资源描述:
《Advances in Minimum Description Length Theory and Applications》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、AdvancesinMinimumDescriptionLengthAdvancesinMinimumDescriptionAdvancesinMinimumDescriptionLengthLengthTheoryandApplicationseditedbyPeterD.Grünwald,InJaeMyung,andMarkA.PittPeterD.GrünwaldisaresearcheratCWI,theNationalTheprocessofinductiveinference—toinfergeneralResearch
2、InstituteforMathematicsandComputerTheoryandApplicationslawsandprinciplesfromparticularinstances—istheScience,Amsterdam,theNetherlands.Heisalsoaffili-computerscience/machinelearningbasisofstatisticalmodeling,patternrecognition,andatedwithEURANDOM,theEuropeanResearchInst
3、itutemachinelearning.TheminimumdescriptivelengthfortheStudyofStochasticPhenomena,Eindhoven,theOfrelatedinterest(MDL)principle,apowerfulmethodofinductiveinfer-Netherlands.InJaeMyungandMarkA.Pittareprofes-ence,holdsthatthebestexplanation,givenalimitedsorsintheDepartmento
4、fPsychologyandmembersofProbabilisticModelsoftheBrainsetofobserveddata,istheonethatpermitsthegreat-theCenterforCognitiveScienceatOhioStatePerceptionandNeuralFunctionestcompressionofthedata—thatthemoreweareableGrünwald,Myung,University.editedbyRajeshP.N.Rao,BrunoA.Olshau
5、sen,andMichaelS.Lewickitocompressthedata,themorewelearnabouttheThisbooksurveyssomeofthecurrentprobabilisticapproachestomodelingandunderstandingbrainfunction,regularitiesunderlyingthedata.AdvancesinMinimumpresentingtop-downcomputationalmodelsaswellasbottom-upneurallymot
6、ivatedmodels.ThetopicscoveredDescriptionLengthisasourcebookthatwillintroduceNeuralInformationProcessingseriesincludeBayesianandinformation-theoreticmodelsofperception,probabilistictheoriesofneuralcodingandspikethescientificcommunitytothefoundationsofMDL,ABradfordBookti
7、ming,computationalmodelsoflateralandcortico-corticalfeedbackconnections,andthedevelopmentofrecenttheoreticaladvances,andpracticalapplications.receptivefieldpropertiesfromnaturalsignals.PeterD.GrünwaldThebookbeginswithanextensivetutorialonMDL,coveringitstheoreticalunder
8、pinnings,practicalimpli-cationsaswellasitsvariousinterpretations,anditsLearningwithKernelsInJaeMyungunderlyingphiloso