Abstract: An intelligent hierarchical intrusion detection method is proposed for getting both high precision and high speed. With this method, an improved AdaBoost algorithm is used in selecting intrusion features and constructing an Ada threshold-classifier at every level, and several hierarchical classifiers are combined for detection. A Linux IDS experimental platform is designed and implemented to train and test the intelligent intrusion detector. Experimental results show that the method reduces the complexity of computation, and that the false negative rate is reduced greatly while maintaining the high detection rate. Moreover, the method improves the processing speed and is especially appealing for the real-time processing of the intrusion detection system.

Key words: intrusion detection, AdaBoost algorithm, feature selection, Ada threshold-classifier, hierarchical structure