nemos.regularizer.Regularizer#

class nemos.regularizer.Regularizer(**kwargs)[source]#

Bases: Base, ABC

Abstract base class for regularized solvers.

This class is designed to provide a consistent interface for optimization solvers, enabling users to easily switch between different regularizers, ensuring compatibility with various loss functions and optimization algorithms.

allowed_solvers#

Tuple of solver names that are allowed for use with this regularizer.

default_solver#

String of the default solver name allowed for use with this regularizer.

Attributes

allowed_solvers

default_solver

__init__(**kwargs)[source]#

Methods

__init__(**kwargs)

get_params([deep])

From scikit-learn, get parameters by inspecting init.

get_proximal_operator()

Abstract method to retrieve the proximal operator for this solver.

penalized_loss(loss, regularizer_strength)

Abstract method to penalize loss functions.

set_params(**params)

Set the parameters of this estimator.
property allowed_solvers: Tuple[str]#
property default_solver: str#
get_params(deep=True)#

From scikit-learn, get parameters by inspecting init.

Parameters:

deep

Return type:

dict

Returns:

out:

A dictionary containing the parameters. Key is the parameter name, value is the parameter value.

abstract get_proximal_operator()[source]#

Abstract method to retrieve the proximal operator for this solver.

Return type:

Callable[[Any, float, float], Tuple[Array, Array]]

Returns:

The proximal operator, which typically applies a form of regularization.

abstract penalized_loss(loss, regularizer_strength)[source]#

Abstract method to penalize loss functions.

Parameters:

  • loss (Callable) – Callable loss function.

  • regularizer_strength (float) – Float the indicates the regularization strength.

Return type:

Callable

Returns:

A modified version of the loss function including any relevant penalization based on the regularizer type.

set_params(**params)#

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:

**params (dict) – Estimator parameters.

Returns:

self – Estimator instance.

Return type:

estimator instance