數(shù)據(jù)挖掘中聚類中心問題的光滑化和填充函數(shù)方法.pdf

1、Y907733中圈分類號(hào)：022l單位代號(hào)：11903密級(jí)=學(xué)號(hào)：03720791●—————————●———_叫———__l。_l一上海大學(xué)⑧碩士學(xué)位論文。_。。_。——’_?！瘛猒___●_一●。SHANGHAIUNIVERSITYMASTER’SDissertation題’數(shù)據(jù)挖掘中聚類中心問題目f的光滑化和填充函數(shù)方法作者學(xué)科專業(yè)祝麗華—。。。___●?！猒_●。__●●。__●——?！褚幌到y(tǒng)分析與集成導(dǎo)痧塾坐墮夔攮完成日

2、期2006，5AbstractClusteringcenterpr。blemisanimport村ltclassofDataMillingprogr鋤mingproblemswhichcanbede丘nedastheminimizationofanunconstrainteddob出functionDuetoitswideapplicationsininformatiOnretrie州，矗leextracting，objectandch

3、aracterrecognitionandimagesegmentation，iti80fgreatiIlteresttostudydusteringceⅡterproblemsAclusteringcenterproble珊canbeexpressedasfbl】ows：(』=))minm1∽=去耋。璺劃扎州7n，。，s=l，，口stz=(z1，，z9)∈R”。gwherez1，，，zqaren—dimensionalvariable

4、representingtheceⅡtersofthe口clus—ters，misthenumberofdata，n?！蔙“isthei—thpointindatabaSesincetheobjectivefunction，(茁)isnonconvexandnonsmooth，theprob1em(P)isanonsmoothglobalopti】11iza七ionproblemInthisthesis，weproposeasmooth

5、ingandflⅡedfunctionmethodfor(P)byexploitingthespecialstructureofth8problem，weproposeasmoothjngandoftheproblemUsingthesmooth_ingfIlnction，weappr似imatethenonsm00thnonconvexobjectivefunctionbyasmoothfunctionFilledfunctionme

6、thodisthenadoptedto8earchforaglo—be】optjmalsolutionof亡he印pro](imationpr。blemNumeric缸experimeⅡtsondiflbreIltd礎(chǔ)abase盯ereported如rtheproposedmethodThethesisisorganizedasfollawsInChapter1，we百vesomebriefintr0ductionofDataMinin

7、gandclusteringcenterproblemSomegeneralmethodsarediscussedandrecentresearchprogressonclusteringcenterproblemsisbrieflyintroducedInChapter2wediscusssomeoftheexistingmethodsforsolvingtheclusteringcenterproblemsincludingboth

8、localandtheglobaloptiⅡlizationmechodsbased0ncuttingangelme七h(yuǎn)eodmddiscretegradicn七AnewconvergentsmoothingandfllledfunctionmethodispresentedinChapter3forsolvingthenonsmootjlcIusteringcenterplobIemsNumericaIresuItson8everal