Stochastic Gradient MCMC Methods for Hidden Markov Models

Yi-An Ma, Nicholas J. Foti, Emily B. Fox
Proceedings of the 34th International Conference on Machine Learning, PMLR 70:2265-2274, 2017.

Abstract

Stochastic gradient MCMC (SG-MCMC) algorithms have proven useful in scaling Bayesian inference to large datasets under an assumption of i.i.d data. We instead develop an SG-MCMC algorithm to learn the parameters of hidden Markov models (HMMs) for time-dependent data. There are two challenges to applying SG-MCMC in this setting: The latent discrete states, and needing to break dependencies when considering minibatches. We consider a marginal likelihood representation of the HMM and propose an algorithm that harnesses the inherent memory decay of the process. We demonstrate the effectiveness of our algorithm on synthetic experiments and an ion channel recording data, with runtimes significantly outperforming batch MCMC.

Cite this Paper

BibTeX
@InProceedings{pmlr-v70-ma17a, title = {Stochastic Gradient {MCMC} Methods for Hidden {M}arkov Models}, author = {Yi-An Ma and Nicholas J. Foti and Emily B. Fox}, booktitle = {Proceedings of the 34th International Conference on Machine Learning}, pages = {2265--2274}, year = {2017}, editor = {Precup, Doina and Teh, Yee Whye}, volume = {70}, series = {Proceedings of Machine Learning Research}, month = {06--11 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v70/ma17a/ma17a.pdf}, url = {https://proceedings.mlr.press/v70/ma17a.html}, abstract = {Stochastic gradient MCMC (SG-MCMC) algorithms have proven useful in scaling Bayesian inference to large datasets under an assumption of i.i.d data. We instead develop an SG-MCMC algorithm to learn the parameters of hidden Markov models (HMMs) for time-dependent data. There are two challenges to applying SG-MCMC in this setting: The latent discrete states, and needing to break dependencies when considering minibatches. We consider a marginal likelihood representation of the HMM and propose an algorithm that harnesses the inherent memory decay of the process. We demonstrate the effectiveness of our algorithm on synthetic experiments and an ion channel recording data, with runtimes significantly outperforming batch MCMC.} }
Endnote
%0 Conference Paper %T Stochastic Gradient MCMC Methods for Hidden Markov Models %A Yi-An Ma %A Nicholas J. Foti %A Emily B. Fox %B Proceedings of the 34th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2017 %E Doina Precup %E Yee Whye Teh %F pmlr-v70-ma17a %I PMLR %P 2265--2274 %U https://proceedings.mlr.press/v70/ma17a.html %V 70 %X Stochastic gradient MCMC (SG-MCMC) algorithms have proven useful in scaling Bayesian inference to large datasets under an assumption of i.i.d data. We instead develop an SG-MCMC algorithm to learn the parameters of hidden Markov models (HMMs) for time-dependent data. There are two challenges to applying SG-MCMC in this setting: The latent discrete states, and needing to break dependencies when considering minibatches. We consider a marginal likelihood representation of the HMM and propose an algorithm that harnesses the inherent memory decay of the process. We demonstrate the effectiveness of our algorithm on synthetic experiments and an ion channel recording data, with runtimes significantly outperforming batch MCMC.
APA
Ma, Y., Foti, N.J. & Fox, E.B.. (2017). Stochastic Gradient MCMC Methods for Hidden Markov Models. Proceedings of the 34th International Conference on Machine Learning, in Proceedings of Machine Learning Research 70:2265-2274 Available from https://proceedings.mlr.press/v70/ma17a.html.

Related Material