基于改进SVM主动学习的网络入侵检测  

An Improved Incremental Bayes Classificaiton Model

作　　者：苏志同[1] 刘芳正[1] 

机构地区：[1]北方工业大学计算机学院,北京100144

出　　处：《计算机与数字工程》2016年第9期1770-1773,共4页Computer & Digital Engineering

摘　　要：支持向量机(SVM)主动学习模型能够很好地解决入侵检测系统的小样本学习的问题,提高入侵检测系统中分类器的性能。针对SVM主动学习模型对于构建初始训练集具有随机性,采用核空间聚类的初始训练集构建方法进行优化,并引入蚁群聚类算法减轻样本选择规则对分类性能的影响,结果表明改进后的模型可以有效提高入侵检测系统的分类性能。Support vector machine（SVM）active learning model can solve the problem of small sample learning of intrusion detection system,and improve the performance of the classifier in intrusion detection system.For SVM active learngin model to construct the initial training set is random,the nuclear spatial clustering of the initial training set construction method were optimized,and the introduction of ant colony clustering algorithm reducing sample selection rules on the classification performance effect.The results show that the improved model can effectively improve the intrusion detection system of classification performance.

关 键 词：入侵检测 支持向量机 主动学习 分类性能 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]