DistantModes.h

Go to the documentation of this file.

/*!
 * \brief       Implements the DistantModes/ArtificialModes benchmark problem
 * 
 * \author      O. Krause, A.Fischer, K.Bruegge
 * \date        2012
 *
 *
 * \par Copyright 1995-2017 Shark Development Team
 * 
 * <BR><HR>
 * This file is part of Shark.
 * <http://shark-ml.org/>
 * 
 * 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 UNSUPERVISED_RBM_PROBLEMS_DISTANTMODES_H
#define UNSUPERVISED_RBM_PROBLEMS_DISTANTMODES_H

#include  <shark/Data/Dataset.h>
#include  <shark/LinAlg/Base.h>
#include <shark/Core/Random.h>
#include <shark/Models/AbstractModel.h>

namespace shark{


///\brief Creates a set of pattern (each later representing a mode)
/// which than are randomly perturbed to create the data set.
/// The dataset was introduced in Desjardins et al. (2010) (Parallel Tempering for training restricted Boltzmann machines, AISTATS 2010) 
///
///The higher the perturbation is the harder it is to classify, 
///but the closer are the modes and thus the easier the data distribution is to learn.
class DistantModes{
private:
    UnlabeledData<RealVector> m_data;
    
    double m_p;
    unsigned m_dim;
    unsigned m_modes;
    unsigned m_copies;
    std::size_t m_batchSize;
    
    //Generates a basic pattern representing the "center" of a mode.
    void modePrototype(RealVector& pattern, unsigned mode) const {
        for (std::size_t i = 0; i != pattern.size(); ++i){
            pattern(i) = (mode % 2) ^ (i * (mode / 2 + 1) / pattern.size()) % 2;
        }
    }


    ///Perturbates the pattern by randomly flipping pixels
    ///@param pattern the pattern
    ///@param p the flipping probability
    void perturbate(RealVector& pattern, double p)const{
        for (std::size_t i = 0; i < pattern.size(); ++i){
            if (random::uni(random::globalRng, 0,1) > p){
                pattern(i) = !pattern(i);
            }
        }
    }
    
    void init() {
        std::vector<RealVector> data(m_modes * m_copies,RealVector(m_dim));
        for (std::size_t i = 0; i != data.size(); ++i) {
            RealVector& element=data[i];
            unsigned mode = i % m_modes;
            modePrototype(element, mode);
            perturbate(element, m_p);
        }
        m_data = createDataFromRange(data, m_batchSize);
    }
    
public:
    ///generates the DistantModes distribution.
    ///
    ///\param p the probability of changing a input neuron
    ///\param dim the dimensionality of the data.
    ///\param modes the number of modes, should be a multiple of 2
    ///\param copies the number of disturbed copies for each mode
    ///\param batchSize the size of the batches in which the generated data set is organized
    DistantModes(double p = 0, unsigned dim = 16, unsigned modes=4, unsigned copies =2500, size_t batchSize=0)
        :m_p(p), m_dim(dim), m_modes(modes), m_copies(copies), m_batchSize(batchSize) {
        init();
    }
    
    ///returns the generated dataset
    UnlabeledData<RealVector> data() const{
        return m_data;
    };
    
    ///returns the dimensionality of the data
    std::size_t inputDimension() const {
        return m_dim;
    }
};

}
#endif