面向异常检测的关联模式挖掘算法研究

《面向异常检测的关联模式挖掘算法研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、东南大学博士学位论文面向异常检测的关联模式挖掘算法研究姓名：朱玉全申请学位级别：博士专业：计算机应用技术指导教师：孙志挥20030615东南_人学博士学位论文ABSTRACTAlongwiththeprogressofscience&technologyandtherapiddevelopmentofcomputernetwork，networkande-businesshavebecometheonlywaybywhichenterprisesgaincompetitiveadvantages．Howeveahighlyadvancednetworkhasbroughtushigh

2、risk．Thesecurityofthenetworkandinformationsystemhasbecomeasocialproblemwhichpeoplepaymuchattentionto．Howtopredict，identify,evaluatethesepotentialproblems，andmanagethemwitheconomicandeffectivecountermeasureswillbeatechnicalproblemtoberesolvedformodembusinessmanagement，whichwilldeterminatetheexis

3、tenceanddevelopmentofanenterprise．Themostimportantsignificanceofestablishingdefensesystemsofnetworksafetyforenterprises，especiallye—businessenterprises，liesinriskmanagement．Theabnonnitydetectiontechnology,whichistheessentialcomponentofthedefensesystemsofnetworksafety,hasbroughttomanyresearchers

4、’attention，andhaswidemarketforeground．Researchesfromhomeandabroadshowthatitisfeasibleintheorytoapplydataminingtectmologytoabnormitydetection，anditispossibletoconstructsuchasystemintechnologytOO．Theminingforassociationpatternsisquiteanimportantresearchtaskintheresearchondataminingtechnologyforab

5、nomtitydetection．Inthispaper,themodelofassociationpattemsminingforabnormitydetectionanditsalgorithmsareproposed，whichestabIishcorrespondingalgorithmselementsforthedefensesystemsofnetworksafety．Themainresultsandcontentsobtainedinthisdissertationareasfclllows：Firstly,themodelofassociationpatterns

6、miningforabnormitydetectionisproposed．Theproblemsandtechnicaldifficultyofitsalgorithmareanalyzed，whichincludedispersion，efficiency,paralleldatamining，pertinence，incrementalupdatingandSOon．Thepattemcodingandthemethodtocomparetwokindsofpatternsareputforward．Secondly,forthecharacteristicofsecurity

7、auditdata,analgorithmforminingassociationpatternsonabnormitydetectionisproposed．Twokindsofincrementalupdatingalgorithmsarealsopresented．Bythesealgorithms，wecannotonlyfindassociationpattemsoruserbehavior’Spatternsefficientlyfromsec