资源描述:
《基于分布式学习的大规模网络入侵检测算法》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、ISSN1000-9825,CODENRUXUEWE-mail:jos@iscas.ac.cnJournalofSoftware,Vol.19,No.4,April2008,pp.993−1003http://www.jos.org.cnDOI:10.3724/SP.J.1001.2008.00993Tel/Fax:+86-10-62562563©2008byJournalofSoftware.Allrightsreserved.∗基于分布式学习的大规模网络入侵检测算法1,21,2+1,21,2刘衍珩,田大新,余雪岗,
2、王健1(吉林大学计算机科学与技术学院,吉林长春130012)2(吉林大学符号计算与知识工程教育部重点实验室,吉林长春130012)Large-ScaleNetworkIntrusionDetectionAlgorithmBasedonDistributedLearning1,21,2+1,21,2LIUYan-Heng,TIANDa-Xin,YUXue-Gang,WANGJian1(CollegeofComputerScienceandTechnology,JilinUniversity,Changchun130012
3、,China)2(KeyLaboratoryofSymbolicComputationandKnowledgeEngineeringofMinistryofEducation,JilinUniversity,Changchun130012,China)+Correspondingauthor:Phn:+86-431-85168355,E-mail:daxin222@163.com,daxin222@email.jlu.edu.cnLiuYH,TianDX,YuXG,WangJ.Large-Scalenetworkint
4、rusiondetectionalgorithmbasedondistributedlearning.JournalofSoftware,2008,19(4):993−1003.http://www.jos.org.cn/1000-9825/19/993.htmAbstract:AsInternetbandwidthisincreasingatanexponentialrate,it’simpossibletokeepupwiththespeedofnetworksbyjustincreasingthespeedofp
5、rocessors.Inaddition,thosecomplexintrusiondetectionmethodsalsofurtheraddtothepressureonnetworkintrusiondetectionsystem(NIDS)platforms,andthenthecontinuousincreasingspeedandthroughputofnetworkposenewchallengestoNIDS.InordertomakeNIDSeffectiveinGigabitEthernet,the
6、idealpolicyistousealoadbalancertosplitthetrafficandforwardthemtodifferentdetectionsensors,andthesesensorscananalyzethesplittingdatainparallel.Iftheloadbalancerisrequiredtomakeeachslicecontainingallthenecessaryevidencetodetectaspecificattack,ithastobedesignedcomp
7、licatedlyandbecomesanewbottleneckofNIDS.Tosimplifytheloadbalancer,thispaperputsforwardadistributedneuralnetworklearningalgorithm.Byusingthelearningalgorithm,alargedatasetcanbesplitrandomlyandeachslicedataishandledbyanindependentneuralnetworkinparallel.Thefirstex
8、perimentteststhealgorithm’slearningabilityonthebenchmarkofcircle-in-the-squareandcomparesitwithARTMAP(adaptiveresonancetheorysupervisedpredictivemapping)andBP(backpro