example_perceptron_kernel_gaussian_yinyang2.cpp

Programa para el testeo de las funciones.Para compilar o código example_perceptron_kernel_gaussian_yinyang2.cpp:

g++ -static -o example_perceptron_kernel_gaussian_yinyang2 example_perceptron_kernel_gaussian_yinyang2.cpp -lpdsmlmm -lpdsramm -lpdsspmm

Para executar o programa:

./example_perceptron_kernel_gaussian_yinyang2

Retornando por consola:

Neurona:
0.1758083022248  -8.6487555419947 -8.788829119114  -7.5034303296872 -7.221256389084  -7.4351142358589 -7.1386765267093 -8.5061559836684 -8.8781505169577 -8.7493849843645 -4.9570411895064 -7.3867368563384 -7.0667960156638 6.9274895852939  6.4736966944438  7.6067216407531  6.3736354528218  7.8450425370139  7.3352561755174  7.3670027696057  6.1262235652442  4.9086447179916  7.5959828499847  7.2198125494198  7.6606725092723  

╔═══════════════════════════════════╗
║    ClassificationMetrics data     ║
╠═══════════════════════════════════╣
║        Threshold: 0.5             ║
║          Samples: 2000            ║
╠═══════════════════════════════════╣
║ Pred.[0]/Act.[*]: 952     47      ║
║ Pred.[1]/Act.[*]: 48      953     ║
╠═══════════════════════════════════╣
║         Accuracy: 95.25 %         ║
║        Precision: 95.2  %         ║
║           Recall: 95.3  %         ║
╠═══════════════════════════════════╣
║           FScore: 95.25 %         ║
╚═══════════════════════════════════╝

╔═══════════════════════════════════╗
║    ClassificationMetrics data     ║
╠═══════════════════════════════════╣
║        Threshold: 0.5             ║
║          Samples: 2000            ║
╠═══════════════════════════════════╣
║ Pred.[0]/Act.[*]: 950     64      ║
║ Pred.[1]/Act.[*]: 50      936     ║
╠═══════════════════════════════════╣
║         Accuracy: 94.3  %         ║
║        Precision: 94.93 %         ║
║           Recall: 93.6  %         ║
╠═══════════════════════════════════╣
║           FScore: 94.26 %         ║
╚═══════════════════════════════════╝

Classification - data

Classification - training

Classification - testing

Código example_perceptron_kernel_gaussian_yinyang2.cpp:

 
#include <iostream>
#include <Pds/Ra>
#include <Pds/Ml>
 
    
int main(void)
{
    Pds::ClassificationMetrics Metrics;
    Pds::Vector Yp;
    Pds::Matrix F;
    Pds::IterationConf Conf;
    Conf.SetMaxIter(10000); Conf.Show=false;
    unsigned int K=12;
    
    // Generating data
    unsigned int L=1000;
    Pds::Matrix X;
    Pds::Vector Y;
    Pds::Matrix G;
    Pds::DataSet::Classification::LoadDataYinYang(L,X,Y);
    
    Pds::Octave::XLabel="x_1";
    Pds::Octave::YLabel="x_2";
    Pds::Octave::Plot::PointsX2DY(X,Y,"testando.m","example_perceptron_kernel_gaussian_yinyang2_data.png");
    
    // Create Perceptron
    Pds::Array<unsigned int> ID0,ID1;
    std::vector<Pds::Matrix> Block=X.FindRowsBlock(Y);
    Pds::Matrix G0=Pds::Clustering::Kmeans(Conf,Block[0],K,ID0);
    Pds::Matrix G1=Pds::Clustering::Kmeans(Conf,Block[1],K,ID1);
    std::vector<double> Sigma0=Pds::Clustering::KmeansSigma(Block[0],ID0);
    std::vector<double> Sigma1=Pds::Clustering::KmeansSigma(Block[1],ID1);
    
    G=Pds::MergeVer({G0,G1});
    Sigma0.insert(Sigma0.end()  , Sigma1.begin()  , Sigma1.end());
    
    
    F=Pds::Kernel::Gaussian(X,G,Sigma0);
 
    Pds::Perceptron Neurona(Conf,F,Y);
    Neurona.Print("\nNeurona:\n");
    
    // Predict training data
    Yp=Neurona.Predict(F);
    
    // Metrics of training
    Metrics = Pds::ClassificationMetrics::Calculate(0.5,Yp,Y);
    Metrics.Print("\n");
    
    Pds::Octave::Plot::PointsX2DY(X,Yp,"testando.m","example_perceptron_kernel_gaussian_yinyang2_training.png");
        
    // Predict testing data
    Pds::DataSet::Classification::LoadDataYinYang(L,X,Y); // Load new data
    
    F=Pds::Kernel::Gaussian(X,G,Sigma0);
    Yp=Neurona.Predict(F);
    
    // Metrics testing
    Metrics = Pds::ClassificationMetrics::Calculate(Metrics.Threshold,Yp,Y);
    Metrics.Print("\n");
    
    Pds::Octave::Plot::PointsX2DY(X,Yp,"testando.m","example_perceptron_kernel_gaussian_yinyang2_testing.png");
    return 0;
}