文献类型：期刊文献

中文题名：基于约简支持向量机的快速入侵检测算法

英文题名：Fast Intrusion Detection Algorithm Based on Reduced SVM

作者：张雪芹[1];顾春华[1];吴吉义[2]

机构：[1]华东理工大学信息科学与工程学院,上海200237;[2]杭州师范大学电子商务与信息安全重点实验室,浙江杭州310036

年份：2011

卷号：39

期号：2

起止页码：108

中文期刊名：华南理工大学学报（自然科学版）

外文期刊名：Journal of South China University of Technology(Natural Science Edition)

收录：CSTPCD;;EI(收录号：20111713936162);Scopus;北大核心:【北大核心2008】；CSCD:【CSCD2011_2012】；

基金：国家自然科学基金资助项目(60773094);杭州市电子商务与信息安全重点实验室开放课题项目(HZEB201009)

语种：中文

中文关键词：入侵检测;支持向量机;样本选择;凸包

外文关键词：intrusion detection; support vector machine; sample selection; convex hull

摘要：标准支持向量机(SVM)算法受时间和空间复杂度约束,无法有效地处理大规模网络入侵检测问题.文中基于SVM的几何解释,提出了一种基于并行凸包分解计算和支持向量机的入侵检测分类算法(PCH-SVM).该算法借助凸包的分解和并行计算快速提取训练样本空间几何凸包的顶点,构建约简SVM训练样本集.实验结果表明,该算法可以在不造成精度损失的前提下,降低SVM训练的时空复杂度,加速入侵检测分类器的建模和检测.
Owing to the constraints of time and space complexity,the standard SVM（Support Vector Machine） algorithm cannot effectively deal with large-scale network intrusion detection.In order to solve this problem and in view of the geometric interpretation of SVM,an intrusion detection classification algorithm named PCH-SVM is proposed based on the parallel convex hull decomposition and the SVM.With the help of convex hull decomposition and parallel computing,this algorithm can fast extract the vertices of convex hull of the original training samples to build a reduced SVM training set.Experimental results show that the proposed algorithm can effectively reduce the time and space complexity during SVM training,and speeds up the modeling and detection of intrusion detection classifier without any accuracy loss.