Hidden Markov models with stick-breaking priors

Authors:
John Paisley;Lawrence Carin
Affiliations:
Department of Electrical and Computer Engineering, Duke University, Durham, NC;Department of Electrical and Computer Engineering, Duke University, Durham, NC
Venue:
IEEE Transactions on Signal Processing
Year:
2009

Citing 4
Cited 1

Generalized Dirichlet distribution in Bayesian analysis

Applied Mathematics and Computation
Multiaspect Target Identification with Wave-Based Matched Pursuits and Continuous Hidden Markov Models

IEEE Transactions on Pattern Analysis and Machine Intelligence
Variational Message Passing

The Journal of Machine Learning Research
Hidden Markov models for multiaspect target classification

IEEE Transactions on Signal Processing

A reservoir-driven non-stationary hidden Markov model

Pattern Recognition

Quantified Score

Hi-index	35.68

Visualization

Abstract

The number of states in a hidden Markov model (HMM) is an Important parameter that has a critical impact on the inferred model. Bayesian approaches to addressing this issue include the nonparametric hierarchical Dirichlet process, which does not extend to a variational Bayesian (VB) solution. We present a fully conjugate, Bayesian approach to determining the number of states in a HMM, which does have a variational solution. The infinite-state HMM presented here utilizes a stick-breaking construction for each row of the state transition matrix, which allows for a sparse utilization of the same subset of observation parameters by all states. In addition to our variational solution, we discuss retrospective and collapsed Gibbs sampling methods for MCMC inference. We demonstrate our model on a music recommendation problem containing 2250 pieces of music from the classical, jazz, and rock genres.