Research On Intrusion Detection System Based On Data Mining

As the development of technology and application of network, it is important that how to quickly and accurately detects the intrusion of network while more and more attack to the computer and network happened. So, it is urgent to study how to find the intrusion in computer network precisely and rapidly.This paper focuses on designing a model for intrusion and the application of data mining on intrusion detection and picking up the feature of intrusion. Association rules and frequent episode algorithm are applied to data aggregation, feature construction and feature selection in data mining, and every recorder represented by these features are analyzed by machine learning to create new rules to the rule library. Axis attribute, reference attribute and level-wise approximate mining algorithm are used to extend to data mining algorithm.This thesis is comprised of three parts. The first one introduces the source of network attack, the composition, function and classification of the intrusion detection system. Then some rudimentary knowledge needed is introduced briefly, which includes the basic conception and the improved algorithms of data mining.The second part focuses on designing the framework of intrusion detection and the application of data mining algorithm in intrusion detection. The intrusion detection system based on data mining is proposed. Association rule algorithm and sequential pattern algorithm are merged into axis attributes and reference feature. Managing the rules and patterns of the knowledge database in combination and generalization makes a great progress of accuracy of IDS. The level-wise approximate mining algorithm proposed to mine the low support frequent feature can upgrades the veracity of the system.In the third part, the network traffic proved by DARPA (Defense Advanced Research Projects Agency) has been analyzed detailed. In the...