xref: /dports/misc/opennn/opennn-5.0.5/opennn/optimization_algorithm.cpp

//   OpenNN: Open Neural Networks Library
//   www.opennn.net
//
//   O P T I M I Z A T I O N   A L G O R I T H M   C L A S S
//
//   Artificial Intelligence Techniques SL
//   artelnics@artelnics.com

#include "optimization_algorithm.h"

namespace OpenNN
{

/// Default constructor.
/// It creates a optimization algorithm object not associated to any loss index object.

OptimizationAlgorithm::OptimizationAlgorithm()
    : loss_index_pointer(nullptr)
{
    const int n = omp_get_max_threads();
    non_blocking_thread_pool = new NonBlockingThreadPool(n);
    thread_pool_device = new ThreadPoolDevice(non_blocking_thread_pool, n);

    set_default();
}


/// It creates a optimization algorithm object associated to a loss index object.
/// @param new_loss_index_pointer Pointer to a loss index object.

OptimizationAlgorithm::OptimizationAlgorithm(LossIndex* new_loss_index_pointer)
    : loss_index_pointer(new_loss_index_pointer)
{
    const int n = omp_get_max_threads();
    non_blocking_thread_pool = new NonBlockingThreadPool(n);
    thread_pool_device = new ThreadPoolDevice(non_blocking_thread_pool, n);

    set_default();
}


/// Destructor.

OptimizationAlgorithm::~OptimizationAlgorithm()
{
    delete non_blocking_thread_pool;
    delete thread_pool_device;
}


/// Returns a pointer to the loss index object to which the optimization algorithm is
/// associated.

LossIndex* OptimizationAlgorithm::get_loss_index_pointer() const
{
#ifdef __OPENNN_DEBUG__

    if(!loss_index_pointer)
    {
        ostringstream buffer;

        buffer << "OpenNN Exception: OptimizationAlgorithm class.\n"
               << "LossIndex* get_loss_index_pointer() const method.\n"
               << "Loss index pointer is nullptr.\n";

        throw logic_error(buffer.str());
    }

#endif

    return loss_index_pointer;
}


/// Returns the hardware used. Default: Multi-core

string OptimizationAlgorithm::get_hardware_use() const
{
    return hardware_use;
}


/// Set hardware to use. Default: Multi-core.

void OptimizationAlgorithm::set_hardware_use(const string& new_hardware_use)
{
    hardware_use = new_hardware_use;
}

/// Returns true if this optimization algorithm object has an associated loss index object,
/// and false otherwise.

bool OptimizationAlgorithm::has_loss_index() const
{
    if(loss_index_pointer)
    {
        return true;
    }
    else
    {
        return false;
    }
}


/// 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.

const bool& OptimizationAlgorithm::get_display() const
{
    return display;
}


/// Returns the number of iterations between the training showing progress.

const Index& OptimizationAlgorithm::get_display_period() const
{
    return display_period;
}


/// Returns the number of iterations between the training saving progress.

const Index& OptimizationAlgorithm::get_save_period() const
{
    return save_period;
}


/// Returns the file name where the neural network will be saved.

const string& OptimizationAlgorithm::get_neural_network_file_name() const
{
    return neural_network_file_name;
}


/// Sets the loss index pointer to nullptr.
/// It also sets the rest of members to their default values.

void OptimizationAlgorithm::set()
{
    loss_index_pointer = nullptr;

    set_default();
}


/// Sets a new loss index pointer.
/// It also sets the rest of members to their default values.
/// @param new_loss_index_pointer Pointer to a loss index object.

void OptimizationAlgorithm::set(LossIndex* new_loss_index_pointer)
{
    loss_index_pointer = new_loss_index_pointer;

    set_default();
}


void OptimizationAlgorithm::set_threads_number(const int& new_threads_number)
{
    if(non_blocking_thread_pool != nullptr) delete this->non_blocking_thread_pool;
    if(thread_pool_device != nullptr) delete this->thread_pool_device;

    non_blocking_thread_pool = new NonBlockingThreadPool(new_threads_number);
    thread_pool_device = new ThreadPoolDevice(non_blocking_thread_pool, new_threads_number);
}


/// Sets a pointer to a loss index object to be associated to the optimization algorithm.
/// @param new_loss_index_pointer Pointer to a loss index object.

void OptimizationAlgorithm::set_loss_index_pointer(LossIndex* new_loss_index_pointer)
{
    loss_index_pointer = new_loss_index_pointer;
}


/// 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.
/// @param new_display Display value.

void OptimizationAlgorithm::set_display(const bool& new_display)
{
    display = new_display;
}


/// Sets a new number of iterations between the training showing progress.
/// @param new_display_period
/// Number of iterations between the training showing progress.

void OptimizationAlgorithm::set_display_period(const Index& new_display_period)
{


#ifdef __OPENNN_DEBUG__

    if(new_display_period <= 0)
    {
        ostringstream buffer;

        buffer << "OpenNN Exception: ConjugateGradient class.\n"
               << "void set_display_period(const Index&) method.\n"
               << "Display period must be greater than 0.\n";

        throw logic_error(buffer.str());
    }

#endif

    display_period = new_display_period;
}


/// Sets a new number of iterations between the training saving progress.
/// @param new_save_period
/// Number of iterations between the training saving progress.

void OptimizationAlgorithm::set_save_period(const Index& new_save_period)
{


#ifdef __OPENNN_DEBUG__

    if(new_save_period <= 0)
    {
        ostringstream buffer;

        buffer << "OpenNN Exception: ConjugateGradient class.\n"
               << "void set_save_period(const Index&) method.\n"
               << "Save period must be greater than 0.\n";

        throw logic_error(buffer.str());
    }

#endif

    save_period = new_save_period;
}


/// Sets a new file name where the neural network will be saved.
/// @param new_neural_network_file_name
/// File name for the neural network object.

void OptimizationAlgorithm::set_neural_network_file_name(const string& new_neural_network_file_name)
{
    neural_network_file_name = new_neural_network_file_name;
}


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

void OptimizationAlgorithm::set_default()
{
    display = true;

    display_period = 5;

    save_period = UINT_MAX;

    neural_network_file_name = "neural_network.xml";
}


/// Performs a default checking for optimization algorithms.
/// In particular, it checks that the loss index pointer associated to the optimization algorithm is not nullptr,
/// and that the neural network associated to that loss index is neither nullptr.
/// If that checkings are not hold, an exception is thrown.

void OptimizationAlgorithm::check() const
{
#ifdef __OPENNN_DEBUG__

    ostringstream buffer;

    if(!loss_index_pointer)
    {
        buffer << "OpenNN Exception: OptimizationAlgorithm class.\n"
               << "void check() const method.\n"
               << "Pointer to loss index is nullptr.\n";

        throw logic_error(buffer.str());
    }

    const NeuralNetwork* neural_network_pointer = loss_index_pointer->get_neural_network_pointer();

    if(neural_network_pointer == nullptr)
    {
        buffer << "OpenNN Exception: OptimizationAlgorithm class.\n"
               << "void check() const method.\n"
               << "Pointer to neural network is nullptr.\n";

        throw logic_error(buffer.str());
    }

#endif
}


/// Serializes the optimization algorithm 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.

void OptimizationAlgorithm::write_XML(tinyxml2::XMLPrinter& file_stream) const
{
    ostringstream buffer;

    file_stream.OpenElement("OptimizationAlgorithm");

    // Display

    file_stream.OpenElement("Display");

    buffer.str("");
    buffer << display;

    file_stream.PushText(buffer.str().c_str());

    file_stream.CloseElement();


    file_stream.CloseElement();
}


/// Loads a default optimization algorithm from a XML document.
/// @param document TinyXML document containing the error term members.

void OptimizationAlgorithm::from_XML(const tinyxml2::XMLDocument& document)
{
    const tinyxml2::XMLElement* root_element = document.FirstChildElement("OptimizationAlgorithm");

    if(!root_element)
    {
        ostringstream buffer;

        buffer << "OpenNN Exception: OptimizationAlgorithm class.\n"
               << "void from_XML(const tinyxml2::XMLDocument&) method.\n"
               << "Optimization algorithm element is nullptr.\n";

        throw logic_error(buffer.str());
    }

    // Display
    {
        const tinyxml2::XMLElement* display_element = root_element->FirstChildElement("Display");

        if(display_element)
        {
            const string new_display_string = display_element->GetText();

            try
            {
                set_display(new_display_string != "0");
            }
            catch(const logic_error& e)
            {
                cerr << e.what() << endl;
            }
        }
    }
}


/// Returns a default(empty) string matrix containing the members
/// of the optimization algorithm object.

Tensor<string, 2> OptimizationAlgorithm::to_string_matrix() const
{
    Tensor<string, 2> string_matrix;

    return string_matrix;
}


/// Prints to the screen the XML-type representation of the optimization algorithm object.

void OptimizationAlgorithm::print() const
{


}


/// Saves to a XML-type file the members of the optimization algorithm object.
/// @param file_name Name of optimization algorithm XML-type file.

void OptimizationAlgorithm::save(const string& file_name) const
{
//    tinyxml2::XMLDocument* document = to_XML();

//    document->SaveFile(file_name.c_str());

//    delete document;
}


/// Loads a gradient descent object from a XML-type file.
/// Please mind about the file format, wich is specified in the User's Guide.
/// @param file_name Name of optimization algorithm XML-type file.

void OptimizationAlgorithm::load(const string& file_name)
{
    set_default();

    tinyxml2::XMLDocument document;

    if(document.LoadFile(file_name.c_str()))
    {
        ostringstream buffer;

        buffer << "OpenNN Exception: OptimizationAlgorithm class.\n"
               << "void load(const string&) method.\n"
               << "Cannot load XML file " << file_name << ".\n";

        throw logic_error(buffer.str());
    }

    from_XML(document);
}


/// Return a string with the stopping condition of the Results

string OptimizationAlgorithm::Results::write_stopping_condition() const
{
    switch(stopping_condition)
    {
    case MinimumParametersIncrementNorm:
        return "Minimum parameters increment norm";

    case MinimumLossDecrease:
        return "Minimum loss decrease";

    case LossGoal:
        return "Loss goal";

    case GradientNormGoal:
        return "Gradient norm goal";

    case MaximumSelectionErrorIncreases:
        return "Maximum selection error increases";

    case MaximumEpochsNumber:
        return "Maximum number of epochs";

    case MaximumTime:
        return "Maximum training time";
    }

    return string();
}


/// Resizes all the training history variables.
/// @param new_size Size of training history variables.

void OptimizationAlgorithm::Results::resize_training_history(const Index& new_size)
{
    training_error_history.resize(new_size);
}


/// Resizes all the selection history variables.
/// @param new_size Size of selection history variables.

void OptimizationAlgorithm::Results::resize_selection_history(const Index& new_size)
{
    selection_error_history.resize(new_size);
}


/// Resizes the training error history keeping the values.
/// @param new_size Size of training history variables.

void OptimizationAlgorithm::Results::resize_training_error_history(const Index& new_size)
{
    const Tensor<type, 1> old_training_error_history = training_error_history;

    training_error_history.resize(new_size);

    for(Index i = 0; i < new_size; i++)
    {
        training_error_history(i) = old_training_error_history(i);
    }
}


/// Resizes the training error history keeping the values.
/// @param new_size Size of training history variables.

void OptimizationAlgorithm::Results::resize_selection_error_history(const Index& new_size)
{
    const Tensor<type, 1> old_selection_error_history = selection_error_history;

    selection_error_history.resize(new_size);

    for(Index i = 0; i < new_size; i++)
    {
        selection_error_history(i) = old_selection_error_history(i);
    }
}


/// Writes the time from seconds in format HH:mm:ss.

const string OptimizationAlgorithm::write_elapsed_time(const type& time) const
{

#ifdef __OPENNN_DEBUG__

    if(time > static_cast<type>(3600e5))
    {
        ostringstream buffer;

        buffer << "OpenNN Exception: OptimizationAlgorithm class.\n"
               << "const string write_elapsed_time(const type& time) const method.\n"
               << "Time must be lower than 10e5 seconds.\n";

        throw logic_error(buffer.str());
    }

    if(time < static_cast<type>(0))
    {
        ostringstream buffer;

        buffer << "OpenNN Exception: OptimizationAlgorithm class.\n"
               << "const string write_elapsed_time(const type& time) const method.\n"
               << "Time must be greater than 0.\n";

        throw logic_error(buffer.str());
    }
#endif

    int hours = static_cast<int>(time) / 3600;
    int seconds = static_cast<int>(time) % 3600;
    int minutes = seconds / 60;
    seconds = seconds % 60;

    ostringstream elapsed_time;

    elapsed_time << setfill('0') << setw(2) << hours << ":"
                 << setfill('0') << setw(2) << minutes << ":"
                 << setfill('0') << setw(2) << seconds;

    return elapsed_time.str();
}


/// @todo

void OptimizationAlgorithm::Results::save(const string&) const
{

}



Tensor<string, 2> OptimizationAlgorithm::Results::write_final_results(const Index& precision) const
{
    ostringstream buffer;

    Tensor<string, 2> final_results(7, 2);

    // Final parameters norm

    final_results(0,0) = "Final parameters norm";

    buffer.str("");
    buffer << setprecision(precision) << final_parameters_norm;

    final_results(0,1) = buffer.str();

    // Final loss

    final_results(1,0) = "Final training error";

    buffer.str("");
    buffer << setprecision(precision) << final_training_error;

    final_results(1,1) = buffer.str();

    // Final selection error

    final_results(2,0) = "Final selection error";

    buffer.str("");
    buffer << setprecision(precision) << final_selection_error;

    final_results(2,1) = buffer.str();

    // Final gradient norm

    final_results(3,0) = "Final gradient norm";

    buffer.str("");
    buffer << setprecision(precision) << final_gradient_norm;

    final_results(3,1) = buffer.str();

    // Final learning rate

    //   names.push_back("Final learning rate");

    //   buffer.str("");
    //   buffer << setprecision(precision) << final_learning_rate;

    //   values.push_back(buffer.str());

    // Epochs number

    final_results(4,0) = "Epochs number";

    buffer.str("");
    buffer << epochs_number;

    final_results(4,1) = buffer.str();

    // Elapsed time

    final_results(5,0) = "Elapsed time";

    buffer.str("");
    buffer << setprecision(precision) << elapsed_time;

    final_results(5,1) = buffer.str();

    // Stopping criteria

    final_results(6,0) = "Stopping criterion";

    final_results(6,1) = write_stopping_condition();

    return final_results;
}

}


// OpenNN: Open Neural Networks Library.
// Copyright(C) 2005-2020 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