| Network is suffering from endless new attacks with increasing amount of data.Traditional network intrusion detective techniques which employ manual analysis and coding can not adapt to the trend. Information discovery can identify effective, innovative, and potentially useful patterns from data sets.Data mining, as the core of the discovery, can extract hidden patterns from a large amount of data for machine learning and pattern updating, and thus reduce the manual and expertise work. Rough sets theory, as a data mining tool, can effectively deal with the expression and reasoning of incomplete and uncertain information, which makes it feasible for the theory to be used in intrusion detection, so it is important to know how to achieve effective reduction under the premise of not affecting the detective rate.This paper discusses the relationship rough sets theory in dealing with the large amount of data to eliminate redundancies in areas such as the excellent performance through attribute reduction algorithm analysis . Starting from the correspondence in equivalence determined by the domain classification and conditional attribute subsets.The author proposes the attribute reduction algorithm of rough sets directly from the angle of domain classification, discussing the invariance property of positive region of standard rough sets and theβ? variable precision rough sets reduction distributed and reduced under theβvariable precision rough sets. Through a variety of attribute reduction algorithm for analysis and comparison, Discussion of rough sets theory in data mining algorithms to highlight aspects of the network intrusion advantage. In order to verify the end of this article proposed by the reduction algorithm in the intrusion detection aspects of the detection efficiency. Through the KDDCup1999 simulation tests, it proves that the algorithm offers high rate of detecting attacks on DOS and Probe, and U2R and R2L as well. |
PDF Full Text Request