剪枝和網(wǎng)格采樣相結(jié)合的非平衡數(shù)據(jù)集分類(lèi)方法.pdf

1、獨(dú)創(chuàng)性聲明本人聲明所呈交的學(xué)位論文是本人在導(dǎo)師指導(dǎo)下進(jìn)行的研究工作及取得的研究成果據(jù)我所知，除了文中特別加以標(biāo)注和致謝的地方外，論文中不包含其他人已經(jīng)發(fā)表或撰寫(xiě)過(guò)的研究成果，也不包含為獲得蝴其他教育機(jī)構(gòu)的學(xué)位或證書(shū)而使用過(guò)的材料。與我一同工作的同志對(duì)本研究所做的任何貢獻(xiàn)均已在論文中作了明確的說(shuō)明并表示謝意學(xué)位論文作者簽名：掀簽字日期：幽p年、廣月，。日學(xué)位論文版權(quán)使用授權(quán)書(shū)本學(xué)位論文作者完全了解燃有關(guān)保留、使用學(xué)位論文的規(guī)定，有權(quán)保留并

2、向國(guó)家有關(guān)部門(mén)或機(jī)構(gòu)送交論文的復(fù)印件和磁盤(pán)，允許論文被查閱和借閱。本人授杈竊麴啦以將學(xué)位論文的全部或部分內(nèi)容編入有關(guān)數(shù)據(jù)庫(kù)進(jìn)行檢搴，可以采用影印，縮印或掃描等復(fù)制手段保存、匯編學(xué)位論文，。(保密的學(xué)位論文在解密后適用本授權(quán)書(shū))學(xué)位論文作者簽名：烈導(dǎo)師簽名：秀’老—弦3簽字日期妒p年』月，礦日簽字日期：o趴年鋤／曰學(xué)位論文作者畢業(yè)去向：工作單位：電話(huà)：通訊地址：郵編：AbstractImbalanceddatasetsclassifica

3、tionproblemiscommonproblemsinthefieldofpaRernrecognition，machinelearninganddataminingaswellasahotissueImbalanceddatasetisadatasetofcategoriesbecauseofthepresenceofskewnamelyakindofcategorysamplesmorethanothercategoriesof

4、sampleThetraditionalclassifiersinordertopursueahighrateofaccuracyfocusonclassificationaccuracyofthemajorityclasssamplesofImbalanceddatasets，ontheotherhandtheminorityclasssamplesofimbalanceddatasetsshouldbcconsideredbecau

5、seofthecostofclassificationanditstrueinformationTherefore，researchofImbalanceddataprocessingproblemisveryimportantAtpresentdomesticandforeignscholarshaveobtainedsomeachievementsindatapreprocessingandalgorithmsoftwolevela

6、boutimbalanceddatasetsclassificationproblemScholarsaretryingtoimprovethetraditionalalgorithmsandimprovetheclassificationperformanceinImbalanceddatasetonthealgorithmlevelInthedatapretreatmentlevel，scholarsgenerallyremovet

7、henegativesamplesofnoisedataandseparatefromtheclassificationofsurfacedatainundersampling，otherwisetheyaddnoisedatatooversamplingdatainordertobalanceInaword，manymethodsaredifferentondatareductionordataadditionInthispaper’

8、newsamplingmethodsaboutimbalanceddatasetsclassificationareconsideredinordertopreventthefromgeneralundersamplingmethodonthebasisofpreviousstudiesGridsamplingmethodbypruningputsforward，namelythemajorityclassofthesampleswil

9、lbedividedintoabsolutesafetydatadataofedgeandnoisedatabefore鰣dsamplingbasingonadaptiveboostingmethodtocarryOUtsamplingdatalearningTheartificialdataandtypicalUCIdatasetsareverifiedwithROCcurveastheevaluationcriterionbytes

10、tInlightoftestconclusion，theAUCvalueisgreaterthantheothertypesofalgorithms，whichshowsthemodelhasgoodperformanceTheothernewmethod，namelythemixedsamplingontheRandomSMOTEmethod，putforwardandisvalidbytestKeywords：Imbalancedd