基于核函数Fisher鉴别的异常入侵检测  

An Anomaly Intrusion Detection Based on Kernel Fisher Discriminant

作　　者：周鸣争[1] 

机构地区：[1]安徽工程科技学院计算机科学与工程系,芜湖241000

出　　处：《电子与信息学报》2006年第9期1727-1730,共4页Journal of Electronics & Information Technology

基　　金：安徽省教育厅自然科学重点基金(2004kj033zd)资助课题

摘　　要：将核函数方法引入入侵检测研究中,提出了一种基于核函数Fisher鉴别的异常入侵检测算法,用于监控进程的非正常行为。首先分析了核函数Fisher鉴别分类算法应用于入侵检测的可能性,然后具体描述了核函数Fisher鉴别算法在异构数据集下的推广,提出了基于核函数Fisher鉴别的异常入侵检测模型。并以Sendmail系统调用序列数据集为例,详细讨论了该模型的工作过程。最后将实验仿真结果与其它方法进行了比较,结果表明,该方法的检测效果优于同类的其它方法。Kernel method is introduced to intrusion detection and an anomaly intrusion detection method based on kernel Fisher discriminant is presented in this paper. This method is applied for monitoring the abnormal behavior of processes. Firstly, this paper presents the possible of kernel Fisher method applied to intrusion detection. Secondly, this paper descriptions the kernel Fisher algorithm is generalized for heterogeneous datasets. A model of anomaly intrusion detection based on kernel Fisher is given and the working process of this model is used with sendmail system call in detail discussion; Finally, the simulation result is compared with other methods, The measuring result of this method is superior to other similar methods.

关 键 词：异常入侵检测 核函数Fisher鉴别 异构数据集 系统调用 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]