test_working_newhb_der.cpp

 
#include <iostream>
#include <Pds/Ra>
#include <Pds/Ml>
#include <Pds/Sp>
 
 
#include "extra_newhb.h"
 
bool get_W1_W2_IS_for_given_W_X_Y(  double b,unsigned int N,
                                    Pds::Vector W,Pds::Matrix X,Pds::Vector Y,
                                    Pds::Matrix &W1,Pds::Matrix &W2,Pds::Matrix &IS)
{
    double val;
    Pds::Vector w1=Pds::LinSpace(-b,b,N);
    Pds::Vector w2=Pds::LinSpace(-b,b,N);
    
    Pds::Meshgrid(w1,w2,W1,W2);
    IS.Reshape(N,N);
    
    for(unsigned int i=0;i<N;i++)
    for(unsigned int j=0;j<N;j++)
    {
        W.Set(1,w1[i]);
        W.Set(2,w2[j]);
        
        val=Pds::LogisticModel::CostXqEntropy( W,X,Y,q_factor_val); //w*(100/w.Norm())
        IS.SetRaw(i,j,val);
    }
    
    return true;
}
 
 
// ./test/test_working_newhb_der
int main(void)
{
    Pds::Ra::Randomize();
    
    Pds::Octave::DAspect=true;
    Pds::Octave::YLimits=true;
    
    // Generating data
    // Data 1
    Pds::Matrix X(Pds::Ra::TextFormat,"../test/hard-data-x.txt");
    Pds::Vector Y(Pds::Ra::TextFormat,"../test/hard-data-y.txt");
    
    // Data 2
    //unsigned int L=1000;    Pds::Matrix X;  Pds::Vector Y;
    //Pds::DataSet::Classification::LoadDataBand(L,X,Y);
    
    
    //double d=Pds::DataSet::InterDistance(X);
    
    
    // Seleciono punto de inicio
    Pds::Vector W=Pds::LogisticModel::GetW0MeanMethod(X);
    W.Normalize();
    //W=W/W.Max();
    Pds::Octave::Plot::PointsX2DYW(X,Y,W,"testando_w.m","test_working_steps_0.png");
    
    Pds::Matrix W1,W2,IS;
    unsigned int N=64; double b=2.0;
    
    
    Pds::Vector P(W.Nel());
    Pds::Vector DW;
    double E1;
    for(unsigned int n=1;n<=60;n++)
    {
        double h=W.RMS()*0.5;
        std::cout<<"\n n: "<<n<<"\n";
        std::cout<<" h: "<<h<<"\n";
        
 
        if(Pds::Hb(Pds::LogisticModel::Classify(W,X).Geq(0.5).Mean())<=0.0)
        {
            W=Pds::LogisticModel::GetW0MeanMethod(X);
            W.Normalize();
        }
        
        DW=Pds::LogisticModel::GradientCostXqEntropy(W,X,Y,q_factor_val,h);
        //DW.Normalize();
        //DW=DW*W.Norm();
        
        get_W1_W2_IS_for_given_W_X_Y(   b,N,W,X,Y,   W1,W2,IS);
        
        DW.T().Print("DW: ");
        W=W-0.05*DW;
        W.T().Print(" W: ");
        
        //W.Normalize();
        //W.T().Print(" W: ");
        
        E1=Pds::LogisticModel::CostXqEntropy(W,X,Y,q_factor_val);
        Pds::Octave::Title="NHb:"+std::to_string(E1);
        Pds::Octave::Plot::PointsX2DYW(X,Y,W,"testando_w.m","test_working_steps_"+std::to_string(n)+".png");
        
        P.SetRaw(0,W[1]);
        P.SetRaw(1,W[2]);
        P.SetRaw(2,1);
        Pds::Octave::Plot::SurfCWithPoint(  W1,W2,IS,P,
                                            "testando_is"+std::to_string(n)+".m",
                                            "is_matrix_"+std::to_string(n)+".png"); 
    }
 
    return 0;
}