OneVersusOne.cpp

Go to the documentation of this file.

//===========================================================================
/*!
 * 
 *
 * \brief       Example program for the one-versus-one classifier based on SVMs
 * 
 * 
 *
 * \author      T. Glasmachers
 * \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/>.
 *
 */
//===========================================================================


#include <shark/Core/Random.h>
#include <shark/Data/DataDistribution.h>
#include <shark/Models/OneVersusOneClassifier.h>
#include <shark/Models/Kernels/GaussianRbfKernel.h>
#include <shark/Algorithms/Trainers/CSvmTrainer.h>
#include <shark/ObjectiveFunctions/Loss/ZeroOneLoss.h>
#include <iostream>


using namespace shark;
using namespace std;


// data generating distribution for our toy
// multi-category classification problem
/// @cond EXAMPLE_SYMBOLS
class Problem : public LabeledDataDistribution<RealVector, unsigned int>
{
public:
    void draw(RealVector& input, unsigned int& label)const
    {
        label = random::discrete(random::globalRng, 0, 4);
        input.resize(1);
        input(0) = random::gauss(random::globalRng) + 3.0 * label;
    }
};
/// @endcond

int main()
{
    // experiment settings
    unsigned int classes = 5;
    std::size_t ell = 100;
    std::size_t tests = 10000;
    double C = 10.0;
    double gamma = 0.5;

    // generate a very simple dataset with a little noise
    Problem problem;
    ClassificationDataset training = problem.generateDataset(ell);
    ClassificationDataset test = problem.generateDataset(tests);
    repartitionByClass(training);

    // kernel function
    GaussianRbfKernel<> kernel(gamma);
    // train the OVO machine
    OneVersusOneClassifier<RealVector> ovo;
    unsigned int pairs = classes * (classes - 1) / 2;
    std::vector< KernelClassifier<RealVector> > bin_svm(pairs);
    for (std::size_t n=0, c=1; c<classes; c++)
    {
        std::vector< OneVersusOneClassifier<RealVector>::binary_classifier_type* > vs_c;
        for (std::size_t e=0; e<c; e++, n++)
        {
            //get the binary subproblem
            ClassificationDataset bindata = binarySubProblem(training,e,c);
                
            // train the binary machine
            CSvmTrainer<RealVector> trainer(&kernel, C,false);
            trainer.train(bin_svm[n], bindata);
            vs_c.push_back(&bin_svm[n]);
        }
        ovo.addClass(vs_c);
    }

    // compute errors
    ZeroOneLoss<unsigned int> loss;
    Data<unsigned int> output = ovo(training.inputs());
    double train_error = loss.eval(training.labels(), output);
    output = ovo(test.inputs());
    double test_error = loss.eval(test.labels(), output);
    cout << "training error: " << 100.0 * train_error << "%" << endl;
    cout << "    test error: " << 100.0 *  test_error << "%" << endl;
}