition approach based on classifier selection to network intrusion detection and proposes a clustering-based classifier selection method. In the method, multiple clusters are selected for a test sample. Then, the average performance of each classifier on selected clusters is calculated and the classifier with the best average performance is chosen to classify the test sample. In the calculation of average performance, weighted average is adopted. Weight values are calculated according to the distances between the test sample and each selected cluster. Experiments were done on the intrusion detection dataset of KDD'99 to compare the method with the Clustering and Selection (CS) method. The experimental results show the presented method is effective.