基于Choquet模糊積分的多分類器系統(tǒng)多樣性研究.pdf

1、河北大學(xué)碩士學(xué)位論文基于Choquet模糊積分的多分類器系統(tǒng)多樣性研究姓名：張國(guó)防申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：應(yīng)用數(shù)學(xué)指導(dǎo)教師：王熙照20070601Abstract量曼曼曼曼曼曼皇皇曼魯量曼I皇曼舅曼皇曼曼皇曼鼉曼皇寡!曼曼曼曼鼉曼曼皇曼皇曼皇量曼曼曼皇笪曼曼皇!曼曼曼曼曼曼曼曼曼曼皇曼量曼曼穹曼曼曼皇曼曼曼皇曼曼曼!曼曼曼量曼AbstractComparedtosingleclassifiermultipleclassifierssys

2、temcallmakeuptheshortthatinstancessetcan’tsatisfyconsistencyassumption(thesamedistribution)andhavehighertrainingandtestingaccuracyCommonintuitionsuggeststhattheclassifiersintheensembleshouldbeasaccurateaspossibleandshoul

3、dnotmakecoincidenterrors，thediversityoftheclassifieroutputsisthereforeafactoronimprovingclassifieraccuracyanditisalsoavitalrequirementfortheSUCCESSofensembleInthisthesiswedefineadiversitymeasureofthefusionsystemonthebase

4、ofthedecisionprofileofmultipleclassifierssystembasedontheChoquetfuzzyintegral，andintroducethissystem，what’Smoreweproposeerrorinequalityandproveit’SsufficiencyInadditionwegivenonlinearityprogrammingwithrespecttomaximumoft

5、hediversityofclassifieroutputs，whichCanmakefuzzymeasurefrominstancesset，atthesametimewediscussanalgorithmforlearningfuzzymeasurebyapplyingthediversityAtlastwehaveallexperimentthatshowthatthediversitymeasureinthethesisisf

6、easible，andmakeaconclusiononthediversityandgeneralizationaccuracyofmultipleclassifierssystem，thatis，thegreaterthediversityofclassifieroutputs，thegreaterthegeneralizationaccuracyofsystemKeyWords：diversity；multipleclassifi