levenberg_marquardt_algorithm.h

//   OpenNN: Open Neural Networks Library
//   www.opennn.net
//
//   L E V E N B E R G - M A R Q U A R D T   A L G O R I T H M   C L A S S   H E A D E R
//
//   Artificial Intelligence Techniques SL
//   artelnics@artelnics.com
 
#ifndef LEVENBERGMARQUARDTALGORITHM_H
#define LEVENBERGMARQUARDTALGORITHM_H
 
// System includes
 
#include <string>
#include <sstream>
#include <iostream>
#include <fstream>
#include <algorithm>
#include <functional>
#include <limits>
#include <math.h>
#include <time.h>
 
// OpenNN includes
 
#include "config.h"
#include "tensor_utilities.h"
#include "optimization_algorithm.h"
 
// Eigen includes
 
#include "../eigen/Eigen/Dense"
 
namespace OpenNN
{
 
struct LevenbergMarquardtAlgorithmData;
 
 
 
 
class LevenbergMarquardtAlgorithm : public OptimizationAlgorithm
{
 
public:
 
   // Constructors
 
   explicit LevenbergMarquardtAlgorithm();
 
   explicit LevenbergMarquardtAlgorithm(LossIndex*);
 
   // Destructor
 
   virtual ~LevenbergMarquardtAlgorithm();
 
   // Get methods
 
   // Stopping criteria
 
   const type& get_minimum_loss_decrease() const;
   const type& get_loss_goal() const;
 
   const Index& get_maximum_selection_failures() const;
 
   const Index& get_maximum_epochs_number() const;
   const type& get_maximum_time() const;
 
   // Utilities
 
   const type& get_damping_parameter() const;
 
   const type& get_damping_parameter_factor() const;
 
   const type& get_minimum_damping_parameter() const;
   const type& get_maximum_damping_parameter() const;
 
   // Set methods
 
   void set_default();
 
   void set_damping_parameter(const type&);
 
   void set_damping_parameter_factor(const type&);
 
   void set_minimum_damping_parameter(const type&);
   void set_maximum_damping_parameter(const type&);
 
   // Stopping criteria
 
   void set_minimum_loss_decrease(const type&);
   void set_loss_goal(const type&);
 
   void set_maximum_selection_failures(const Index&);
 
   void set_maximum_epochs_number(const Index&);
   void set_maximum_time(const type&);
 
   // Training methods
 
   void check() const;
 
   TrainingResults perform_training();
 
   void update_parameters(
           const DataSetBatch&,
           NeuralNetworkForwardPropagation&,
           LossIndexBackPropagationLM&,
           LevenbergMarquardtAlgorithmData&);
 
   string write_optimization_algorithm_type() const;
 
   // Serialization methods
 
   Tensor<string, 2> to_string_matrix() const;
   
   void from_XML(const tinyxml2::XMLDocument&);
 
   void write_XML(tinyxml2::XMLPrinter&) const;
   
private:
 
   // MEMBERS
 
 
   type damping_parameter;
 
 
   type minimum_damping_parameter;
 
 
   type maximum_damping_parameter;
 
 
   type damping_parameter_factor;
 
   // Stopping criteria 
 
 
   type minimum_loss_decrease;
 
 
   type training_loss_goal;
 
 
   Index maximum_selection_failures;
 
 
   Index maximum_epochs_number;
 
 
   type maximum_time;
};
 
 
struct LevenbergMarquardtAlgorithmData : public OptimizationAlgorithmData
{
 
    explicit LevenbergMarquardtAlgorithmData()
    {
    }
 
    explicit LevenbergMarquardtAlgorithmData(LevenbergMarquardtAlgorithm* new_Levenberg_Marquardt_method_pointer)
    {
        set(new_Levenberg_Marquardt_method_pointer);
    }
 
    virtual ~LevenbergMarquardtAlgorithmData() {}
 
    void set(LevenbergMarquardtAlgorithm* new_Levenberg_Marquardt_method_pointer)
    {
        Levenberg_Marquardt_algorithm = new_Levenberg_Marquardt_method_pointer;
 
        LossIndex* loss_index_pointer = Levenberg_Marquardt_algorithm->get_loss_index_pointer();
 
        NeuralNetwork* neural_network_pointer = loss_index_pointer->get_neural_network_pointer();
 
        const Index parameters_number = neural_network_pointer->get_parameters_number();
 
        // Neural network data
 
        old_parameters.resize(parameters_number);
 
        parameters_difference.resize(parameters_number);
 
        potential_parameters.resize(parameters_number);
        parameters_increment.resize(parameters_number);
    }
 
    LevenbergMarquardtAlgorithm* Levenberg_Marquardt_algorithm = nullptr;
 
    // Neural network data
 
    Tensor<type, 1> old_parameters;
    Tensor<type, 1> parameters_difference;
 
    Tensor<type, 1> parameters_increment;
 
    // Loss index data
 
    type old_loss = type(0);
 
    // Optimization algorithm data
 
    Index epoch = 0;
};
 
 
}
 
#endif
 
 
// OpenNN: Open Neural Networks Library.
// Copyright(C) 2005-2021 Artificial Intelligence Techniques, SL.
//
// This library is free software; you can redistribute it and/or
// modify it under the terms of the GNU Lesser General Public
// License as published by the Free Software Foundation; either
// version 2.1 of the License, or any later version.
//
// This library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
// Lesser General Public License for more details.
 
// You should have received a copy of the GNU Lesser General Public
// License along with this library; if not, write to the Free Software
// Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA