多通道分割
Observationdensityfunctions:
j
p(Oji|Si=m, )=
τi
(j,i)jj
N(µ|µµ,m,Σµ,m)N(Ok|µ,Σjm)dµ,
k=1
(3)
wherep(S,F,O|Θ)isgivenin(4)andPe(Θ)istheentropic
priorofmodelΘthat,inthiswork,dependsonAas
w
w
Pe(Θ)∝expPp,qlogPp,q,(7)
w
p
q
w
wherePp,qdenotesatransitionprobabilityinAw.Accord-ingly,intheM-stepoftheEMalgorithm,wewillupdatethetransitionprobabilitiesbymaximizingtheentropicprioras,
w w
A =argmax{log(Pe(Θ|O))+λp(Pp,q 1)},
A
w
p
q
jj
where ={µjµ,m,Σµ,m,Σm|j=1,2;m=1,...,Nj}speci esthetwosegmentalmodels,Njisthenumber
(j,i)
ofstatesinchannelj(hereN1=N2=4),andOkdenotestheobservationofframekinshotiandchan-nelj.Wereferthereaderto[4]formoredetails.
Giventhedual-channelobservationsofTshots,O={Oji|i=1,...,T,j=1,2},thejointlikelihoodisde nedas:
p(S,F,O|Θ)=
1212P(S1)P(S1)P(F1|S1,S1)
T
12
P(Fi|Si,Si)
i=1
T 2
jj
P(Si|Si 1,Fi 1)
T 2
j
p(Oji|Si).
(4)
i=2j=1i=1j=1
TheExpectationMaximization(EM)algorithmisoftenused
forlearningaDBN,suchastheMCSHMM,which ndstheoptimalparametersbymaximizingthelikelihoodfunction(e.g.,(4))throughiterativeExpectation(theE-step)andMaximization(theM-step).However,themaximumlikeli-hood(ML)learningoftheMCSHMMcouldbeproblematicduetothecomplexstatespacespeci edbySandF,andalargesetofmodelparameters,suchasAgivenin(2).
3.2LearningoftheMCSHMM
Therearetwoaspectsinmodellearning,structurelearn-ingandparameterlearning.Theformeroneaimsat ndingacompactande ectivemodelstructurebysimplifyingthestatespaceandreducingtheparameterset,andthelat-teronetriestooptimizethemodelparametersgivenapre-de nedmodelstructure.Moreadvancingly,thetwolearn-ingprocessescouldbeuni edintooneframeworkwherethemodelstructureandparameterscanbeoptimizedsimultane-ously[13,2].Inthiswork,wewillusetheideasofentropicpriorandparameterextinctionproposedin[2]thatresultinamaximumaposteriori(MAP)estimator.ThisMAPestimatorcanbeincorporatedintotheM-stepoftheEMalgorithmwhichwillencourageamaximallystructuredandminimallyambiguousmodel.Thisisaccomplishedbytrim-mingtheweaklysupportedparametersandstates,leadingtoacompactmodelwithgooddeterminism.
Inthiswork,theMAPestimatorfocusesonAde nedin(2)thathasmanypossiblestatetransitionsduetoalargenumberofpossiblestatesofF(N1×N2=16here).Inotherwords,wewanttosimplifyAbykeepingonlyimportantstatetransitions,thatwoulde ectivelyreducethenumberofusefulstatesinFandbalancetheincomingandoutgo-ingtransitionsbetweenthetwolayers.Consequentially,theMAP-basedEMestimator ndtheoptimalparameterby
Θ =argmaxPe(Θ|O),
Θ
(8)
whereistheLagrangemultipliertoensure=1.Usingasimilaroptimizationtechniquediscussedin[2],theMAPestimateofΘcanbeachievedbysettingthederiva-tiveoflog-posteriorgivenin(8)tozero.Consequentially,thisstepenhancestheimportantstatesinFanddrivestheweaklysupportedonestowardszero.Accordingtothetran-sitionsofsmallprobabilities,thestatesinFthatareseldomvisitedcouldbefoundandeventuallyeliminated.Herebythestatespaceisoptimizedbybalancingthestatetransi-tionsbetweenthetwolayers.AfterAisoptimized,otherparametersinΘcanalsobeupdatedintheM-step.
Weresorttothejunctiontreealgorithmin[10]toimple-menttheMAP-basedEMalgorithm.ThejunctiontreeisanauxiliarydatastructurethatcanconvertaDirectedAcyclicGraph(DAG),suchastheMCSHMM,intoanundirectedgraphbyeliminatingcyclesinthegraph,sothatbeliefprop-agationcanbee ectivelyperformedonthemodi edgraphforinferenceandlearning.OurEMalgorithmimplementa-tionwasbasedontheBayesNetMatlabToolboxdevelopedbyKPMurphy1.Afterlearning,theViterbialgorithmcanbeusedtoobtainoptimalstatesequencesforbothchannelsatthe rstlayer,i.e.,S,whichencodestwosemanticstruc-tures,i.e.,cameraviewsandplaytypes.ThesecondlayerstatesetFisavailablethatshowsthedependencybetweentwosemanticstructuresanditistrimmedtosixstatesthatislessthan40%ofthecompletestateset.
λwp
w
qPp,q
4.EXPERIMENTALRESULTS
ThealgorithmwasimplementedinMatlab7.0andtestedonaPCwith3.2GCPUand1Gmemoryforsix30-minNFLvideos.TheproposedMCSHMMiscomparedwithourpreviousmethodsthatincludetheHMMwithGaussianemission(HMM(1))andGMMemission(HMM(2))[3],andtheembeddedGMM-HMM(HMM(3))in[5]andtheSHMMin[4].WealsoimplementedtwoCHMM-basedapproaches[1],amongwhichCHMM(1)usesshot-wisefeaturesliketheHMM(1)andCHMM(2)frame-wisefeaturesliketheSHMM.Weadoptedacoarse-to- neprogressiveinitializationstrat-egythatusesthetrainingresultofasimplermodeltoini-tializeamorecomplexmodel.Forexample,weusedtheK-meanresulttoinitializeHMM(1)whosetrainingresultwasusedtoinitializeHMM(2),andthetrainingresultofSHMMwasusedtoinitializeMCSHMM.Additionally,somepriorknowledgecanalsobeused.Forexample,the rstshotisalwaysinthecentralview.TheinitializationofAmayre-quiresomeactualcalculationofthefrequenciesofdi erentstatetransitionsinacoupleofgames.
(5)
where
Pe(Θ|O)∝P(O|Θ)Pe(Θ)=
S,F
p(S,F,O|Θ)Pe(Θ),
(6)
The theory of segmental ... 暂无评价 33页 免费 Multi-channel Segmental ... 6页 免费 Hidden Markov models for... 暂无评价 7页 1财富值喜欢此文档的...
Hidden Markov models for... 暂无评价 23页 免费 Segmental Hidden Markov ... 暂无评价 8页 免费 Multimedia Communication... 暂无评价 24页 免费 ...
In this Article, a novel monitoring method that combines dynamic multiway principal component analysis (DMPCA) and hidden segmental semi-Markov models (HSMM...
Hidden Markov Models A Hidden Markov Model consists of 1. A sequence of ...The Segmental K-means Algorithm 2. The Baum-Welch (E-M) Algorithm In ...
Hidden Markov Models for Information Extraction Probabilistic Logic Learning Seminar - WS03/04 Supervisor : Prof. Dr. Luc De Raedt Dipl. Inf. Kristian ...
Motivates hidden Markov models (HMM): hidden ...Initial state distribution: Gilbert-Elliott Channel ...Multi-user communication networks Hierarchical ...
Hidden Markov Models I... 1人阅读 13页 2.00 Subspace Distribution ... 3人阅读 4页 2.00 Data-Driven Temporal F... 1人阅读 6页 2...
Motif-based Hidden Markov Models for Multiple Sequence Alignment William N. Grundy ? Charles P. Elkan Dept. of Computer Science & Engineering University ...
龙星计划机器学习 Lecture13 Structured Models Hidden Markov Models versus ...NLP example: In a sentence segmentation task, each segmental state may ...
An Introduction to Hidden Markov Models L. R. Rabiner B. H. juang The basic theory of Markov chains has been known to mathematicians and engineers ...
热门文档