Abstract: The security of computer networks plays a strategic role in modern society. This paper applies pattern recognition approach based on classifier selection to intrusion detection and presents a network intrusion detection approach based on static classifier selection. The approach reduces the errors of static classifier selection and improves the detection performance by partitioning every clustering area with a new method，selecting a classifier according to the sub-areas, and combing the nearest neighbor rule. Clustering and selection (CS) is a typical method of static classifier selection. Experiments have been done on the intrusion detection dataset of KDD’99 and the results show that the proposed approach is superior to the one based on clustering and selection.

Key words: Static classifier selection, Network intrusion detection, Clustering and selection, Pattern recognition

摘要： 计算机网络的安全在当今社会起着举足轻重的作用。该文将基于分类器选择的模式识别方法应用于入侵检测，提出了一种基于静态分类器选择的网络入侵检测方法。该方法对经过聚类获得的各个区域采用新的策略进一步划分，在划分后的子区域上选择分类器，结合了最近邻规则，减小静态分类器选择方法的误差，提高了检测性能。聚类选择(CS)是典型的静态分类器选择方法，在KDD’99的入侵检测数据集上的实验表明，该方法的性能优于基于聚类选择的网络入侵检测方法。

关键词: 静态分类器选择, 网络入侵检测, 聚类选择, 模式识别