数据挖掘分类算法在入侵检测应用中的比较  

Comparison of Data Mining Algorithms in Intrusion Detection Application

作　　者：张虹波[1] 匡银虎[1] 

机构地区：[1]宁夏大学物理电气信息学院,银川750021

出　　处：《现代计算机》2009年第11期14-18,共5页Modern Computer

基　　金：宁夏高等学校科学研究自然科学基金项目(No.2007027);宁夏大学自然科学基金项目(No.ZR0629)

摘　　要：针对当前入侵检测系统普遍存在的误报、漏报和缺乏自适应性问题,采用ODM的分类算法中的决策树分类算法、支持向量机分类算法、朴素贝叶斯算法和二元变量逻辑回归算法等四种重点技术对实验数据进行模型建立和测试,并通过对四种应用结果准确度的对比分析证明本文所采用的技术路线的可行性和生成结果的准确性,从中找出在实际应用中较为常用、直接、有效的和有一定通用价值的方法。Aiming at the current problem of false positives, omissions and lack of self-adapt ability in the intrusion detection system, the four kinds of important classification algorithms in ODM, i.e.,Decision Tree Classification Algorithm,Support Vector Machine Classification Algorithm, Naive Bayes Algorithm and Binary Logistic Regression Algorithm. Uses them to model and test on the experimental data. Through comparing and analyzing accuracy of the results of the four methods, it proves that the technology used in this article is feasible and the result is accurate, then finds the most usually used,the most direct, the most effective and a cer- tain common-valued algorithm for the practical applications.

关 键 词：数据挖掘 决策树 支持向量机 入侵检测 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术] TP311.13[自动化与计算机技术—计算机科学与技术]