Daniel@0: O = 3;
Daniel@0: Q = 2;
Daniel@0: 
Daniel@0: % "true" parameters
Daniel@0: prior0 = normalise(rand(Q,1));
Daniel@0: transmat0 = mk_stochastic(rand(Q,Q));
Daniel@0: obsmat0 = mk_stochastic(rand(Q,O));
Daniel@0: 
Daniel@0: % training data
Daniel@0: T = 5;
Daniel@0: nex = 10;
Daniel@0: data = dhmm_sample(prior0, transmat0, obsmat0, T, nex);
Daniel@0: 
Daniel@0: % initial guess of parameters
Daniel@0: prior1 = normalise(rand(Q,1));
Daniel@0: transmat1 = mk_stochastic(rand(Q,Q));
Daniel@0: obsmat1 = mk_stochastic(rand(Q,O));
Daniel@0: 
Daniel@0: % improve guess of parameters using EM
Daniel@0: [LL, prior2, transmat2, obsmat2] = dhmm_em(data, prior1, transmat1, obsmat1, 'max_iter', 5);
Daniel@0: LL
Daniel@0: 
Daniel@0: % use model to compute log likelihood
Daniel@0: loglik = dhmm_logprob(data, prior2, transmat2, obsmat2)
Daniel@0: % log lik is slightly different than LL(end), since it is computed after the final M step