基因數(shù)據(jù)分析中聚類算法的研究與應(yīng)用.pdf

1、廈門大學(xué)碩士學(xué)位論文基因數(shù)據(jù)分析中聚類算法的研究與應(yīng)用姓名：葉興甫申請學(xué)位級別：碩士專業(yè)：人工智能基礎(chǔ)指導(dǎo)教師：周昌樂201105AbstractAstwoofthemostimportanttoolsofthedataanalysis，ClusteringandFormalConceptAnalysis(FCA)havebeenwidelyusedintheareaofMathematics，ComputerScience，Biolo

2、gyandSOonForthebiologicalresearchareagenechipshaveproducedaconsiderableamountofgeneexpressiondataAdditionallythesedatacarrymuchimportantinformationTherefore，theproblemsofhowtollsethesetwopowerfultoolshavebecometheresearc

3、hfocusesCurrentlytherearemanyclassicalgorithmsonClusteringandapproachesofconstructingConceptLatticesincludingtheHierarchicalClusteringAnalysisKmeansClusteringAlgorithmSOMsandSOonHoweverallofthesealgorithmsandapproachesha

4、vethdrdisadvantagesrespectivelylikethelargetimecomplexitythesensitivitytotheinitialdustercenters，theinstabilityoftheclusteringresultsandSOonInadditiontothattheHierarchicalClusteringAlgorithmandKmeansClusteringAlgorithmal

5、ebo廿lrequiredtheappropriateKfortheclusteringwhichishardtobedetermined，evenfortheresearchers，whocan’tknowthestructuresanddistributionsofthelargedatasetsbeforealotofexperimentsMeanwhilegeneexpressiondatahavethefeatureoffuz

6、ziness，whichmakesithardertobeprocessedbytheclassicalclusteringalgorithmsAlthough，tosomeextent，theFuzzyCmeans(FCM)algorithmcansolvetheproblemoffuzziness，westillcan’tgettoknowtheinnerrelationsbetweenthegenesamplesfromthatA

7、ccordingtotheadvantagesanddisadvantagesofthealgorithmswementionedabove，thisthesisdiscussestheapplicationofthecombinationofthesetwoimportanttools(FCMClusteringAlgorithmandFormalConceptAnalysis)togeneexpressiondataFirst，we

8、discussandanalyzetheprincipleoftheclassicalclusteringalgorithmsandtheapplicationtogeneexpressiondataThen，accordingtothefuzzinessofgeneexpressiondataweusetheFCMclusteringalgorithmtosolvetheproblemoffuzzinesstosomeextentAd