欢迎来到天天文库
浏览记录
ID:40086420
大小:1.23 MB
页数:48页
时间:2019-07-20
《Nonparamater_Bayesian_Model》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、HierarchicalBayesianNonparametricModelswithApplicationsYeeWhyeTehGatsbyComputationalNeuroscienceUnitUniversityCollegeLondon17QueenSquareLondonWC1N3AR,UnitedKingdomMichaelI.JordanDepartmentofStatisticsDepartmentofElectricalEngineeringandComputerScienceUniversityofCalifornia,BerkeleyBerkeley,CA94
2、720,USAFebruary14,2009AbstractHierarchicalmodelingisafundamentalconceptinBayesianstatistics.Thebasicideaisthatparametersareendowedwithdistributionswhichmaythemselvesintroducenewparameters,andthisconstructionrecurses.InthisreviewwediscusstheroleofhierarchicalmodelinginBayesiannon-parametrics,foc
3、usingonmodelsinwhichtheinnite-dimensionalparame-tersaretreatedhierarchically.Forexample,weconsideramodelinwhichthebasemeasureforaDirichletprocessisitselftreatedasadrawfromanotherDirichletprocess.ThisyieldsanaturalrecursionthatwerefertoasahierarchicalDirichletprocess.Wealsodiscusshierarchiesbas
4、edonthePitman-Yorprocessandoncompletelyrandomprocesses.Wedemonstratethevalueofthesehierarchicalconstructionsinawiderangeofpracticalapplications,inproblemsincomputationalbiology,computervisionandnaturallanguageprocessing.1IntroductionHierarchicalmodelingisafundamentalconceptinBayesianstatistics.
5、Thebasicideaisthatparametersareendowedwithdistributionswhichmaythemselvesintroducenewparameters,andthisconstructionrecurses.Acommonmotifinhierarchicalmodelingisthatoftheconditionallyindependenthierarchy,inwhichasetofparametersarecoupledbymakingtheirdistributionsdepend1onasharedunderlyingparamet
6、er.Thesedistributionsareoftentakentobeidentical,basedonanassertionofexchangeabilityandanappealtodeFinetti'stheorem.Hierarchieshelptounifystatistics,providingaBayesianinterpretationoffrequentistconceptssuchasshrinkageandrandomeects.Hierarchiesalsoprovidewaystospecifynon-standarddistributionalfo
7、rms,obtainedasintegralsoverunderlyingparameters.Theyplayaroleincomputationalpracticeintheguiseofvariableaugmentation.Theseadvantagesarewellappreciatedintheworldofparametricmodeling,andfewBayesianparametricmodelersfailtomakeuseofso
此文档下载收益归作者所有
点击更多查看相关文章~~