I am trying to train a Hidden Markov Chain model with different Mixuture Gaussian emission distribution for different states. What I want is the number of mixtures of gaussian for each state is different. For example, for state 1, the emission distribution has 3 mixed gaussian, for state 2, the emission distribution has 6 mixed gaussion.
However, I have tried the hmm.learn package and use GMMHMM class. class hmmlearn.hmm.GMMHMM(n_components=1, n_mix=1, min_covar=0.001, startprob_prior=1.0, transmat_prior=1.0, weights_prior=1.0, means_prior=0.0, means_weight=0.0, covars_prior=None, covars_weight=None, algorithm='viterbi', covariance_type='diag', random_state=None, n_iter=10, tol=0.01, verbose=False, params='stmcw', init_params='stmcw', implementation='log') It seems the n_mix here is set for all states. If I set n_mix = 3, after training, I got 3 mixtures gaussian for state 1 and 3 mixtures gaussian for state 2.
But what I am expecting is 3 mixtures gaussian for state 1, and 5 mixtures gaussian for state 2. Any hints for any pakcage/tools that can do this? python or matlab language is the best.
