Awesome

Informational Mixture of Hidden Markov Models

The iMHMM implementation is based on the standard EM for MHMM from this paper:

Pernes, D.; and Cardoso, J. S. Spamhmm: Sparse mixture of hidden markov models for graph connected entities. 2019 International Joint Conference on Neural Networks(IJCNN) pp. 1–10 (2019)

and also oMHMM from this paper:

Safinianaini, N.; de Souza, C.; Bostr̈om, H.; and Lagergren,J. Orthogonal mixture of hidden markov models (ECML-PKDD) (2020)

The softwares needed to run the experiments:

Python 3.6.2 hmmlearn 0.2.1 cvxpy 1.0.21 numpy 1.16.2 scikit-learn 0.19.1 scipy 1.1.0

The computing infrastructure which we use:

OS: OS X

Processor: 2,8 GHz Intel Core i7

Memory: 16 GB 2133 MHz LPDDR3

Graphics: Radeon Pro 560 4 GB Intel HD Graphics 630 1536 MB

Note: To see the regularization introduced by iMHMM, see line 59 of the file "imhmm.py" in the test folder.