“…Thebasicideaoftrainingandtestingalgorithmusedin (Wangetal.,2008;Wang,Shao,&Luo, 1998)fortheMNNinFigure1isgivenasfollows.First,thesamplesetisdividedintothetraining setandthetestingset.Then,thetrainingsetisclusteredwithfuzzyC-meansClusteringmethod whichfunctionsasthetaskdistributorandthecenterC i ,i K = 1, ..., ofeachclusterisobtained.Then, K subnetsareconstructedtotraineveryclassofthetrainingset.Afterthetraining,eachmemberof thetestingset,i.e.,testingdatum,issentintoallthesubnetsandtheoutputofthei-thsubnetis y i . Themembershipdegreeofthistestdatumtothei-thsubnetisdenotedby…”