返回
data mining neural networks with genetic algorithms

data mining neural networks with genetic algorithms

ID：14201263

大小：75.91 KB

页数：12页

时间：2018-07-26

data mining neural networks with genetic algorithms_第1页
data mining neural networks with genetic algorithms_第2页
data mining neural networks with genetic algorithms_第3页
data mining neural networks with genetic algorithms_第4页
data mining neural networks with genetic algorithms_第5页
资源描述：

《data mining neural networks with genetic algorithms》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、DataminingneuralnetworkswithgeneticalgorithmsAjitNarayanan,EdwardKeedwellandDraganSavicSchoolofEngineeringandComputerScienceUniversityofExeterExeterEX44PTUnitedKingdomajit@dcs.ex.ac.uktel:(+)1392264064AbstractItisanopenquestionastowhatisthebestwaytoextractsymbolicrulesfromtrainedneuralnetwork

2、sindomainsinvolvingclassification.Previousapproachesbasedonanexhaustiveanalysisofnetworkconnectionandoutputvalueshavealreadybeendemonstratedtobeintractableinthatthescale-upfactorincreasesexponentiallywiththenumberofnodesandconnectionsinthenetwork.Anovelapproachusinggeneticalgorithmstosearchfo

3、rsymbolicrulesinatrainedneuralnetworkisdemonstratedinthispaper.Preliminaryexperimentsinvolvingclassificationarereportedhere,withtheresultsindicatingthatourproposedapproachissuccessfulinextractingrules.Whileitisacceptedthatfurtherworkisrequiredtoconvincinglydemonstratethesuperiorityofourapproa

4、choverothers,thereisneverthelesssufficientnoveltyintheseresultstojustifyearlydissemination.(Ifthepaperisaccepted,thelatestresultswillbereported,togetherwithsufficientinformationtoaidreplicabilityandverification.)IntroductionArtificialneuralnetworks(ANNs)areincreasinglyusedinproblemdomainsinvo

5、lvingclassification.Theyareadeptatfindingcommonalitiesinasetofseeminglyunrelateddataandforthisreasonareusedinagrowingnumberofclassificationtasks.Unfortunately,acommonlyperceivedproblemwithANNswhenusedforclassificationisthat,whileatrainedANNcanindeedclassifythedata,sometimeswithmoreaccuracytha

6、natraditional,symbolicmachinelearningapproach,thereasonsfortheirclassificationcannotbefoundeasily.TrainedANNsarecommonlyperceivedtobe‘blackboxes’whichmapinputdataontoaclassthroughanumberofmathematicallyweightedconnectionsbetweenlayersofneurons.WhiletheideaofANNsasblackboxesmaynotbeaproblemina

7、pplicationswherethereislittleinterestinthereasonsbehindclassification,thiscanbeamajorobstacleinapplicationswhereitisimportanttohavesymbolicrulesorotherformsofknowledgestructure,suchasidentificationordecisiontrees,whichareeasilyinterpretableby

当前文档最多预览五页，下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页，下载文档查看全文
温馨提示：
1. 部分包含数学公式或PPT动画的文件，查看预览时可能会显示错乱或异常，文件下载后无此问题，请放心下载。
2. 本文档由用户上传，版权归属用户，天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容，确认文档内容符合您的需求后进行下载，若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误，付费完成后未能成功下载的用户请联系客服处理。