The Application Of Fuzzy Classifying Technique On Intrusion Detection

Intrusion detection is a technique in network security system, which can find out unauthorized access and illegitimate using by legal users, take active actions, and protect itself from outside and inside intrusions. This technique is the complement of passive actions, such as system scanner or firewall.The analyze engine design for IDS has close relation with artificial intelligence, and so many advanced techniques is applied including data-mining, clustering, neural network, etc. After studying the intrusion detection system and fuzzy theory, the thesis does some research on the application of feature selection and fuzzy classifying rules in IDS.This thesis is organized as follows. It first studies the application of feature selection in IDS. With feature selection, it can compress the feature space effectively to achieve the goal of depressing the dimension. After combining the feature selection, fuzzy technology and evolutionary strategy together, it then presents a fuzzy classifying algorithm. By turning characteristics of the things fuzzily into several fuzzy characteristics, the fuzzy technology can produce a lot of categorized surfaces and thus improve the accuracy of categorizing. Finally, the thesis draws conclusions by experiments and indicated that the method is of great advantages.