基于KPCA和SVM的网络入侵检测研究  

Research on Intrusion Detection System Based on KPCA and SVM

作　　者：顾钧[1] 

机构地区：[1]上海第二工业大学网络信息中心

出　　处：《计算机仿真》2010年第7期105-107,共3页Computer Simulation

摘　　要：研究无线网络安全检测,针对入侵检测存在先验知识少的情况下推广能力差的问题,为了保证网络运行的安全性,提出了利用核主成分分析(KPCA)和支持向量机(SVM)相结合进行入侵检测的方法。首先用核主元分析对输入变量进行特征提取,消除变量之间的相关性,然后运用网格算法对核参数进行了寻优,通过交叉验证的方法对支持向量机进行参数选择,最后利用所建立好的模型进行预测。利用方法对KDD CUP99数据集进行仿真实验,与传统算法相比,方法对网络入侵检测有很高的识别率,为网络入侵检测提供了依据。Current IDS is an important part of network security.Current IDS has poor generalization ability when given less priority knowledge.In this paper,the KPCA and SVM are adopted to implement intrusion detection.The Kernel Principal Component Analysis method can not only solve the linear correlation of the input but also compress the data.The kernel parameters of KPCA are optimized by grid algorithm and the parameters of support vector machine model are selected by cross validation method.Compared with traditional algorithms,this method can achieve higher detection rate and better generalization,and decrease the time of performance.In the end of the paper,the experiment on KDD CUP99 data set shows the effectiveness and excellent performance of the method.

关 键 词：核主成分分析 支持向量机 入侵检测 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]