高密度特征的网络入侵检测算法研究  

Research on network intrusion detection algorithm for high density features

作　　者：张志华[1] 

机构地区：[1]肇庆学院教育技术与计算机中心,广东肇庆526061

出　　处：《激光杂志》2015年第2期100-103,共4页Laser Journal

基　　金：广东省科技项目(2012B061700063)

摘　　要：为了提高网络入侵检测的正确率,针对特征优化和训练样本选择问题,提出一种高密度的网络入侵特征检测算法。首先提取网络状态特征,然后将特征编码成为粒子的位置向量,通过粒子之间信息共享找到最优特征子集,删除冗余和无效特征,降低特征维数,最后采用模糊均值聚类算法选择最优训练样本,并通过支持向量机建立网络入侵检测器。在Matlab 2012平台上采用标准网络入侵数据库对算法性能进行测试,实验结果表明,相对于其它网络入侵检测算法,本文算法提高了网络入侵检测的正确率和检测效率,获得更加理想的网络入侵检测结果。In order to improve the accuracy of network intrusion detection and solve the feature optimization and training samples selection problem,this paper puts forward a network intrusion detection algorithm for high density features. Firstly,the features of network are extracted,and then features are encoded as position vector of particle,and the best subset of features are found by information sharing among particles to delete the redundant and ineffective features and reduce feature dimension,and finally fuzzy c-means clustering algorithm is used to selecting training samples and support vector machine is used to establish the network intrusion classifier. The model performance is test by using network intrusion data on matlab 2012,and the experimental results show that the proposed algorithm has improved detection accuracy of network intrusion and the detection efficiency and obtain more ideal result of network intrusion compared with other algorithms.

关 键 词：网络入侵检测 特征选择 训练样本集 入侵检测器 

分 类 号：TN911.7[电子电信—通信与信息系统]