Simulated annealingbasedautomaticfuzzyclusteringcombinedwithANN classification foranalyzingmicroarraydata

Microarray technologyhasmadeitpossibletomonitortheexpressionlevelsofmanygenessimultane- ously acrossanumberofexperimentalconditions.Fuzzyclusteringisanimportanttoolforanalyzing microarray geneexpressiondata.Inthisarticle,areal-codedSimulatedAnnealing(VSA)basedfuzzy clustering methodwithvariablelengthconfigurationisdevelopedandcombinedwithpopularArtificial Neural Network(ANN)basedclassifier.TheideaistorefinetheclusteringproducedbyVSAusingANN classifier toobtainimprovedclusteringperformance.Theproposedtechniqueisusedtoclusterthree publicly availablereallifemicroarraydatasets.Thesuperiorperformanceoftheproposedtechniquehas been demonstratedbycomparingwithsomewidelyusedexistingclusteringalgorithms.Alsostatistical significance testhasbeenconductedtoestablishthestatisticalsignificanceofthesuperiorperformanceof the proposedclusteringalgorithm.Finallybiologicalrelevanceoftheclusteringsolutionsareestablished.