粒子群选择特征和信息增益确定特征权值的入侵检测  

Network intrusion detection based on particle swarm optimization algorithm and information gain

作　　者：黄会群[1,2] 孙虹[1] 

机构地区：[1]中南大学公共卫生学院,长沙410078 [2]湖南财政经济学院信息管理系,长沙410205

出　　处：《计算机应用》2014年第6期1686-1688,1693,共4页journal of Computer Applications

基　　金：湖南省教育厅科研项目(12C0561)

摘　　要：为了提高网络入侵检测正确率,提出一种粒子群算法(PSO)选择特征和信息增益(IG)法确定特征权值的网络入侵检测模型(PSO-IG)。首先采用PSO选择网络入侵特征子集,消除冗余特征;然后采用IG法确定特征子集中的特征权重,并采用支持向量机(SVM)建立分类模型;最后采用KDD CUP 99数据集对PSO-IG的性能进行测试。测试结果表明:PSO-IG消除了冗余特征,降低了输入维数,提高了网络入侵检测速度;通过合理确定特征权值,提高了入侵检测正确率。In order to improve the detection accuracy of network intrusion, a network intrusion detection model named PSO-IG was proposed based on Particle Swarm Optimization （PSO） algorithm and Information Gain （IG）. Firstly, PSO algorithm was used to eliminate redundant features of original network data, and then the weight values of selection features were obtained using IG, and Support Vector Machine （SVM） was used to establish intrusion detection model. Finally, the KDD CUP 99 dataset was used to test the performance of PSO-IG. The results show that the proposed model can eliminate redundant features and reduce the input dimension to improve the detection speed of network intrusion, and it can improve the network intrusion detection accuracy by reasonable selecting weight values.

关 键 词：特征选择 特征权值 入侵检测 粒子群算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]