TrainingError.h

Go to the documentation of this file.

/*!
 * 
 *
 * \brief       Stopping Criterion which stops, when the trainign error seems to converge
 * 
 * 
 *
 * \author      O. Krause
 * \date        2010
 *
 *
 * \par Copyright 1995-2017 Shark Development Team
 * 
 * <BR><HR>
 * This file is part of Shark.
 * <https://shark-ml.github.io/Shark/>
 * 
 * Shark 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 3 of the License, or
 * (at your option) any later version.
 * 
 * Shark 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 Shark.  If not, see <http://www.gnu.org/licenses/>.
 *
 */
 
#ifndef SHARK_TRAINERS_STOPPINGCRITERA_TRAININGERROR_H
#define SHARK_TRAINERS_STOPPINGCRITERA_TRAININGERROR_H
 
#include "AbstractStoppingCriterion.h"
#include <shark/Core/ResultSets.h>
#include <queue>
#include <numeric>
#include <shark/LinAlg/Base.h>
namespace shark{
 
///  \brief This stopping criterion tracks the improvement of the error function of the training error over an interval of iterations.
///
/// If at one point, the difference between the error values of the beginning and the end of the interval are smaller
/// than a certain value, this stopping criterion assumes convergence and stops.
/// Of course, this may be misleading, when the algorithm temporarily gets stuck at a saddle point of the error surface.
/// The functions assumes that the algorithm is minimizing. For details, see:
///
/// Lutz Prechelt. Early Stopping - but when? In Genevieve B. Orr and
/// Klaus-Robert Müller: Neural Networks: Tricks of the Trade, volume
/// 1524 of LNCS, Springer, 1997.
///
template<class PointType = RealVector>
class TrainingError: public AbstractStoppingCriterion< SingleObjectiveResultSet<PointType> >{
public:
    /// constructs the TrainingError generalization loss
    /// @param intervalSize size of the interval over which the progress is monitored
    /// @param minDifference minimum difference between start and end of the interval allowed before training stops
    TrainingError(size_t intervalSize, double minDifference){
        m_minDifference = minDifference;
        m_intervalSize = intervalSize;
        reset();
    }
    /// returns true if training should stop
    bool stop(const SingleObjectiveResultSet<PointType>& set){
 
        m_interval.pop();
        m_interval.push(set.value);
        return (m_interval.front()-set.value) >= 0
            && (m_interval.front()-set.value) < m_minDifference;
    }
    /// resets the internal state
    void reset(){
        m_interval = std::queue<double>();
        for(size_t i = 0; i != m_intervalSize;++i) {
            m_interval.push(std::numeric_limits<double>::max());
        }
    }
protected:
    /// monitored training interval
    std::queue<double> m_interval;
    /// minmum difference allowed
    double m_minDifference;
    /// size of the interval
    size_t m_intervalSize;
};
}
 
 
#endif