基于集成学习的入侵检测方法  

Ensemble Learning Based Intrusion Detection Method

作　　者：徐冲[1] 王汝传[1] 任勋益[1] 

机构地区：[1]南京邮电大学计算机学院,南京210003

出　　处：《计算机科学》2010年第7期217-219,224,共4页Computer Science

基　　金：国家自然科学基金(60973139;60773041);江苏省自然科学基金(BK2008451);国家高科技863项目(2007AA01Z404;2007AA01Z478);现代通信国家重点实验室基金(9140C1105040805);国家和江苏省博士后基金(0801019C;20090451240;20090451241);江苏高校科技创新计划项目(CX08B-085Z;CX08B-086Z);江苏省六大高峰人才项目(2008118)资助

摘　　要：为解决传统入侵检测中存在的检测效率低、对未知的入侵行为检测困难等问题,提出了将改进的BP神经网络算法和支持向量机集成的入侵检测模型。实验表明,集成改进的BP神经网络和支持向量机与检出率最好的单个神经网络、单个SVM相比检测率有所提高,同时提高了对未知入侵行为的识别。In order to solve the problem of low detection rate for novel attacks and the difficulties in detecting unknown intrusions existing in traditional intrusion systems, the paper proposed a model based on ensemble learning in improved BP neural networks and support vector machines. Experiments show that using the ensemble learning method, the detection rate is higher than that of using any individual networks and svm. So it has a better detection rate not only to the known intrusion, but also to the unknown intrusion.

关 键 词：入侵检测 集成学习 BP神经网络 支持向量机 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]