twodlearn.templates.bayesnet module

class twodlearn.templates.bayesnet.EVGPEstimator(options=None, logger_path='tmp', session=None, **kargs)

Bases: twodlearn.templates.bayesnet.VGPEstimator

model

Autoinit with arguments [‘dataset’, ‘prior’]

train

Autoinit with arguments [‘dataset’, ‘batch_size’]

class twodlearn.templates.bayesnet.ExplicitGpEstimator(options=None, logger_path='tmp', session=None, **kargs)

Bases: twodlearn.templates.bayesnet.GpEstimator

Explicit gp estimator

fit(train_x, train_y, max_iter=100)

fit the gaussian process to the training data. By default, the parameters of the kernel are optimized.

Parameters

• train_x – input training data [batch_size, x_size].

• train_y – output training data [batch_size, y_size].

• max_iter – number of maximum iterations to run the optimization. Defaults to 100.

model

Autoinit with arguments [‘train_x’, ‘train_y’]

test

class twodlearn.templates.bayesnet.GpEstimator(options=None, logger_path='tmp', session=None, **kargs)

Bases: twodlearn.templates.supervised.SupervisedEstimator

fit(train_x, train_y, max_iter=100)

fit the gaussian process to the training data. By default, the parameters of the kernel are optimized.

Parameters

• train_x – input training data [batch_size, x_size].

• train_y – output training data [batch_size, y_size].

• max_iter – number of maximum iterations to run the optimization. Defaults to 100.

get_save_data()

model

Autoinit with arguments [‘train_x’, ‘train_y’]

normalizer

optimizer

Autoinit with arguments [‘learning_rate’, ‘valid_freq’]

predict(x)

test

train

valid

class twodlearn.templates.bayesnet.VGPEstimator(options=None, logger_path='tmp', session=None, **kargs)

Bases: twodlearn.templates.supervised.SupervisedEstimator

fit(dataset, max_iter=100, feed_dict=None)

m

model

Autoinit with arguments [‘dataset’]

normalizer

optimizer

Autoinit with arguments [‘learning_rate’, ‘valid_freq’]

predict(x)

test

Autoinit with arguments [‘dataset’, ‘tolerance’]

train

Autoinit with arguments [‘dataset’, ‘batch_size’]

valid