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
{
 
 
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();
}
 
 
 
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();
}
 
 
 
OptimizationAlgorithm::~OptimizationAlgorithm()
{
    delete non_blocking_thread_pool;
    delete thread_pool_device;
}
 
 
 
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;
}
 
 
 
string OptimizationAlgorithm::get_hardware_use() const
{
    return hardware_use;
}
 
 
 
void OptimizationAlgorithm::set_hardware_use(const string& new_hardware_use)
{
    hardware_use = new_hardware_use;
}
 
 
 
bool OptimizationAlgorithm::has_loss_index() const
{
    if(loss_index_pointer)
    {
        return true;
    }
    else
    {
        return false;
    }
}
 
 
 
const bool& OptimizationAlgorithm::get_display() const
{
    return display;
}
 
 
 
const Index& OptimizationAlgorithm::get_display_period() const
{
    return display_period;
}
 
 
 
const Index& OptimizationAlgorithm::get_save_period() const
{
    return save_period;
}
 
 
 
const string& OptimizationAlgorithm::get_neural_network_file_name() const
{
    return neural_network_file_name;
}
 
 
 
void OptimizationAlgorithm::set()
{
    loss_index_pointer = nullptr;
 
    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);
}
 
 
 
void OptimizationAlgorithm::set_loss_index_pointer(LossIndex* new_loss_index_pointer)
{
    loss_index_pointer = new_loss_index_pointer;
}
 
 
 
void OptimizationAlgorithm::set_display(const bool& new_display)
{
    display = new_display;
}
 
 
 
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;
}
 
 
 
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;
}
 
 
 
void OptimizationAlgorithm::set_neural_network_file_name(const string& new_neural_network_file_name)
{
    neural_network_file_name = new_neural_network_file_name;
}
 
 
 
void OptimizationAlgorithm::set_default()
{
    display = true;
 
    display_period = 10;
 
    save_period = UINT_MAX;
 
    neural_network_file_name = "neural_network.xml";
}
 
 
 
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
}
 
 
 
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();
}
 
 
 
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;
            }
        }
    }
}
 
 
 
Tensor<string, 2> OptimizationAlgorithm::to_string_matrix() const
{
    return Tensor<string, 2>();
}
 
 
 
void OptimizationAlgorithm::print() const
{
}
 
 
 
void OptimizationAlgorithm::save(const string& file_name) const
{
    FILE * file = fopen(file_name.c_str(), "w");
 
    tinyxml2::XMLPrinter printer(file);
 
    write_XML(printer);
 
    fclose(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);
}
 
 
 
string TrainingResults::write_stopping_condition() const
{
    switch(stopping_condition)
    {
    case OptimizationAlgorithm::StoppingCondition::MinimumLossDecrease:
        return "Minimum loss decrease";
 
    case OptimizationAlgorithm::StoppingCondition::LossGoal:
        return "Loss goal";
 
    case OptimizationAlgorithm::StoppingCondition::MaximumSelectionErrorIncreases:
        return "Maximum selection error increases";
 
    case OptimizationAlgorithm::StoppingCondition::MaximumEpochsNumber:
        return "Maximum number of epochs";
 
    case OptimizationAlgorithm::StoppingCondition::MaximumTime:
        return "Maximum training time";
    }
 
    return string();
}
 
 
 
void TrainingResults::resize_training_error_history(const Index& new_size)
{
    if(training_error_history.size() == 0)
    {
        training_error_history.resize(new_size);
 
        return;
    }
 
    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);
    }
}
 
 
 
void TrainingResults::resize_selection_error_history(const Index& new_size)
{
    if(selection_error_history.size() == 0)
    {
        selection_error_history.resize(new_size);
 
        return;
    }
 
    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);
    }
}
 
 
 
const string OptimizationAlgorithm::write_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_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_time(const type& time) const method.\n"
               << "Time must be greater than 0.\n";
 
        throw logic_error(buffer.str());
    }
#endif
 
    const int hours = static_cast<int>(time) / 3600;
    int seconds = static_cast<int>(time) % 3600;
    const 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();
}
 
 
 
void TrainingResults::save(const string&) const
{
 
}
 
 
Tensor<string, 2> TrainingResults::write_final_results(const Index& precision) const
{
    ostringstream buffer;
 
    Tensor<string, 2> final_results(6, 2);
 
    final_results(0,0) = "Training error";
    final_results(1,0) = "Selection error";
    final_results(2,0) = "Epochs number";
    final_results(3,0) = "Elapsed time";
    final_results(4,0) = "Stopping criterion";
 
    const Index size = training_error_history.size();
 
    if(size == 0)
    {
        final_results(0,1) = "NA";
        final_results(1,1) = "NA";
        final_results(2,1) = "NA";
        final_results(3,1) = "NA";
        final_results(4,1) = "NA";
 
        return final_results;
    }
 
    // Final training error
 
    buffer.str("");
    buffer << setprecision(precision) << training_error_history(size-1);
 
    final_results(0,1) = buffer.str();
 
    // Final selection error
 
    buffer.str("");
 
    selection_error_history.size() == 0
            ? buffer << "NAN"
            : buffer << setprecision(precision) << selection_error_history(size-1);
 
    final_results(1,1) = buffer.str();
 
    // Epochs number
 
    buffer.str("");
    buffer << training_error_history.size()-1;
 
    final_results(2,1) = buffer.str();
 
    // Elapsed time
 
    buffer.str("");
    buffer << setprecision(precision) << elapsed_time;
 
    final_results(3,1) = buffer.str();
 
    // Stopping criteria
 
    final_results(4,1) = write_stopping_condition();
 
    return final_results;
}
 
}
 
 
// 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