欢迎来到天天文库
浏览记录
ID:36621118
大小:877.20 KB
页数:36页
时间:2019-05-13
《基于粗糙集的数据简约算法研究及设计》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、昆明理工大学硕士学位论文基于粗糙集的数据简约算法研究及设计姓名:范敏申请学位级别:硕士专业:系统理论指导教师:刘文奇20031201AbstractRoughsettheory,introducedbyPawlakintheearly1980s,isanewmathematicaltooltodealwithvaguenessanduncertainty.Pawlakfirstlycreatedtheconceptions:aqualityoflowerapproximationandaqualityofupperapproximation,reductiona
2、ttributes,core,boundary—area,etc.Theseconceptionsessentiallyemphasizedthereductionoftherelatedinformationsystem.Meanwhiletheseconceptionsactedasthebaseofthe‘‘KnowledgeDiscoveryinDatabase”(rd)D,orDataMining).Thisapproachseemstobeoffundamentalimportancetoartificialintelligence(AI)andco
3、gnitivesciences,especiallyintheareasofmachinelearning,knowledgeacquisition,decisionanalysis,knowledgediscoveryfromdatabases,expertsystems,decisionsupportsystems,inductivereasoning,andpattemrecognition.Inparticular,theroughsetapproachhasfoundinterestingapplicationssetinmedicine,pharma
4、cology,business,banking,marketresearch,engineeringdesign,meteorology,vibrationanalysis,switchingfunctions,conflictanalysis,imageprocessing,voicerecognition,concurrentsystemanalysis,decisionanalysis,characterrecognition,andotherfieldsThispaperbeginswiththebasictheory,thenintroducesthe
5、mainIIpartsoftheauthorswork.Theyare(1)Firstpart一一researchtheSkowrondiscernibilitymatrix,thesignificantandrelevanceofattributes;Meanwhile,usingtheconditionalentropytoproposeaheuristicalgorithmforattributereduction;Then,correctthemistakeinthereference[17】;finally,theexperimentshowsthat
6、itCangetbettereffectanditalsoshowsthereductionresultsofUCIdatabaseusingthisalgorithm;(2)Secondpa疗一Itbeginswiththevariableuniversaltheory,thenusesthediscemibiliVmatrixtodefinetheundistributedclass,whichproposingaheuristicalgorithmforattributereduction;finally,theexperimentshowsthatitC
7、angetbettereffectanditalsoshowsthereductionresultsofUCIdatabaseusingthisalgorithm.KEYWORDSRoughSettheory,variableuniverse,Attributereduction,relevantreduction,Discernibilitymatrix,Undistributedclass111昆明理工大学学位论文原创性声明Y669296本人郑重声明:所呈交的学位论文,是本人在导师的指导下进行研究工作所取得的成果。除文中已经注明引用的内容外,本论文不合任何其
8、他个人或集体已经发表或撰
此文档下载收益归作者所有