Beam sampling for the infinite hidden Markov model

Gael, Jurgen Van; Saatci, Yunus; Teh, Yee Whye; Ghahramani, Zoubin

doi:10.1145/1390156.1390293

Cited by 166 publications

(158 citation statements)

References 18 publications

(18 reference statements)

Supporting

Mentioning

158

Contrasting

Order By: Relevance

“…This allows standard methods for finite mixture models to be used directly. For example, Van Gael et al (2008) fit an infinite hidden Markov model using the forward-backward sampler for finite hidden Markov model using the slice sampling idea. This would be difficult to implement in a retrospective framework since the truncation point changes when updating the allocations.…”

Section: Conclusion and Discussionmentioning

confidence: 99%

Slice sampling mixture models

2009

View full text Add to dashboard Cite

We propose a more efficient version of the slice sampler for Dirichlet process mixture models described by Walker (2007). This sampler allows the fitting of infinite mixture models with a wide-range of prior specification. To illustrate this flexiblity we develop a new nonparametric prior for mixture models by normalizing an infinite sequence of independent positive random variables and show how the slice sampler can be applied to make inference in this model. Two submodels are studied in detail. The first one assumes that the positive random variables are Gamma distributed and the second assumes that they are inverseGaussian distributed. Both priors have two hyperparameters and we consider their effect on the prior distribution of the number of occupied clusters in a sample. Extensive computational comparisons with alternative "conditional" simulation techniques for mixture models using the standard Dirichlet process prior and our new prior are made. The properties of the new prior are illustrated on a density estimation problem.

show abstract

Section: Conclusion and Discussionmentioning

confidence: 99%

Slice sampling mixture models

2009

View full text Add to dashboard Cite

show abstract

“…One possible way to improve the efficiency of sampling would be to use beam sampling techniques similar to those developed for non-parametric Markov models [39]. Another promising option is parallel sampling, which would allow sampling to be run on a number of different CPUs simultaneously [40].…”

Section: Discussionmentioning

confidence: 99%

Bayesian Learning of a Language Model from Continuous Speech

Neubig

Mimura

Mori

et al. 2012

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYWe propose a novel scheme to learn a language model (LM) for automatic speech recognition (ASR) directly from continuous speech. In the proposed method, we first generate phoneme lattices using an acoustic model with no linguistic constraints, then perform training over these phoneme lattices, simultaneously learning both lexical units and an LM. As a statistical framework for this learning problem, we use non-parametric Bayesian statistics, which make it possible to balance the learned model's complexity (such as the size of the learned vocabulary) and expressive power, and provide a principled learning algorithm through the use of Gibbs sampling. Implementation is performed using weighted finite state transducers (WFSTs), which allow for the simple handling of lattice input. Experimental results on natural, adult-directed speech demonstrate that LMs built using only continuous speech are able to significantly reduce ASR phoneme error rates. The proposed technique of joint Bayesian learning of lexical units and an LM over lattices is shown to significantly contribute to this improvement.

show abstract

“…Beam sampling also consist of slice sampling, so number of states are limits by using dynamic programming considered at each time step to a finite number. Using the beam sampler this presents applications of iHMM inference on change point detection and text prediction problems [2].…”

Section: Analysis Of Search Tasksmentioning

confidence: 99%

Fine Grained Knowledge Sharing Mechanism With Hierarchical Structure Using Iterative D-Ihmm

2017

IJAERD

View full text Add to dashboard Cite

show abstract

Beam sampling for the infinite hidden Markov model

Cited by 166 publications

References 18 publications

Slice sampling mixture models

Slice sampling mixture models

Bayesian Learning of a Language Model from Continuous Speech

Fine Grained Knowledge Sharing Mechanism With Hierarchical Structure Using Iterative D-Ihmm

Contact Info

Product

Resources

About