一種基于密度的動(dòng)態(tài)參數(shù)單元聚類算法.pdf

1、分類號(hào)：TP311密級(jí)：y971269單位代碼：10422學(xué)號(hào)：200211615厶茹只呈碩士學(xué)位論文論文題目：一種基于密度的動(dòng)態(tài)參數(shù)單元聚類算法作者姓名鮑洪慶專業(yè)計(jì)算機(jī)軟件與理論指導(dǎo)教師姓名專業(yè)技術(shù)職務(wù)石冰教授2005年4月5日山東大學(xué)碩士學(xué)位論文^BSTRACTInthisarticle，concepts，techniquesandalgorithmsaboutclusteringwillbediscussedWegiveadyna

2、micparametersolutionforparameterselectionproblemindensitybasedclusteringalgorithmAmongthevariousalgorithmsputforward，amainclassofthemarebasedon“distance”，whetheritisinthesenseoftraditionalEculiddistanceorothers“Kmeans”an

3、d“kmedoids”aretwoofthiskin&HoweverthesealgorithmsareinefficientwhendealingwithlargedatasetsanddatasetsofhighdimensionFurthermore，thenumberofclusterstheycanfindusuallydependsonusers’inputButthistaskisoftenaverytoughonefor

4、theuserInthisarticlewegiveasolutionforthisparameterselectionproblem，calledadynamicparametercomputingalgorithmThealgorithminthisarticlediffersmuchwithaboveonesandittakesatotallydifferentapproach，whichwecallagridanddensity

5、basedalgorithmItcanautomaticallyfindoutsubspacescontaininginterestingpatternswewantanddiscoverallclustersinthatsubspaceBesides，itperformswel1whendealingwithhighdimensionaldataandhasgoodscalabilitywhenthesizeofthedatasets