/HMM

Julia implementation of hidden Markov models

Primary LanguageJuliaMIT LicenseMIT

HMM

Julia implementation of hidden Markov models

Features

  • Works with continuous or discrete observations
  • Customizable emission distributions
  • Pre-implemented emission distribution models:
    • Discrete (on a finite set {1,...,K})
    • Univariate normal
    • Multivariate normal with diagonal covariance

Algorithms implemented

  • Viterbi algorithm
  • Forward algorithm
  • Backward algorithm
  • Baum-Welch algorithm (expectation-maximization) for parameter estimation
  • Generate data from an HMM

Citation and licensing

If you use this software in your research, please cite:
Jeffrey W. Miller (2016). Lecture Notes on Advanced Stochastic Modeling. Duke University, Durham, NC.

This software was written as part of the "Advanced Stochastic Modeling" STA531 course at Duke University, Spring 2016.

Copyright (c) 2016 Jeffrey W. Miller. This software is released under the MIT License (see LICENSE).