#include <quasi_newton_method.h>
Public Member Functions | |
void | resize_training_history (const unsigned int &) |
std::string | to_string (void) const |
Public Attributes | |
Vector< Vector< double > > | parameters_history |
Vector< double > | parameters_norm_history |
Vector< double > | evaluation_history |
Vector< double > | generalization_evaluation_history |
Vector< Vector< double > > | gradient_history |
Vector< double > | gradient_norm_history |
Vector< Matrix< double > > | inverse_Hessian_history |
Vector< Vector< double > > | training_direction_history |
Vector< double > | training_rate_history |
Vector< double > | elapsed_time_history |
Vector< double > | final_parameters |
double | final_parameters_norm |
double | final_evaluation |
double | final_generalization_evaluation |
Vector< double > | final_gradient |
double | final_gradient_norm |
Matrix< double > | final_inverse_Hessian_approximation |
Vector< double > | final_training_direction |
double | final_training_rate |
double | elapsed_time |
Definition at line 72 of file quasi_newton_method.h.
void OpenNN::QuasiNewtonMethod::QuasiNetwonMethodResults::resize_training_history | ( | const unsigned int & | new_size | ) |
This method resizes all the training history variables.
new_size | Size of training history variables. |
Definition at line 1481 of file quasi_newton_method.cpp.
std::string OpenNN::QuasiNewtonMethod::QuasiNetwonMethodResults::to_string | ( | void | ) | const [virtual] |
This method returns a string representation of the current quasi-Newton method results structure.
Reimplemented from OpenNN::TrainingAlgorithm::Results.
Definition at line 1541 of file quasi_newton_method.cpp.
History of the neural network parameters over the training epochs.
Definition at line 78 of file quasi_newton_method.h.
History of the parameters norm over the training epochs.
Definition at line 82 of file quasi_newton_method.h.
History of the performance function evaluation over the training epochs.
Definition at line 86 of file quasi_newton_method.h.
Vector<double> OpenNN::QuasiNewtonMethod::QuasiNetwonMethodResults::generalization_evaluation_history |
History of the generalization evaluation over the training epochs.
Definition at line 90 of file quasi_newton_method.h.
History of the performance function gradient over the training epochs.
Definition at line 94 of file quasi_newton_method.h.
History of the gradient norm over the training epochs.
Definition at line 98 of file quasi_newton_method.h.
Vector< Matrix<double> > OpenNN::QuasiNewtonMethod::QuasiNetwonMethodResults::inverse_Hessian_history |
History of the inverse Hessian approximation over the training epochs.
Definition at line 102 of file quasi_newton_method.h.
Vector< Vector<double> > OpenNN::QuasiNewtonMethod::QuasiNetwonMethodResults::training_direction_history |
History of the random search training direction over the training epochs.
Definition at line 106 of file quasi_newton_method.h.
History of the random search training rate over the training epochs.
Definition at line 110 of file quasi_newton_method.h.
History of the elapsed time over the training epochs.
Definition at line 114 of file quasi_newton_method.h.