02 - A Version Space Approach to Learning Context-free Grammars.pdf

02 - A Version Space Approach to Learning Context-free Grammars.pdf

ID:34815221

大小:1.59 MB

页数:36页

时间:2019-03-11

02 - A Version Space Approach to Learning Context-free Grammars.pdf_第1页
02 - A Version Space Approach to Learning Context-free Grammars.pdf_第2页
02 - A Version Space Approach to Learning Context-free Grammars.pdf_第3页
02 - A Version Space Approach to Learning Context-free Grammars.pdf_第4页
02 - A Version Space Approach to Learning Context-free Grammars.pdf_第5页
资源描述:

《02 - A Version Space Approach to Learning Context-free Grammars.pdf》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、MachineLearning2:39-74,1987©1987KluwerAcademicPublishers,Boston-ManufacturedinTheNetherlandsAVersionSpaceApproachtoLearningContext-freeGrammarsKURTVANLEHN(VANLEHN@A.PSY.CMU.EDU)WILLIAMBALL(BALL@A.PSY.CMU.EDU)DepartmentofPsychology,Carnegie-MellonUnivers

2、ity,Pittsburgh,PA15218U.S.A.(Received:June17,1986)(Revised:January21,1987)Keywords:Induction,grammaticalinference,versionspace,context-freegrammars,learningfromexamples.Abstract.Inprinciple,theversionspaceapproachcanbeappliedtoanyinductionproblem.Howeve

3、r,insomecasestherepresentationlanguageforgeneralizationsissopowerfulthat(1)someoftheupdatefunctionsfortheversionspacearenoteffectivelycomputable,and(2)theversionspacecontainsinfinitelymanygeneralizations.Theclassofcontext-freegrammarsisasimplerepresenta

4、tionthatexhibitstheseproblems.Thispaperpresentsanalgorithmthatsolvesbothproblemsforthisdomain.Givenasequenceofstrings,thealgorithmincrementallyconstructsadatastructurethathasnearlyallthebeneficialpropertiesofaversionspace.Thealgorithmisfastenoughtosolve

5、smallinductionproblemscompletely,anditservesasaframeworkforbiasesthatpermitthesolutionoflargerproblemsheuristically.Thesamebasicapproachmaybeappliedtorepresentationsthatincludecontext-freegrammarsasspecialcases,suchasAnd-Orgraphs,productionsystems,andHo

6、rnclauses.1.IntroductionTheproblemaddressedherearoseinthecourseofstudyinghowpeoplelearnarithmeticproceduresfromexamples(VanLehn,1983a;VanLehn,1983b).Ourdataallowedustoinferapproximationsoftheproceduresthesubjectshadlearnedandtheexamplestheyreceiveddurin

7、gtraining.Thus,theinputsandoutputstothelearningprocesswereknown,andtheproblemwastodescribethelearningprocessindetail.However,becausethesubjects'learningoccurredintermittentlyoverseveralyears,wewerenotimmediatelyinterestedindevelopingadetailedcognitivesi

8、mulationoftheirlearningprocesses.Evenifsuchasimulationcouldbeconstructed,itmightbesocomplicatedthatitwouldnotshedmuchlightonthebasicprinciplesoflearninginthistaskdomain.Therefore,ourinitialobjectivewastofindprinciplesthatcouldact

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。