model.deeplearn.regularizer package

Submodules

model.deeplearn.regularizer.features_orthogonal_regularizer module

class model.deeplearn.regularizer.features_orthogonal_regularizer.FeaturesOrthogonalRegularizer(**kwargs)

Bases: Regularizer

FeaturesOrthogonalRegularizer regularizer to enforce orthogonality in the feature space. Taken from the Keras web documentation on 2023-07-19 https://keras.io/examples/vision/pointnet

__init__(**kwargs)

Initialize the FeaturesOrthogonalRegularizer regularizer. See regularizer.Regularizer.__init__().

__call__(x, training=False, mask=False)

The computational logic of the features orthogonal regularizer. See regularizer.Regularizer.__call__(). :param x: The input tensor.

get_config()

Obtain the dictionary specifying how to serialize the features orthogonal regularizer.

Returns:

The dictionary with the necessary data to serialize the features orthogonal regularizer.

Return type:

dict

classmethod from_config(cfg)

Deserialize a features ortoghonal regularizer from given specification.

Parameters:

cfg – The dictionary specifying how to deserialize the regularizer.

Returns:

The deserialized orthogonal regularizer.

Return type:

:class:.FeaturesOrthogonalRegularizer`

model.deeplearn.regularizer.regularizer module

class model.deeplearn.regularizer.regularizer.Regularizer(**kwargs)

Bases: Regularizer

Author:

Alberto M. Esmoris Pena

A regularizer can be seen as a map \(f\) from an input tensor \(\mathcal{X}\) to an output scalar \(y\) that can be added to the loss function.

The Regularizer class provides an interface that must be realized by any class that must assume the role of a regularizer inside a neural network.

__init__(**kwargs)

Initialize the member attributes of the regularizer.

Parameters:

kwargs – The key-word specification to parametrize the regularizer.

__call__(x)

The regularizer’s computation logic.

Parameters:

x – The input tensor.

Returns:

The output scalar.

get_config()

The dictionary specifying how to serialize the regularizer.

Returns:

The dictionary with the necessary data to serialize the regularizer.

Return type:

dict

Module contents

author:

Alberto M. Esmoris Pena

The layer package contains the logic to handle the regularizer of a neural network.