摘要
针对目前基于异常入侵检测方法所存在的问题,提出了一种基于异类挖掘的聚类方法,该方法通过对采用多种不同类型描述的连接记录对象进行异类数据挖掘,从而实现从大流量网络活动记录数据中快速检测出与正常系统与网络活动相异的已知或未知入侵行为。最后利用KDD'99入侵检测大赛的数据对所提方法进行了检验,实验结果表明了此方法是有效的。
s This paper proposes a new clustering method based on the outlier mining in order to solve the problems of intrusion detection based on outlier. This new method is able to find out novel intrusions involved in large amount of normal system and network activities quickly and effectively. Finally the experiment result made on test data of KDD 1999 intrusion detection competition cup indicates the efficiency of the method.
出处
《计算机工程》
CAS
CSCD
北大核心
2003年第13期125-127,共3页
Computer Engineering