欢迎来到天天文库
浏览记录
ID:37388198
大小:3.79 MB
页数:113页
时间:2019-05-23
《粗糙集扩展模型及其在数据挖掘中的应用研究》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、国防科学技术大学博士学位论文粗糙集扩展模型及其在数据挖掘中的应用研究姓名:赛英申请学位级别:博士专业:管理科学与工程指导教师:陈文伟2002.6.1国防科学技术大学研究生院学位论文ABSTRACTDataminingandknowledgediscoveryindatabasesisanewtechnologyfordrawingknowledgefromdata.Asaneffectiveapproachtoprocessingincomplete,impreciseoruncertaininformation,roughsethasbeenplay
2、inganimportantroleintheareaofdataminingandknowledgediscovery.Thesuccessfulapplicationsofroughsetapproachdependlargelyonthecompletenessofitstheory.Onlyafterthetheoryissystematicallyanddeeplystudied,callroughsetbeappliedtopracticaldomains:Takingasabackgroundtheprojectoftheresearch
3、ofnewtechnologyfordatawarehouseanddataminingintheareaofmanagementdecisionsupport,whichisgrantedbytheNationalNatureScienceFoundafion’thisthesisaddress.bscompletelyandsystematicallythemainresearchcontentsandmethodsofroughsetfromtheaspectsofboththeoryand印plication.Theprimarycontrib
4、utionsofthisthesis~include:‘Havingstudiedtheroughsettheorycarefully,thethesisfindstherelationshipbetweenroughsetandsuchabstracttheoriesasmodallogic,fuzzyset,algebrasystemandinterval-setalgebra,i.e.,roughsetprovidesawelldefinedsemanticalinterpretationtotheseabstracttheories,which
5、enablesUStobetterunderstandthem.Secondly,roughsetbuildsaninter-relationshipamongthesedependenttheoriesandconnectsthemtogether.~Thethesisstudiesthegeneralizedroughsetmodelsandproposesamulti-levelroughsetapproximationmodelCBM-RSbasedonacoveringoftheuniverse.Wevalidatethatthemulti-
6、levelroughsetapproximationinducedbyasequenceofreflexiverelationsisaspecialeaseofCBM-RS.CBM-RSmodelbreaksthroughthelimitofthemulti-levelroughsetmodelsinducedonlybybinaryrelations.Thethesisalsoexaminestheroughsetmodelbasedonclassificationaccoracy.TheMm-RSdataminingapproachgivenlat
7、erisbasedonthemodel.ThethesisproposesaroughsetapproachtominingminimalrulesininconsistentdecisiontablesMIE-RS.Wedealwiththeinconsistencythroughclassificationaccuracy,usingheuristicalgorithmswecangetasetofminimalproductiverulessatisfyingthegivenclassificationaccuracy.Withrespectto
8、theimplementofouralgorithm,weconstructtwoHashfu
此文档下载收益归作者所有