Density Ratio Hidden Markov Models

TL;DR

This paper introduces Density Ratio Hidden Markov Models, which leverage density ratio estimation to improve discriminative performance in sequential classification tasks while maintaining probabilistic interpretability.

Contribution

It proposes a novel reformulation of HMM inference and training using density ratios, enhancing discriminative power and data efficiency.

Findings

Significant increase in classification accuracy compared to traditional HMMs

More efficient use of training data than existing methods

Retains probabilistic interpretability of models

Abstract

Hidden Markov models and their variants are the predominant sequential classification method in such domains as speech recognition, bioinformatics and natural language processing. Being generative rather than discriminative models, however, their classification performance is a drawback. In this paper we apply ideas from the field of density ratio estimation to bypass the difficult step of learning likelihood functions in HMMs. By reformulating inference and model fitting in terms of density ratios and applying a fast kernel-based estimation method, we show that it is possible to obtain a striking increase in discriminative performance while retaining the probabilistic qualities of the HMM. We demonstrate experimentally that this formulation makes more efficient use of training data than alternative approaches.