the dynamics of probabilistic structural relevance

the dynamics of probabilistic structural relevance

ID:33852792

大小:213.21 KB

页数:12页

时间:2019-03-01

the dynamics of probabilistic structural relevance_第1页
the dynamics of probabilistic structural relevance_第2页
the dynamics of probabilistic structural relevance_第3页
the dynamics of probabilistic structural relevance_第4页
the dynamics of probabilistic structural relevance_第5页
资源描述:

《the dynamics of probabilistic structural relevance》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、TheDynamicsofProbabilisticStructuralRelevanceL.C.vanderGaag&J.-J.Ch.MeyerUtrechtUniversity,DepartmentofComputerScienceP.O.Box80.089,3508TBUtrecht,TheNetherlandsAbstractProbabilisticinferencewithabeliefnetworkingeneraliscomputationallyex-pensive.Sincetheconceptofst

2、ructuralrelevanceprovidesforidentifyingpartsofabeliefnetworkthatareirrelevanttoacontextofinterest,itallowsforal-leviatingtosomeextentthecomputationalburdenofinference:inferencecanberestrictedtothenetwork'srelevantpart.Thestructurallyrelevantpartofabeliefnetwork,ho

3、wever,isnotstatic.Itmaychangedynamicallyasreasoningprogresses.Weaddressthedynamicsofstructuralrelevanceandintroducetheconceptofanindependenceprojectiontocapturethesedynamics.1IntroductionComplexproblemdomainsthatarefraughtwithuncertaintyareinthefocusofat-tentionof

4、arti cial-intelligenceresearchandhavebeensoforsometimenow.Oneofthemostpromisingframeworksfordealingwithuncertaintythathaveemergedfromthisresearchistheframeworkof(Bayesian)beliefnetworks[Pearl,1988].Thisframeworkis rmlyrootedinprobabilitytheory.Itprovidesapowerfula

5、ndintuitivelyappealingformalismforrepresentingaprobabilitydistribution;informallyspeaking,abeliefnetworkconsistsofaqualitativepart,encodingadomain'svariablesandtheprobabilisticindependencesamongtheminadirectedgraph,andaquantitativepart,encodingprobabilitiesoverthe

6、sevariables.Inaddition,theframeworko ersasetofalgorithmsforprobabilisticinference.Thebelief-networkframeworkisbecomingin-creasinglypopularforbuildingknowledge-basedsystems,andmoreandmorereal-lifeapplicationsemployingtheframeworkarebeingrealised.Asapplicationsofthe

7、belief-networkframeworkgrowlarger,thenetworksinvolvedincreaseinsizeaccordingly.Forlargebeliefnetworks,probabilisticinferenceshowsatendencytobecomerathertime-consuming.SinceprobabilisticinferenceisknowntobeNP-hard[Cooper,1990],thistendencymaynotbedeniedingeneral.In

8、manyreal-lifeproblemdomains,however,reasoningwithabeliefnetworkconcentratesononlysomevariablesofinterest.Inamedicaldiagnosticapplication,forexample,them

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。