tlda.ThirdOrderCumulant

class tlda.ThirdOrderCumulant(n_topic, alpha_0, n_iter_train, n_iter_test, batch_size, learning_rate, gamma_shape=1.0, theta=1, ortho_loss_criterion=1000, seed=None, n_eigenvec=None, learning_rate_criterion=1e-05)[source]

Class to compute the third order cumulant

Methods

fit(X[, verbose])	Update the factors directly from X using stochastic gradient descent
partial_fit(X_batch[, learning_rate])	Update the factors directly from the batch using stochastic gradient descent
predict(X_test, adjusted_factors, weights)	Infer the document/topic distribution from the factors and weights and make the factor non-negative

partial_fit(X_batch, learning_rate=None)[source]

Update the factors directly from the batch using stochastic gradient descent

Parameters:

X_batchndarray of shape (number_documents, num_topics) equal to the whitened

word counts in each document in the documents used to update the factors

verbosebool, optional

if True, print information about every 200th iteration

fit(X, verbose=True)[source]

Update the factors directly from X using stochastic gradient descent

Parameters:

Xndarray of shape (number_documents, num_topics) equal to the whitened

word counts in each document in the documents used to update the factors

predict(X_test, adjusted_factors, weights)[source]

Infer the document/topic distribution from the factors and weights and make the factor non-negative

Parameters:

X_testndarray of shape (number_documents, vocabulary_size) equal to the word

counts in each test document

Returns:

gammadtensor of shape (number_documents, number_topics) equal to

the normalized document/topic distribution for X_test