C4.5算法连续属性的改良应用【外文翻译】

C4.5算法连续属性的改良应用【外文翻译】

ID:351293

大小:79.50 KB

页数:9页

时间:2017-07-26

C4.5算法连续属性的改良应用【外文翻译】_第1页
C4.5算法连续属性的改良应用【外文翻译】_第2页
C4.5算法连续属性的改良应用【外文翻译】_第3页
C4.5算法连续属性的改良应用【外文翻译】_第4页
C4.5算法连续属性的改良应用【外文翻译】_第5页
资源描述:

《C4.5算法连续属性的改良应用【外文翻译】》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、毕业论文(设计) 外文翻译外文原文ImprovedUseofContinuousAttributesinC4.5J.R.QuinlanBasserDepartmentofComputerScienceUniversityofSydney,SydneyAustralia2006AbstractAreportedweaknessofC4.5indomainswithcontinuousattributesisaddressedbymodifyingtheformationandevaluationoftestsoncontinuousattributes.AnM

2、DL-inspiredpenaltyisappliedtosuchtests,eliminatingsomeofthemfromconsiderationandalteringtherelativedesirabilityofalltests.Empiricaltrialsshowthatthemodificationsleadtosmallerdecisiontreeswithhigherpredictiveaccuracies.ResultsalsoconfirmthatanewversionofC4.5incorporatingthesechanges

3、issuperiortorecentapproachesthatuseglobaldiscriminationandthatconstructsmalltreeswithmulti-intervalsplits.1.IntroductionMostempiricallearningsystemsaregivenasetofpre-classifiedcases,eachdescribedbyavectorofattributevalues,andconstructfromthemamappingfromattributevaluestoclasses.The

4、attributesusedtodescribecasescanbegroupedintocontinuousattributes,whosevaluesarenumeric,anddiscreteattributeswithunorderednominalvalues.Forexample,thedescriptionofapersonmightincludeweightinkilograms,withavaluesuchas73.5,andcolorofeyeswhosevalueisoneof`brown',`blue',etc.C4.5(Quinla

5、n,1993)isonesuchsystemthatlearnsdecision-treeclassifiers.SeveralauthorshaverecentlynotedthatC4.5'sperformanceisweakerindomainswithapreponderanceofcontinuousattributesthanforlearningtasksthathavemainlydiscreteattributes.Forexample,Auer,Holte,andMaass(1995)describeT2,asystemthatsearc

6、hesforgoodtwo-leveldecisiontrees,andcomment:TheaccuracyofT2'streesrivaledorsurpassedC4.5'son8ofthe[15]datasets,includingallbutoneofthedatasetshavingonlycontinuousattributes."Discussingtheeffectofreplacingcontinuousattributesbydiscreteattributes,eachofwhosevaluescorrespondstoaninter

7、valofthecontinuousattribute,Dougherty,Kowhai,andSalami(1995)write:C4.5'sperformancewassignificantlyimprovedontwodatasets:::usingtheentropydiscriminationmethodanddidnotsignificantlydegradeonanydataset.::WeconjecturethattheC4.5inductionalgorithmisnottakingfulladvantageofpossibleloca

8、ldiscrimination."Thispaper

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。