traingen.h

Go to the documentation of this file.

/*

Copyright (c) 2003, Cornell University
All rights reserved.

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

   - Redistributions of source code must retain the above copyright notice,
       this list of conditions and the following disclaimer.
   - Redistributions in binary form must reproduce the above copyright
       notice, this list of conditions and the following disclaimer in the
       documentation and/or other materials provided with the distribution.
   - Neither the name of Cornell University nor the names of its
       contributors may be used to endorse or promote products derived from
       this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
THE POSSIBILITY OF SUCH DAMAGE.

*/

// -*- C++ -*-

#ifndef _TRAINGEN_H
#define _TRAINGEN_H

/* This class is the interface for the training data repositories
   Each such an object stores data as a matrix and is able to normalize
   the data. Two parameters per dimmension can be used for normalization.
*/

#include "cmat.h"
#include "vec.h"
#include "normaliz.h"
#include <iostream>
#include <fstream>
#include "general.h"
#include "extravec.h"

using namespace TNT;
using namespace std;

namespace CLUS
{

class TrainingData
{

protected:
    bool normalized;
    Matrix<double> Table;
    
public:
    TrainingData(Subscript M, Subscript N): Table(M,N)
    {
        normalized=false;
    }
    
    void Normalize(Vector<Scale>&, Scale::NormType, int);
    
    virtual void SaveToFile(char* name)
    {
        ofstream ostr(name);
        SaveToStream(ostr);
    }
    
    virtual void SaveToStream(ostream& ostr)
    {
        // @todo set format for ostream
        for(int i=0; i<Table.num_rows(); i++)
        {
            double *row=Table[i];
            for(int j=0; j<Table.num_cols(); j++)
                ostr << row[j] << "\t";
            ostr << endl;
        }
    }
    
    virtual int NumRows(void)
    {
        return Table.num_rows();
    }
    
    virtual int NumCols(void)
    {
        return Table.num_cols();
    }
    
    virtual const Matrix<double>& GetTrainingData(void)
    {
        return Table;
    }

    virtual ~TrainingData()
    {
        // will call Matrix destructor
    }
};

inline void TrainingData::Normalize(Vector<Scale>& scale, Scale::NormType normtype, int nrInputs)
{
    int N=Table.num_cols();
    ExitIf(scale.size()<N,"Wrong scaling vector in normalization");
    Vector<double> Aux1(N,Table[0]), Aux2(N,Table[0]);
    Subscript i=0, j;
    double* data;

    switch(normtype)
    {
    case Scale::Interval:
        for (i=1; i<Table.num_rows(); i++)
        {
            // for each line
            Min(Aux1, N, Table[i]);
            Max(Aux2, N, Table[i]);
        }
        
        for (j=0;j<nrInputs;j++)
            scale[j].SetCoefForInput(Aux1[j],Aux2[j],normtype);
            
        for (j=nrInputs;j<N;j++)
        {
            // Aux1[j]*=/*sqrt*/(1.0*nrInputs/(N-nrInputs));
            // Aux2[j]*=/*sqrt*/(1.0*nrInputs/(N-nrInputs));
            scale[j].SetCoefForOutput(Aux1[j],Aux2[j],normtype);
        }

        break;
    case Scale::Distribution:
        // we have to square the elements of Aux2
        for (j=0;j<N;j++)
            Aux2[i]=pow2(Aux2[i]);
            
        for (i=1; i<Table.num_rows(); i++)
        {
            // for each line
            Sum(Aux1, N, Table[i]);
            SumPow2(Aux2, N, Table[i]);
        }
        
        for (j=0;j<nrInputs;j++)
            scale[j].SetCoefForInput(Aux1[j],Aux2[j],normtype);
            
        for (j=nrInputs;j<N;j++)
            scale[j].SetCoefForOutput(Aux1[j],Aux2[j],normtype);

        break;
    }

    // transform the data
    for (i=0; i<Table.num_rows(); i++)
    {
        data=Table[i];
        
        for (j=0;j<nrInputs;j++)
            data[j]=scale[j].Transform(data[j]);
            
        for (j=nrInputs; j<N; j++)
            data[j]=scale[j].InverseTransform(data[j]);
    }
}

}


#endif // _TRAINGEN_H

---