摘要
针对传统的基于特征检测的入侵检测系统处理的数据常含有大量的冗余特征,使得系统的特征提取和后续处理消耗大量系统资源,导致实时性差,影响检测效果的问题,文章利用粗糙集理论进行特征约简,消除冗余和噪音特征并基于精简后特征子集训练支持向量机,再由训练后的分类器进行入侵检测的方法,以达到提高入侵检测系统的实时性能。实验结果表明了该方法的有效性。
The data based on the traditional feature-based intrusion detection system deal with often contains many redundant features, making the feature extraction and following treatment consume a large amount of system resources, resulting in bad real-time and impacting detection effect. This paper makes use of rough set theory characteristics of reduction to eliminate redundant and noisy features and then trains support vector machine classifier based on the reduced feature subset for intrusion detection in order to improve real-time performance of intrusion detection system. The experimental result shows the effectiveness of the method.
出处
《四川理工学院学报（自然科学版）》
CAS
2009年第5期62-64,共3页
Journal of Sichuan University of Science & Engineering(Natural Science Edition)
基金
国家高技术研究发展计划(863计划)(2008AA11A134)
关键词
粗糙集
特征选择
入侵检测系统
支持向量机
rough sets
feature selection
intrusion detection system
support vector machine
