Python HMM.get_log_likelihood_sample Beispiele

Beispiel #1

Datei: distribution_hmm_modular.py Projekt: manantomar/test

def distribution_hmm_modular(fm_cube, N, M, pseudo, order, gap, reverse, num_examples):
	from modshogun import StringWordFeatures, StringCharFeatures, CUBE
	from modshogun import HMM, BW_NORMAL

	charfeat=StringCharFeatures(CUBE)
	charfeat.set_features(fm_cube)
	feats=StringWordFeatures(charfeat.get_alphabet())
	feats.obtain_from_char(charfeat, order-1, order, gap, reverse)

	hmm=HMM(feats, N, M, pseudo)
	hmm.train()
	hmm.baum_welch_viterbi_train(BW_NORMAL)

	num_examples=feats.get_num_vectors()
	num_param=hmm.get_num_model_parameters()
	for i in range(num_examples):
		for j in range(num_param):
			hmm.get_log_derivative(j, i)

	best_path=0
	best_path_state=0
	for i in range(num_examples):
		best_path+=hmm.best_path(i)
		for j in range(N):
			best_path_state+=hmm.get_best_path_state(i, j)

	lik_example = hmm.get_log_likelihood()
	lik_sample = hmm.get_log_likelihood_sample()

	return lik_example, lik_sample, hmm