資源描述：
《基于遺傳算法的模糊聚類挖掘方法應(yīng)用研究》由會員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、ABSTRACTWithinformationtechnologyandveryfastspeed，informationprocessingdatabasetechnologydevelopingathasbecomeaindispensabletoolforpeopletoacquireusefulmessage．Dataminingisagenericknowledgediscoverytechnology，itisaprocessoffindingmodelandtherelationshipofthedatainalargeamountofdatabyanalyti
2、caltools．Clusteringanalyticalisanimportantcomponentofdataminingtechnology．Dataclusteringminingtechnologyisanemergingareawhichinvolvesvariousareas·FCM(Fuzzyc．means)algorithm，asakindofunsupervisedlearningmethods，itisaresearchhotspotconcerningaboutdataclusteringanalyticaltechnology．FCMisoneofi
3、mportantalgorithmindataclusteringminingmethods．ithasthecharacteristicsassimple，fastconvergenceandstronglocalsearchingpower，etc．However，F(xiàn)CMissensitivetoinitializationandtendstoresultinlocalminimuminiterations．GeneticAlgorithmisarandomsearchingglobaloptimizationalgorithm．Itisacomputationalmod
4、elofthehumanevolution，withimplicitparallelismandcapacityofusingeffectiveglobalinformation．ThecombinationofFCMalgorithmandgeneticalgorithmwillgetahybridalgorithmwhichbenefitstosolveclusteringproblemandmaketremendousimprovementinalgorithmperformance，thehybridalgorithmhasgoodglobalandlocalsear
5、chcapability．ThispaperpresentsahybridFuzzyC-meansalgorithm(IG·FCM)basedonimprovedgeneticalgorithm．Thealgorithmuseglobalsearchabilityofgeneticalgorithmtooptimizetheinitialclustercentersofclusteringalgorithm，andthencarryouttheFCMalgorithmbaseonlocaloptimization·IG．FCMisaHeuristicclusteringalg
6、orithm，itorderlychangesclusteringclassnumber，thenautomaticallydeterminetheoptimalnumberofclusteringclassandtheoptimalclusteringbaseonEvaluationofclusteringvalidityfunction．Becauseofthetraditionalgeneticalgorithmhasshortcomingslikeslowconvergence，poorstabilityandlowaccuracyrate·Thispaperadop
7、tstheoptimumpreservationstrategyintheselectionoperationtomaintaintheoptimumindividualintheprocessofgenetic，andIllthencopytheselectedindividual，thentheoptimumindividualentersthenextgenerationdirectlywithoutparticipatingincrossoverandmutatIonoperation．Thec
