基于融合分类和支持向量机的入侵检测研究

被引:49
作者
肖海军
洪帆
张昭理
廖俊国
机构
[1] 华中科技大学计算机科学与技术学院
关键词
支持向量机; 误用检测; 融合分类;
D O I
暂无
中图分类号
TP393.08 [];
学科分类号
0839 ; 1402 ;
摘要
为了在提高入侵检测的检测率的同时降低虚警率,基于融合分类和支持向量机的异常检测利用融合分类器进行入侵检测。融合分类器包含三个不同的分类器:基于属性选择的支持向量机,基于样本剔除的支持向量机以及标准支持向量机。仿真实验由三部分组成:首先,预处理数据,然后,对完成预处理的数据分别用三个分类器进行预分类,最后,由这三个分类器实际输出的加权和进行融合决策。权值的最优化是一个NP-hard问题,在实验中,利用各分类器预分类的检测率作为其对应的权值简化了权值寻优的过程。实验结论表明,基于融合分类和支持向量机的异常检测可提高入侵检测的整体性能。
引用
收藏
页码:130 / 132+145 +145
页数:4
相关论文
共 8 条
[1]  
Application of Machine Learning Algorithms to KDD Intrusion Detection Dataset within Misuse Detection Context. Maheshkumar Sabhnani,,Gürsel Serpen. Proceedings of the International Conference on Machine Learning, Models, Technologies and Applications, MLMTA‘03 . 2003
[2]  
KDD-99 Classifier Learning Contest LLSoft s Results O-verview. I Levin. SIGKDD Explorations,ACM SIGKDD . 2000
[3]  
Alinear ge-netic programming approach to intrusion detection. D.Song,MI Heywood,and ANur Zincir-Heywood. Proceed-ings of the Genetic and Evolutionary Computation Conference GEC-CO 2003 .
[4]  
PNrule:A new Framework for Learning Clas-sifier Models in Data Mining(A Case-Study in Network IntrusionDetection). Agarwal MV Joshi. Proceedings of the First SIAMConference on Da-ta Mining . 2001
[5]  
Feature ranking and selection for intru-sion detection. S Mukkamala,A HSung. Proceedings of the International Conference onInformation and Knowledge Engineering . 2002
[6]  
Applica-tion of PSVM and Data Processing for Intrusion Detection. Haijun Xiao,Fan Hong,Zhaoli Zhang,and Junguo Liao. Proceedings of International conference on Bio-Inspired Compu-ting:Theory and Applications(BIC-TA2006) .
[7]  
The Nature of Statistical Learning Theory. V N Vapnik. . 1995
[8]  
Ensemble Learning for IntrusionDetection in Computer Networks. L Didaci,G Giacinto,and F Roli. Proceedings of theWorkshopon Machine Learning,Methods and Applications,held in the con-text of the 8th Meeting of the Italian Association of Artificial Intelli-gence(AI.IA) . 2002