%0 Journal Article
%T A rough sets based pruning method for bagging ensemble
%A MIAODuoqian
%A WANGRuizhi
%A DUANQiguo
%A LIUJiming
%J 重庆邮电大学学报(自然科学版)
%D 2008
%X Ensembletechniquestrainasetofcomponentclassifiersandthencombinetheirpredictionstoclassifynewpatterns.Baggingisoneofthemostpopularensembletechniquesforimprovingweakclassifiers.However,itishardtodeployinmanyrealapplicationsbecauseofthelargememoryrequirementandhighcomputationcosttostoreandvotethepredictionsofcomponentclassifiers.Roughsettheoryisaformalmathematicaltooltodealwithincompleteorimpreciseinformation,whichhasattractedalotofattentionfromtheoryandapplicationfields.Inthispaper,anovelroughsetsbasedmethodisproposedtoprunetheclassifiersobtainedfrombaggingensembleandselectasubsetofthecomponentclassifiersforaggregation.Experimentresultsshowthattheproposedmethodnotonlydecreasesthenumberofcomponentclassifiersbutalsoobtainsacceptableperformance.
%K Rough
%K sets
%K Bagging
%K ensemble
%K Pruning
%K method
%U http://journal.cqupt.edu.cn/jcuptnse/jcuptnse/ch/reader/view_abstract.aspx?file_no=20080317&flag=1