Stochastic Collapsed Variational Inference for Hidden Markov Models

TL;DR

This paper introduces a scalable, memory-efficient stochastic collapsed variational inference algorithm for hidden Markov models that improves accuracy over uncollapsed methods in large sequential datasets.

Contribution

It presents a novel stochastic collapsed variational inference method for HMMs that handles long sequences by breaking them into subchains and updating posteriors with a new sum-product algorithm.

Findings

Algorithm is scalable to large datasets

It is more accurate than uncollapsed algorithms

Uses less memory than existing methods

Abstract

Stochastic variational inference for collapsed models has recently been successfully applied to large scale topic modelling. In this paper, we propose a stochastic collapsed variational inference algorithm for hidden Markov models, in a sequential data setting. Given a collapsed hidden Markov Model, we break its long Markov chain into a set of short subchains. We propose a novel sum-product algorithm to update the posteriors of the subchains, taking into account their boundary transitions due to the sequential dependencies. Our experiments on two discrete datasets show that our collapsed algorithm is scalable to very large datasets, memory efficient and significantly more accurate than the existing uncollapsed algorithm.