子空间聚类在入侵检测中的应用  

Application of Sub Space Clustering in Intrusion Detetion

作　　者：张思亮[1] 李广霞[2] 

机构地区：[1]河北广播电视大学,河北石家庄050081 [2]石家庄经济学院信息工程学院,河北石家庄050031

出　　处：《计算机安全》2013年第12期2-6,共5页Network & Computer Security

基　　金：河北省高等学校科学技术研究青年基金项目:2011143;河北省科技计划项目:11213564

摘　　要：在高维数据聚类中,受维度效应的影响,现有的算法聚类效果不佳。在分析现有软子空间聚类算法不足的基础上,引入子空间差异的概念,结合簇内紧凑度的信息来设计新的目标优化函数,提出了一种新的k-means型软子空间聚类算法;针对目前入侵检测实时性和准确性的要求,将离群点扫描技术嵌入新算法中。在KDD Cup1999数据集上的试验表明,该算法能进行高效的特征选择,提高入侵检测的检测精度。In clustering of high dimensional data, most of the existing algorithms can not reach people＇ s expectation due to the curse of dimensionality. By analyzing ]imitations of the existing algorithms, the concept of subspace difference is proposed. Based on these, a new objective function is given by taking into account the compactness of the subspace clusters and sub-space difference of the clusters. And a subspace clustering algorithm based on k-means is presented. Scan processing of outlier is included into this algorithm, which is brought forward for the requirements of real--time characteristic and accuracy in intrusion detection. Experimental results on the KDD CupI999 data show that it can carry out its strong ability against distraction and effectively select feature, thus improving the testing speed and accuracy of intrusion detection.

关 键 词：入侵检测 子空间聚类 检测率 高维数据 特征选择 

分 类 号：TP316.4[自动化与计算机技术—计算机软件与理论]