This concrete class represents the stochastic gradient descent optimization algorithm[1] for a loss index of a neural network. More...

#include <stochastic_gradient_descent.h>

Inheritance diagram for StochasticGradientDescent:

Public Member Functions
	StochasticGradientDescent ()

	StochasticGradientDescent (LossIndex *)

virtual	~StochasticGradientDescent ()
	Destructor. More...

const type &	get_initial_learning_rate () const
	Returns the initial learning rate. More...

const type &	get_initial_decay () const
	Returns the initial decay. More...

const type &	get_momentum () const
	Returns the momentum. More...

const bool &	get_nesterov () const
	Returns true if nesterov is active, and false otherwise. More...

const type &	get_loss_goal () const

const type &	get_maximum_time () const
	Returns the maximum training time. More...

void	set_loss_index_pointer (LossIndex *)

void	set_default ()
	Sets the members of the optimization algorithm object to their default values. More...

void	set_batch_samples_number (const Index &new_batch_samples_number)

Index	get_batch_samples_number () const

void	set_initial_learning_rate (const type &)

void	set_initial_decay (const type &)

void	set_momentum (const type &)

void	set_nesterov (const bool &)

void	set_maximum_epochs_number (const Index &)

void	set_loss_goal (const type &)

void	set_maximum_time (const type &)

void	update_parameters (LossIndexBackPropagation &back_propagation, StochasticGradientDescentData &optimization_data)
	Set hardware to use. Default: Multi-core. More...

TrainingResults	perform_training ()

string	write_optimization_algorithm_type () const

Tensor< string, 2 >	to_string_matrix () const
	Writes as matrix of strings the most representative atributes. More...

void	from_XML (const tinyxml2::XMLDocument &)

void	write_XML (tinyxml2::XMLPrinter &) const

Public Member Functions inherited from OptimizationAlgorithm
	OptimizationAlgorithm ()

	OptimizationAlgorithm (LossIndex *)

virtual	~OptimizationAlgorithm ()
	Destructor. More...

LossIndex *	get_loss_index_pointer () const

string	get_hardware_use () const
	Hardware use. More...

void	set_hardware_use (const string &)
	Set hardware to use. Default: Multi-core. More...

bool	has_loss_index () const

const bool &	get_display () const

const Index &	get_display_period () const
	Returns the number of iterations between the training showing progress. More...

const Index &	get_save_period () const
	Returns the number of iterations between the training saving progress. More...

const string &	get_neural_network_file_name () const
	Returns the file name where the neural network will be saved. More...

const string	write_time (const type &) const
	Writes the time from seconds in format HH:mm:ss. More...

void	set ()

virtual void	set_threads_number (const int &)

virtual void	set_display (const bool &)

void	set_display_period (const Index &)

void	set_save_period (const Index &)

void	set_neural_network_file_name (const string &)

virtual void	check () const

virtual void	print () const
	Prints to the screen the XML-type representation of the optimization algorithm object. More...

void	save (const string &) const

void	load (const string &)

Private Attributes
type	initial_learning_rate
	Initial learning rate. More...

type	initial_decay
	Learning rate decay over each update. More...

type	momentum
	Parameter that accelerates SGD in the relevant direction and dampens oscillations. More...

bool	nesterov
	Boolean. Whether to apply Nesterov momentum. More...

Index	batch_samples_number = 1000
	Number of samples per training batch. More...

type	training_loss_goal = type(0)
	Goal value for the loss. It is used as a stopping criterion. More...

Index	maximum_selection_failures = numeric_limits<Index>::max()
	Maximum selection error allowed. More...

Index	maximum_epochs_number = 10000
	Maximum epochs number. More...

type	maximum_time = type(3600)
	Maximum training time. It is used as a stopping criterion. More...

Additional Inherited Members
Public Types inherited from OptimizationAlgorithm
enum class	StoppingCondition { MinimumLossDecrease , LossGoal , MaximumSelectionErrorIncreases , MaximumEpochsNumber , MaximumTime }
	Enumeration of all possibles condition of stop for the algorithms. More...

Protected Attributes inherited from OptimizationAlgorithm
NonBlockingThreadPool *	non_blocking_thread_pool = nullptr

ThreadPoolDevice *	thread_pool_device

LossIndex *	loss_index_pointer = nullptr
	Pointer to a loss index for a neural network object. More...

Index	epochs_number = 10000
	Number of training epochs in the neural network. More...

string	hardware_use = "Multi-core"
	Hardware use. More...

Index	display_period = 10
	Number of iterations between the training showing progress. More...

Index	save_period = numeric_limits<Index>::max()
	Number of iterations between the training saving progress. More...

string	neural_network_file_name = "neural_network.xml"
	Path where the neural network is saved. More...

bool	display = true
	Display messages to screen. More...

const Eigen::array< IndexPair< Index >, 1 >	AT_B = {IndexPair<Index>(0, 0)}

const Eigen::array< IndexPair< Index >, 1 >	product_vector_matrix = {IndexPair<Index>(0, 1)}

const Eigen::array< IndexPair< Index >, 1 >	A_B = {IndexPair<Index>(1, 0)}

Detailed Description

This concrete class represents the stochastic gradient descent optimization algorithm[1] for a loss index of a neural network.

It supports momentum, learning rate decay, and Nesterov momentum.

[1] Neural Designer "5 Algorithms to Train a Neural Network." https://www.neuraldesigner.com/blog/5_algorithms_to_train_a_neural_network

Definition at line 43 of file stochastic_gradient_descent.h.

Constructor & Destructor Documentation

◆ StochasticGradientDescent() [1/2]

StochasticGradientDescent ( )

explicit

Default constructor. It creates a stochastic gradient descent optimization algorithm not associated to any loss index object. It also initializes the class members to their default values.

Definition at line 18 of file stochastic_gradient_descent.cpp.

◆ StochasticGradientDescent() [2/2]

StochasticGradientDescent ( LossIndex * new_loss_index_pointer )

explicit

Loss index constructor. It creates a stochastic gradient descent optimization algorithm associated to a loss index. It also initializes the class members to their default values.

Parameters

new_loss_index_pointer Pointer to a loss index object.

Definition at line 30 of file stochastic_gradient_descent.cpp.

◆ ~StochasticGradientDescent()

~StochasticGradientDescent ( )

virtual

Destructor.

Definition at line 39 of file stochastic_gradient_descent.cpp.

Member Function Documentation

◆ from_XML()

void from_XML ( const tinyxml2::XMLDocument & document )

virtual

Loads a default optimization algorithm from a XML document.

Parameters

document TinyXML document containing the error term members.

Reimplemented from OptimizationAlgorithm.

Definition at line 723 of file stochastic_gradient_descent.cpp.

◆ get_batch_samples_number()

Index get_batch_samples_number ( ) const

Definition at line 124 of file stochastic_gradient_descent.cpp.

◆ get_initial_decay()

const type & get_initial_decay ( ) const

Returns the initial decay.

Definition at line 54 of file stochastic_gradient_descent.cpp.

◆ get_initial_learning_rate()

const type & get_initial_learning_rate ( ) const

Returns the initial learning rate.

Definition at line 46 of file stochastic_gradient_descent.cpp.

◆ get_loss_goal()

const type & get_loss_goal ( ) const

Returns the goal value for the loss. This is used as a stopping criterion when training a neural network

Definition at line 79 of file stochastic_gradient_descent.cpp.

◆ get_maximum_time()

const type & get_maximum_time ( ) const

Returns the maximum training time.

Definition at line 87 of file stochastic_gradient_descent.cpp.

◆ get_momentum()

const type & get_momentum ( ) const

Returns the momentum.

Definition at line 62 of file stochastic_gradient_descent.cpp.

◆ get_nesterov()

const bool & get_nesterov ( ) const

Returns true if nesterov is active, and false otherwise.

Definition at line 70 of file stochastic_gradient_descent.cpp.

◆ perform_training()

TrainingResults perform_training ( )

virtual

Trains a neural network with an associated loss index, according to the stochastic gradient descent method. Training occurs according to the training parameters and stopping criteria. It returns a results structure with the history and the final values of the reserved variables.

Implements OptimizationAlgorithm.

Definition at line 328 of file stochastic_gradient_descent.cpp.

◆ set_batch_samples_number()

void set_batch_samples_number ( const Index & new_batch_samples_number )

inline

Definition at line 76 of file stochastic_gradient_descent.h.

◆ set_default()

void set_default ( )

virtual

Sets the members of the optimization algorithm object to their default values.

Reimplemented from OptimizationAlgorithm.

Definition at line 103 of file stochastic_gradient_descent.cpp.

◆ set_initial_decay()

void set_initial_decay ( const type & new_dacay )

Set the initial value for the decay.

Parameters

new_initial_learning_rate initial value for the decay.

Definition at line 160 of file stochastic_gradient_descent.cpp.

◆ set_initial_learning_rate()

void set_initial_learning_rate ( const type & new_learning_rate )

Set the initial value for the learning rate. If dacay is not active learning rate will be constant otherwise learning rate will decay over each update.

Parameters

new_initial_learning_rate initial learning rate value.

Definition at line 134 of file stochastic_gradient_descent.cpp.

◆ set_loss_goal()

void set_loss_goal ( const type & new_loss_goal )

Sets a new goal value for the loss. This is used as a stopping criterion when training a neural network

Parameters

new_loss_goal Goal value for the loss.

Definition at line 249 of file stochastic_gradient_descent.cpp.

◆ set_loss_index_pointer()

void set_loss_index_pointer ( LossIndex * new_loss_index_pointer )

virtual

Sets a pointer to a loss index object to be associated to the gradient descent object. It also sets that loss index to the learning rate algorithm.

Parameters

new_loss_index_pointer Pointer to a loss index object.

Reimplemented from OptimizationAlgorithm.

Definition at line 97 of file stochastic_gradient_descent.cpp.

◆ set_maximum_epochs_number()

void set_maximum_epochs_number ( const Index & new_maximum_epochs_number )

Set the a new maximum for the epochs number.

Parameters

new_maximum_epochs number New maximum epochs number.

Definition at line 222 of file stochastic_gradient_descent.cpp.

◆ set_maximum_time()

void set_maximum_time ( const type & new_maximum_time )

Sets a new maximum training time.

Parameters

new_maximum_time Maximum training time.

Definition at line 258 of file stochastic_gradient_descent.cpp.

◆ set_momentum()

void set_momentum ( const type & new_momentum )

Set a new value for momentum, this parameter accelerates SGD in the relevant direction and dampens oscillations.

Parameters

new_momentum initial value for the mometum.

Definition at line 187 of file stochastic_gradient_descent.cpp.

◆ set_nesterov()

void set_nesterov ( const bool & new_nesterov_momentum )

Set nesterov, boolean. Whether to apply Nesterov momentum.

Parameters

new_momentum initial value for the mometum.

Definition at line 213 of file stochastic_gradient_descent.cpp.

◆ to_string_matrix()

Tensor< string, 2 > to_string_matrix ( ) const

virtual

Writes as matrix of strings the most representative atributes.

Reimplemented from OptimizationAlgorithm.

Definition at line 599 of file stochastic_gradient_descent.cpp.

◆ update_parameters()

void update_parameters	(	LossIndexBackPropagation &	back_propagation,
		StochasticGradientDescentData &	optimization_data
	)

Set hardware to use. Default: Multi-core.

Definition at line 283 of file stochastic_gradient_descent.cpp.

◆ write_optimization_algorithm_type()

string write_optimization_algorithm_type ( ) const

virtual

Reimplemented from OptimizationAlgorithm.

Definition at line 591 of file stochastic_gradient_descent.cpp.

◆ write_XML()

void write_XML ( tinyxml2::XMLPrinter & file_stream ) const

virtual

Serializes the gradient descent object into a XML document of the TinyXML library without keep the DOM tree in memory. See the OpenNN manual for more information about the format of this document.

Reimplemented from OptimizationAlgorithm.

Definition at line 645 of file stochastic_gradient_descent.cpp.

Member Data Documentation

◆ batch_samples_number

Index batch_samples_number = 1000

private

Number of samples per training batch.

Definition at line 138 of file stochastic_gradient_descent.h.

◆ initial_decay

type initial_decay

private

Learning rate decay over each update.

Definition at line 126 of file stochastic_gradient_descent.h.

◆ initial_learning_rate

type initial_learning_rate

private

Initial learning rate.

Definition at line 122 of file stochastic_gradient_descent.h.

◆ maximum_epochs_number

Index maximum_epochs_number = 10000

private

Maximum epochs number.

Definition at line 152 of file stochastic_gradient_descent.h.

◆ maximum_selection_failures

Index maximum_selection_failures = numeric_limits<Index>::max()

private

Maximum selection error allowed.

Definition at line 148 of file stochastic_gradient_descent.h.

◆ maximum_time

type maximum_time = type(3600)

private

Maximum training time. It is used as a stopping criterion.

Definition at line 156 of file stochastic_gradient_descent.h.

◆ momentum

type momentum

private

Parameter that accelerates SGD in the relevant direction and dampens oscillations.

Definition at line 130 of file stochastic_gradient_descent.h.

◆ nesterov

bool nesterov

private

Boolean. Whether to apply Nesterov momentum.

Definition at line 134 of file stochastic_gradient_descent.h.

◆ training_loss_goal

type training_loss_goal = type(0)

private

Goal value for the loss. It is used as a stopping criterion.

Definition at line 144 of file stochastic_gradient_descent.h.

The documentation for this class was generated from the following files:

Public Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ StochasticGradientDescent() [1/2]

◆ StochasticGradientDescent() [2/2]

◆ ~StochasticGradientDescent()

Member Function Documentation

◆ from_XML()

◆ get_batch_samples_number()

◆ get_initial_decay()

◆ get_initial_learning_rate()

◆ get_loss_goal()

◆ get_maximum_time()

◆ get_momentum()

◆ get_nesterov()

◆ perform_training()

◆ set_batch_samples_number()

◆ set_default()

◆ set_initial_decay()

◆ set_initial_learning_rate()

◆ set_loss_goal()

◆ set_loss_index_pointer()

◆ set_maximum_epochs_number()

◆ set_maximum_time()

◆ set_momentum()

◆ set_nesterov()

◆ to_string_matrix()

◆ update_parameters()

◆ write_optimization_algorithm_type()

◆ write_XML()

Member Data Documentation

◆ batch_samples_number

◆ initial_decay

◆ initial_learning_rate

◆ maximum_epochs_number

◆ maximum_selection_failures

◆ maximum_time

◆ momentum

◆ nesterov

◆ training_loss_goal