OpenNN::ModelSelection Class Reference

#include <model_selection.h>

Classes
struct	ModelSelectionResults

Public Types
enum	InputsSelectionMethod { MaximumLinearCorrelation, MaximumLogisticCorrelation, Exhaustive }

Public Member Functions
	ModelSelection (void)
	ModelSelection (TrainingStrategy *)
	ModelSelection (const std::string &)
	ModelSelection (const tinyxml2::XMLDocument &)
virtual	~ModelSelection (void)
TrainingStrategy *	get_training_strategy_pointer (void) const
const Vector< size_t > &	get_hidden_perceptrons_numbers (void) const
const size_t &	get_parameters_assays_number (void) const
const bool &	get_reserve_parameters_data (void) const
const bool &	get_reserve_performance_data (void) const
const bool &	get_reserve_generalization_performance_data (void) const
const bool &	get_reserve_minimal_parameters (void) const
const bool &	get_reserve_performance_data_statistics (void) const
const bool &	get_reserve_generalization_performance_data_statistics (void) const
const bool &	get_reserve_model_order_selection_plot (void) const
const InputsSelectionMethod &	get_inputs_selection_method (void) const
const bool &	get_display (void) const
void	set_training_strategy_pointer (TrainingStrategy *)
void	set_default (void)
void	set_hidden_perceptrons_numbers (const Vector< size_t > &)
void	set_parameters_assays_number (const size_t &)
void	set_assays_numbers (const size_t &, const size_t &)
void	set_reserve_parameters_data (const bool &)
void	set_reserve_performance_data (const bool &)
void	set_reserve_generalization_performance_data (const bool &)
void	set_reserve_minimal_parameters (const bool &)
void	set_reserve_performance_data_statistics (const bool &)
void	set_reserve_generalization_performance_data_statistics (const bool &)
void	set_reserve_model_order_selection_plot (const bool &)
void	set_inputs_selection_method (const InputsSelectionMethod &)
void	set_display (const bool &)
Matrix< double >	calculate_linear_correlations (void) const
Matrix< double >	calculate_logistic_correlations (void) const
void	check (void) const
void	perform_maximum_linear_correlation_inputs_selection (void) const
void	perform_maximum_logistic_correlation_inputs_selection (void) const
void	perform_exhaustive_inputs_selection (void) const
void	perform_inputs_selection (void) const
ModelSelectionResults	perform_order_selection (void) const
ModelSelectionResults	perform_model_selection (void) const
tinyxml2::XMLDocument *	to_XML (void) const
void	from_XML (const tinyxml2::XMLDocument &)
void	print (void) const
void	save (const std::string &) const
void	load (const std::string &)

Private Attributes
TrainingStrategy *	training_strategy_pointer
Vector< Vector< size_t > >	inputs_indices
Vector< size_t >	hidden_perceptrons_numbers
size_t	parameters_assays_number
bool	reserve_parameters_data
bool	reserve_performance_data
bool	reserve_generalization_performance_data
bool	reserve_minimal_parameters
bool	reserve_performance_data_statistics
bool	reserve_generalization_performance_data_statistics
bool	reserve_model_order_selection_plot
double	correlation_goal
InputsSelectionMethod	inputs_selection_method
bool	display

Detailed Description

This class represents the concept of model selection algorithm. It is used for finding a network architecture with maximum generalization capabilities.

Definition at line 45 of file model_selection.h.

Constructor & Destructor Documentation

OpenNN::ModelSelection::ModelSelection ( TrainingStrategy *  new_training_strategy_pointer )

explicit

Training strategy constructor.

Parameters

new_training_strategy_pointer

Pointer to a gradient descent object.

Definition at line 37 of file model_selection.cpp.

OpenNN::ModelSelection::ModelSelection ( const std::string &  file_name )

explicit

File constructor.

Parameters

file_name

Name of XML model selection file.

Definition at line 49 of file model_selection.cpp.

OpenNN::ModelSelection::ModelSelection ( const tinyxml2::XMLDocument &  model_selection_document )

explicit

XML constructor.

Parameters

model_selection_document

Pointer to a TinyXML document containing the model selection data.

Definition at line 61 of file model_selection.cpp.

Member Function Documentation

Matrix< double > OpenNN::ModelSelection::calculate_linear_correlations

(

void

)

const

Returns a matrix with the linear correlations between all input and target variables. The number of rows is the number of input variables. The number of columns is the number of target variables.

Definition at line 529 of file model_selection.cpp.

Matrix< double > OpenNN::ModelSelection::calculate_logistic_correlations

(

void

)

const

Returns a matrix with the logistic correlations between all input and target variables. The number of rows is the number of input variables. The number of columns is the number of target variables.

Definition at line 578 of file model_selection.cpp.

void OpenNN::ModelSelection::from_XML

(

const tinyxml2::XMLDocument &

)

Todo:

Definition at line 1112 of file model_selection.cpp.

const bool & OpenNN::ModelSelection::get_display

(

void

)

const

Returns true if messages from this class can be displayed on the screen, or false if messages from this class can't be displayed on the screen.

Definition at line 209 of file model_selection.cpp.

void OpenNN::ModelSelection::load

(

const std::string &

file_name

)

Loads the model selection members from a XML file.

Parameters

file_name

Name of model selection XML file.

Definition at line 1147 of file model_selection.cpp.

void OpenNN::ModelSelection::perform_exhaustive_inputs_selection

(

void

)

const

Todo:

This method is not implemented

Definition at line 691 of file model_selection.cpp.

void OpenNN::ModelSelection::perform_inputs_selection

(

void

)

const

Todo:

Definition at line 815 of file model_selection.cpp.

void OpenNN::ModelSelection::perform_maximum_linear_correlation_inputs_selection

(

void

)

const

Todo:

This method is not implemented

Definition at line 667 of file model_selection.cpp.

void OpenNN::ModelSelection::perform_maximum_logistic_correlation_inputs_selection

(

void

)

const

Todo:

This method is not implemented

Definition at line 679 of file model_selection.cpp.

ModelSelection::ModelSelectionResults OpenNN::ModelSelection::perform_model_selection

(

void

)

const

Todo:

Definition at line 1049 of file model_selection.cpp.

ModelSelection::ModelSelectionResults OpenNN::ModelSelection::perform_order_selection

(

void

)

const

Todo:

Definition at line 856 of file model_selection.cpp.

void OpenNN::ModelSelection::save

(

const std::string &

file_name

)

const

Saves the model selection members to a XML file.

Parameters

file_name

Name of model selection XML file.

Definition at line 1132 of file model_selection.cpp.

void OpenNN::ModelSelection::set_assays_numbers	(	const size_t &	new_complexity_assays_number,
		const size_t &	new_parameters_assays_number
	)

Sets the numbers of complexities and assays.

Parameters

new_complexity_assays_number	Number of hidden neurons.
new_parameters_assays_number	Number of trainings for each different neural network.

Definition at line 268 of file model_selection.cpp.

void OpenNN::ModelSelection::set_default

(

void

)

Todo:

Definition at line 230 of file model_selection.cpp.

void OpenNN::ModelSelection::set_display

(

const bool &

new_display

)

Sets a new display value. If it is set to true messages from this class are to be displayed on the screen; if it is set to false messages from this class are not to be displayed on the screen.

Parameters

new_display

Display value.

Definition at line 432 of file model_selection.cpp.

void OpenNN::ModelSelection::set_hidden_perceptrons_numbers

(

const Vector< size_t > &

new_hidden_perceptrons_numbers

)

Sets the number of complexities to be compared in the model order selection process.

Parameters

new_hidden_perceptrons_numbers

Vector with different hidden layers sizes.

Definition at line 245 of file model_selection.cpp.

void OpenNN::ModelSelection::set_inputs_selection_method

(

const InputsSelectionMethod &

new_inputs_selection_method

)

Sets a new method for selecting the inputs which have more impact on the targets.

Parameters

new_inputs_selection_method

Method for selecting the inputs (MaximumLinearCorrelation, MaximumLogisticCorrelation or Exhaustive).

Definition at line 419 of file model_selection.cpp.

void OpenNN::ModelSelection::set_parameters_assays_number

(

const size_t &

new_parameters_assays_number

)

Sets the number of times that each different neural network is to be trained.

Parameters

new_parameters_assays_number

Number of assays for each set of parameters.

Definition at line 256 of file model_selection.cpp.

void OpenNN::ModelSelection::set_reserve_generalization_performance_data

(

const bool &

new_reserve_generalization_performance_data

)

Sets the reserve flag for the generalization performance data.

Parameters

new_reserve_generalization_performance_data

Flag value.

Definition at line 364 of file model_selection.cpp.

void OpenNN::ModelSelection::set_reserve_generalization_performance_data_statistics

(

const bool &

new_reserve_generalization_performance_data_statistics

)

Sets the reserve flag for the generalization performance data statistics.

Parameters

new_reserve_generalization_performance_data_statistics

Flag value.

Definition at line 397 of file model_selection.cpp.

void OpenNN::ModelSelection::set_reserve_minimal_parameters

(

const bool &

new_reserve_minimal_parameters

)

Sets the reserve flag for the minimal parameters.

Parameters

new_reserve_minimal_parameters

Flag value.

Definition at line 375 of file model_selection.cpp.

void OpenNN::ModelSelection::set_reserve_model_order_selection_plot

(

const bool &

new_reserve_model_order_selection_plot

)

Sets the reserve flag for the model order selection plot.

Parameters

new_reserve_model_order_selection_plot

Flag value.

Definition at line 408 of file model_selection.cpp.

void OpenNN::ModelSelection::set_reserve_parameters_data

(

const bool &

new_reserve_parameters_data

)

Sets the reserve flag for the parameters data.

Parameters

new_reserve_parameters_data

Flag value.

Definition at line 342 of file model_selection.cpp.

void OpenNN::ModelSelection::set_reserve_performance_data

(

const bool &

new_reserve_performance_data

)

Sets the reserve flag for the performance data.

Parameters

new_reserve_performance_data

Flag value.

Definition at line 353 of file model_selection.cpp.

void OpenNN::ModelSelection::set_reserve_performance_data_statistics

(

const bool &

new_reserve_performance_data_statistics

)

Sets the reserve flag for the performance data statistics.

Parameters

new_reserve_performance_data_statistics

Flag value.

Definition at line 386 of file model_selection.cpp.

void OpenNN::ModelSelection::set_training_strategy_pointer

(

TrainingStrategy *

new_training_strategy_pointer

)

Sets a new training strategy pointer.

Parameters

new_training_strategy_pointer

Pointer to a training strategy object.

Definition at line 220 of file model_selection.cpp.

tinyxml2::XMLDocument * OpenNN::ModelSelection::to_XML

(

void

)

const

Serializes the model selection object into a XML document of the TinyXML library. See the OpenNN manual for more information about the format of this document.

Definition at line 1070 of file model_selection.cpp.

---

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

• model_selection.h
• model_selection.cpp