An_introduction_to_graphical_models

An_introduction_to_graphical_models

ID:40879017

大小:185.53 KB

页数:19页

时间:2019-08-09

An_introduction_to_graphical_models_第1页
An_introduction_to_graphical_models_第2页
An_introduction_to_graphical_models_第3页
An_introduction_to_graphical_models_第4页
An_introduction_to_graphical_models_第5页
资源描述:

《An_introduction_to_graphical_models》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、AnintroductiontographicalmodelsKevinP.Murphy10May20011IntroductionThefollowingquotation,fromthePrefaceof[Jor99],providesaveryconciseintroductiontographicalmodels.Graphicalmodelsareamarriagebetweenprobabilitytheoryandgraphtheory.Theyprovideanaturaltoolfordealingwithtwoproblemstha

2、toccurthroughoutappliedmathematicsandengineering{uncertaintyandcomplexity{andinparticulartheyareplayinganincreasinglyimportantroleinthedesignandanalysisofmachinelearningalgorithms.Fundamentaltotheideaofagraphicalmodelisthenotionofmodularity{acomplexsystemisbuiltbycombiningsimple

3、rparts.Probabilitytheoryprovidesthegluewherebythepartsarecombined,ensuringthatthesystemasawholeisconsistent,andprovidingwaystointerfacemodelstodata.Thegraphtheoreticsideofgraphicalmodelsprovidesbothanintuitivelyappealinginterfacebywhichhumanscanmodelhighly-interactingsetsofvaria

4、blesaswellasadatastructurethatlendsitselfnaturallytothedesignofecientgeneral-purposealgorithms.Manyoftheclassicalmultivariateprobabalisticsystemsstudiedin eldssuchasstatistics,systemsengineering,informationtheory,patternrecognitionandstatisticalmechanicsarespecialcasesofthegene

5、ralgraphicalmodelformalism{examplesincludemixturemodels,factoranalysis,hiddenMarkovmodels,Kalman ltersandIsingmodels.Thegraphicalmodelframeworkprovidesawaytoviewallofthesesystemsasinstancesofacommonunderlyingformalism.Thisviewhasmanyadvantages{inparticular,specializedtechniquest

6、hathavebeendevelopedinone eldcanbetransferredbetweenresearchcommunitiesandexploitedmorewidely.Moreover,thegraphicalmodelformalismprovidesanaturalframeworkforthedesignofnewsystems.Inthispaper,wewill eshoutthisremarkbydiscussingthefollowingtopics:Representation:howcanagraphicalmo

7、delcompactlyrepresentajointprobabilitydistribution?Inference:howcanweecientlyinferthehiddenstatesofasystem,givenpartialandpossiblynoisyobservations?Learning:howdoweestimatetheparametersandstructureofthemodel?Decisiontheory:whathappenswhenitistimetoconvertbeliefsintoactions?

8、Applications:whathasthismachinerybeenusedfor?1P

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。
相关文章
更多
相关标签