Online Learning in Discrete Hidden Markov Models

TL;DR

This paper introduces and compares three online algorithms for learning in discrete Hidden Markov Models, analyzing their performance and generalization error, especially in drifting concept scenarios, with insights on learning dynamics.

Contribution

The paper presents three novel online algorithms for discrete HMMs and compares their effectiveness with the existing Baldi-Chauvin algorithm using KL divergence.

Findings

One of the algorithms effectively learns drifting concepts.

Performance varies depending on the algorithm and scenario.

Discussion on learning dynamics and symmetry breaking is included.

Abstract

We present and analyse three online algorithms for learning in discrete Hidden Markov Models (HMMs) and compare them with the Baldi-Chauvin Algorithm. Using the Kullback-Leibler divergence as a measure of generalisation error we draw learning curves in simplified situations. The performance for learning drifting concepts of one of the presented algorithms is analysed and compared with the Baldi-Chauvin algorithm in the same situations. A brief discussion about learning and symmetry breaking based on our results is also presented.