主动学习半监督聚类入侵检测算法  

Research of Intrusion Detection Algorithm Based on Semi-Supervised Clustering

作　　者：李永忠[1] 李正洁[1] 荆春伟[1] 胡翰[1] 

机构地区：[1]江苏科技大学计算机科学与工程学院,江苏镇江212003

出　　处：《微电子学与计算机》2011年第10期28-31,共4页Microelectronics & Computer

基　　金：江苏省高校自然科学基金资助项目(05KJD52006);江苏科技大学科研资助项目(2005DX006J)

摘　　要：针对聚类的入侵检测算法误报率高的问题,提出一种主动学习半监督聚类入侵检测算法.在半监督聚类过程中应用主动学习策略,主动查询网络中未标记数据与标记数据的约束关系,利用少量的标记数据生成正确的样本模型来指导大量的未标记数据聚类,对聚类后仍未能标记的数据采用改进的K-近邻法进一步确定未标记数据的类型,实现对新攻击类型的检测.实验结果表明了算法的可行性及有效性.Aiming at the problem for labeled data that intrusion detection algorithms based on supervised learning,the ASCID algorithm for intrusion detection based on semi-supervised is proposed in this paper,by appling active learning strategy to semi-supervised clustering process.Active learning queries constrains on labeled data and unlabeled data,which uses minimal labeled data to generate the correct sample data model and guide lots of unlabelled data clustering,the algorithm performance is improved by use an improved K-nearest neighbor algorithm to further define the type of unlabeled data after clustering.Finally,the experiment results show the feasibility and validity of the algorithm.

关 键 词：入侵检测 半监督聚类 主动学习 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]