Citation
Nick Foti, Jason Xu, Dillon Laird, Emily B. Fox. "Stochastic Variational Inference for Hidden Markov Models". Neural Information Processing Systems (NIPS), 8, December, 2014.

Abstract
Variational inference algorithms have proven successful for Bayesian analysis in large data settings, with recent advances using stochastic variational inference (SVI). However, such methods have largely been studied in independent or exchangeable data settings. We develop an SVI algorithm to learn the parameters of hidden Markov models (HMMs) in a time-dependent data setting. The challenge in applying stochastic optimization in this setting arises from dependencies in the chain, which must be broken to consider minibatches of observations. We propose an algorithm that harnesses the memory decay of the chain to adaptively bound errors arising from edge effects. We demonstrate the effectiveness of our algorithm on synthetic experiments and a large genomics dataset where a batch algorithm is computationally infeasible.

Electronic downloads

• https://www.src.org/library/publication/p072549/. (Link to publication on SRC website.)
• http://arxiv.org/pdf/1411.1670v1.pdf

• HTML

  Nick Foti, Jason Xu, Dillon Laird, Emily B. Fox. <a
  href="http://www.terraswarm.org/pubs/423.html"
  >Stochastic Variational Inference for Hidden Markov
  Models</a>, Neural Information Processing Systems
  (NIPS), 8, December, 2014.

• Plain text

  Nick Foti, Jason Xu, Dillon Laird, Emily B. Fox.
  "Stochastic Variational Inference for Hidden Markov
  Models". Neural Information Processing Systems (NIPS),
  8, December, 2014.

• BibTeX

  @inproceedings{FotiXuLairdFox14_StochasticVariationalInferenceForHiddenMarkovModels,
      author = {Nick Foti and Jason Xu and Dillon Laird and Emily
                B. Fox},
      title = {Stochastic Variational Inference for Hidden Markov
                Models},
      booktitle = {Neural Information Processing Systems (NIPS)},
      day = {8},
      month = {December},
      year = {2014},
      abstract = {Variational inference algorithms have proven
                successful for Bayesian analysis in large data
                settings, with recent advances using stochastic
                variational inference (SVI). However, such methods
                have largely been studied in independent or
                exchangeable data settings. We develop an SVI
                algorithm to learn the parameters of hidden Markov
                models (HMMs) in a time-dependent data setting.
                The challenge in applying stochastic optimization
                in this setting arises from dependencies in the
                chain, which must be broken to consider
                minibatches of observations. We propose an
                algorithm that harnesses the memory decay of the
                chain to adaptively bound errors arising from edge
                effects. We demonstrate the effectiveness of our
                algorithm on synthetic experiments and a large
                genomics dataset where a batch algorithm is
                computationally infeasible.},
      URL = {http://terraswarm.org/pubs/423.html}
  }
  

Posted by Barb Hoversten on 3 Nov 2014.
Groups: services