rough neural intelligent approach for classification

《rough neural intelligent approach for classification》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、RoughNeuralIntelligentApproachforClassificationSanaaRashedAbdallahandYasserFouadHassanDepartmentofMathematics(computerscience),FacultyofScience,AlexUniversityThispaperdescribesroughneuralnetworkwhereneuralnetworksystemsandroughsetstheoryarecompletelyintegratedint

2、oahybridsystemandareusedcooperativelyfordecisionandclassificationsupport.Also,roughsetsandneuralnetworkarechosenforthecombinedmethodbecausetheycandiscoverpatternsinambiguousandimperfectdata,andprovidetoolsfordataandpatternanalysis.Thecommoncharacteristicofroughse

3、tsandneuralnetworksisthatbothapproacheshavetheabilitytolearndecisionmodelsbyexamples.1IntroductionIntelligentsystemscomprisevariousparadigmsdedicatedtoapproximatelysolvingreal-worldproblems,e.g.,indecisionmaking,classificationorlearning;amongtheseparadigmsarefuzz

4、ysets,roughsets,neuralnetworks,andgeneticalgorithms.Fuzzysetsprovideanaturalframeworkfortheprocessindealingwithuncertainty.Itoffersaproblem-solvingtoolbetweentheprecisionofclassicalmathematicsandtheinherentimprecisionoftherealworld.Theimprecisioninanimagecontaine

5、dwithincolorvaluecanbehandledus-ingfuzzysets.Neuralnetworksandroughsetsarewidelyusedforclassificationandrulegeneration.Geneticalgorithms(GAs)areinvolvedinvariousoptimizationandsearchpro-cesses,likequeryoptimizationandtemplateselection.Otherapproacheslikecasebased

6、reasoninganddecisiontreesarealsowidelyusedtosolvedataanalysisproblems.Eachoneofthesetechniqueshasitsownpropertiesandfeaturesincludingtheirabilityoffind-ingimportantrulesandinformationthatcouldbeusefulfordataclassification.Neuralnetworksprovidearobustapproachtoappr

7、oximatingreal-valued,discrete-valuedandvector-valuedfunctions.Thewell-knownalgorithmBack-propagation,whichusesgra-dientdescenttotunenetworkparameterstobestfitthetrainingsetwithinput-outputpair,hasbeenappliedasalearningtechniquefortheneuralnetworks.Roughsetsbaseds

8、ystemsprovidedomainknowledgeexpressedintheformofIf-thenrulesandtoolsfordataanalysis.Unlikeotherintelligentsystems,roughsetanalysisrequiresnoexternalparametersanduseso