基于原子分类关联规则的网络入侵检测研究  

Study of Network Intrusion Detection Based on Atomic Classified Association Rules

作　　者：郭红艳[1] 谷保平[1] 

机构地区：[1]河南广播电视大学,郑州450000

出　　处：《河南广播电视大学学报》2014年第2期106-109,共4页Journal of Henan Radio & TV University

基　　金：2014年河南省教育厅科学技术研究重点项目(项目编号:14B520052);2014年河南省教育厅人文社会科学研究项目(项目编号:2014-gh-215)阶段性研究成果

摘　　要：由于较好的准确性和鲁棒性表现,关联分类算法已经广泛应用于网络入侵检测领域。然而由于生成的关联规则数目较多,分类模型难以识别,同时导致运行速度慢、占用系统内存大。在本文中,对分类关联规则算法进行了改进,解决了上述问题,并且应用于入侵检测中,使用KDD99数据集进行验证,结果表明系统在保持误报率低的情况下,入侵检测率有所提高。Because of the good accuracy and robustness performance, associative classification algorithm has been widely used in network intrusion detection field. However, due to the number of generated association rules,classification model is difficult to identify. At the same time it can lead to slowly running and take up system memory. In this article, the classification of association rule algorithm is improved, and these problems will be solved. Meanwhile,the classification of association rule is applied to intrusion detection, validated using KDD99 data set. The results finally show that intrusion detection rate is improved while the system keeps the low rate of false positives.

关 键 词：入侵检测 关联规则 原子分类关联规则 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]