onyx.am.gmm_trainer – Training Gaussian mixture models.

class onyx.am.gmm_trainer.GmmAccumSet(num_comps, dimension, covariance_type)

Bases: object

Internal class used for accumulation for GMM training - you shouldn’t be making these yourself.

>>> g0 = GmmAccumSet(10, 33, gaussian.GaussianModelBase.DIAGONAL_COVARIANCE)

accum_sequence(seq, comp_scores, gammas=None)

clear()

estimate_gmm_probs()

Compute weights, means, and vars from accum set.

merge(other)

Merge another accum set into this one

class onyx.am.gmm_trainer.GmmTrainer(gmm, mode='train', relevances=(0.0, 0.0, 0.0))

Bases: onyx.am.gmm_trainer.Trainer

>>> m0 = gaussian.GaussianMixtureModel(3, gaussian.GaussianModelBase.DIAGONAL_COVARIANCE, 2)
>>> m0.set_weights(np.array((0.75, 0.25)))
>>> mu = np.array(((1, 1, 1), (3, 3, 3)))
>>> m0.set_means(mu)
>>> v = np.array(((1, 1, 1), (1, 1, 1)))
>>> m0.set_vars(v)
>>> m0.seed(0)
>>> data0 = np.array([m0.sample() for _ in xrange(200)], dtype=float).transpose()
>>> m1 = gaussian.GaussianMixtureModel(3, gaussian.GaussianModelBase.DIAGONAL_COVARIANCE, 2)
>>> m1.init_models()
>>> print m1
Gmm: (Type = diagonal, Dim = 3, NumComps = 2)
Weights   Models
 0.5000       Means: -0.3997 0.1183 0.9441    Vars: 0.7000 0.7000 0.7000
 0.5000       Means: -1.0006 1.5618 -0.7573    Vars: 0.7000 0.7000 0.7000
>>> gmm_tr1 = GmmTrainer(m1)
>>> gmm_tr1.train(data0)
>>> print m1
Gmm: (Type = diagonal, Dim = 3, NumComps = 2)
Weights   Models
 0.4641       Means: 1.9168 2.2232 2.2034    Vars: 2.0687 1.7396 1.1752
 0.5359       Means: 0.7015 0.8667 0.8024    Vars: 0.8496 0.7484 0.7699

>>> m2 = gaussian.GaussianMixtureModel(3, gaussian.GaussianModelBase.FULL_COVARIANCE, 2)
>>> m2.init_models()
>>> print m2
Gmm: (Type = full, Dim = 3, NumComps = 2)
Weights   Models
 0.5000       Means: 1.4930 0.0120 0.1124    Vars:
 0.7000 0.0000 0.0000
 0.0000 0.7000 0.0000
 0.0000 0.0000 0.7000
 0.5000       Means: -0.1175 -0.2294 -0.6498    Vars:
 0.7000 0.0000 0.0000
 0.0000 0.7000 0.0000
 0.0000 0.0000 0.7000
>>> data1 = np.array([m0.sample() for _ in xrange(800)], dtype=float).transpose()
>>> gmm_tr2 = GmmTrainer(m2)
>>> gmm_tr2.train(data1)
>>> print m2
Gmm: (Type = full, Dim = 3, NumComps = 2)
Weights   Models
 0.6864       Means: 1.8985 1.7657 1.8274    Vars:
 1.8164 0.9724 0.8152
 0.9724 2.0158 0.9202
 0.8152 0.9202 1.7895
 0.3136       Means: 0.5914 1.0580 0.8863    Vars:
 0.5956 -0.2138 -0.0857
 -0.2138 0.7284 -0.1301
 -0.0857 -0.1301 0.7710

train(data, max_iters=10, ll_change_threshold=None)

train_gaussian(data)

train_gmm(data, max_iters, ll_change_threshold)

update_gmm(est_weights, est_means, est_vars)

class onyx.am.gmm_trainer.Trainer

Bases: object