Stochastic Variational Inference for Hidden Markov Models

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.

Appears in Advances in Neural Information Processing Systems (NIPS), 2014

References in corpus (2)

• Streaming Variational Bayes
• Parallelizing MCMC via Weierstrass Sampler

Cited by in corpus (11)

• Variational Inference: A Review for Statisticians
• Amortized Variational Inference: A Systematic Review
• Advances in Variational Inference
• Hidden Markov Nonlinear ICA: Unsupervised Learning from Nonstationary Time Series
• Automatic structured variational inference
• The Neural Moving Average Model for Scalable Variational Inference of State Space Models
• Divide-and-Conquer Bayesian Inference in Hidden Markov Models
• Automatic variational inference with cascading flows
• Targeted stochastic gradient Markov chain Monte Carlo for hidden Markov models with rare latent states
• Stochastic Collapsed Variational Inference for Sequential Data
• Scalable Hybrid HMM with Gaussian Process Emission for Sequential Time-series Data Clustering