分布式環(huán)境下多分類器識(shí)別和應(yīng)用.pdf

1、西安電子科技大學(xué)碩士學(xué)位論文分布式環(huán)境下多分類器識(shí)別和應(yīng)用姓名：高鳳娥申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：方敏20060101ABSTRACTABSTRACTTheclassificationisbasicproblemofthemachinestudyandpatternrecognition，butwiththechangesoftechnologyandrealisticenvironmentthetraditiona

2、lcategorisedmethodbasedonsinglejointhasalreadyreceivedthegreatchallenge，needtoimproveitinordertomeetdemandsofrealityTothedeficiencyofthetraditionalrecognitionsystem，weproposedthenewalgorithminthispaperThegroundworkofthis

3、textistostructurealotofclassifierbaseonBoostingalgorithmunderthedistributedenvironmentandmergeBoostinganddistributedtoforanorganicwhole，thispaperdealswithsomeprimaryexplorationsbasedonmultiplesitesandmultipleclassifiersu

4、chasanalyticalmethods，frameworkdesignandSOonFirstlyStandardBoostingalgorithmisintroducedindemil，andweproposedtheimprovingalgorithmArecognitionmethodofmultipleclassifiersbaseondistributedBoostingalgorithmhasdesignedandrea

5、lizedInthesystemusingthismethod，F(xiàn)eatureextractingtheoryisdiscussedcomprehensivelyThedatasetofthesystemispretreated，andthehighdimensionsdataisprojectedontolowdimensionsdataSOaStaisetherecognitionratioAsthebasicclassifiera

6、lgnrithmofBoosting，theBPneuralnetworkisalSOdiscussedindetail，andarationalneuralnetworkconfigurationismadesuccessfullywhichissuitableforBoostingInthelast，wetreattheresearchaboutmultipleclassifiersunderthedistributedenviro

7、nmentsuchasdesign，constructionandimprovingetcThenweverifiedsystematicdependabilityfurtherbytestingitwitllthepublicdataaggregatemadeveryeffectiveIthasreachedtherequirementdesignedsystematicallyAtthesanletime，recognitionti