/***********************************************************************
This file is part of KEEL-software, the Data Mining tool for regression,
classification, clustering, pattern mining and so on.
Copyright (C) 2004-2010
F. Herrera (herrera@decsai.ugr.es)
L. S�nchez (luciano@uniovi.es)
J. Alcal�-Fdez (jalcala@decsai.ugr.es)
S. Garc�a (sglopez@ujaen.es)
A. Fern�ndez (alberto.fernandez@ujaen.es)
J. Luengo (julianlm@decsai.ugr.es)
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program 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 General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see http://www.gnu.org/licenses/
**********************************************************************/
//
// SVBPS.java
//
// Salvador Garc�a L�pez
//
// Created by Salvador Garc�a L�pez 16-7-2004.
// Copyright (c) 2004 __MyCompanyName__. All rights reserved.
//
package keel.Algorithms.Preprocess.Instance_Selection.SVBPS;
import keel.Algorithms.Preprocess.Basic.*;
import keel.Dataset.*;
import org.core.*;
import java.util.StringTokenizer;
import java.util.Vector;
import java.util.Arrays;
import org.libsvm.*;
public class SVBPS extends Metodo {
/*Own parameters of the algorithm*/
private int k;
private String kernelType;
private double C;
private double eps;
private int degree;
private double gamma;
private double nu;
private double p;
private int shrinking;
public SVBPS (String ficheroScript) {
super (ficheroScript);
}
public void ejecutar () {
int i, j, l, m, n, o;
int nClases;
svm_parameter SVMparam= new svm_parameter();
svm_problem SVMp = null;
svm_model svr = null;
double exTmp[];
boolean marcas[];
boolean coincide, igual;
Instance inst;
int nSel;
double conjS[][];
double conjR[][];
int conjN[][];
boolean conjM[][];
int clasesS[];
Referencia orden[];
int vecinos[][];
Vector asociados[];
double dist, bestD;
int vecinosTemp[];
double distTemp[];
int aciertosSin;
int claseObt;
int mayoria;
boolean parar;
long tiempo = System.currentTimeMillis();
/*Getting the number of different classes*/
nClases = 0;
for (i=0; i<clasesTrain.length; i++)
if (clasesTrain[i] > nClases)
nClases = clasesTrain[i];
nClases++;
//SVM PARAMETERS
SVMparam.C = C;
SVMparam.cache_size = 10; //10MB of cache
SVMparam.degree = degree;
SVMparam.eps = eps;
SVMparam.gamma = gamma;
SVMparam.nr_weight = 0;
SVMparam.nu = nu;
SVMparam.p = p;
SVMparam.shrinking = shrinking;
SVMparam.probability = 0;
if(kernelType.compareTo("LINEAR")==0){
SVMparam.kernel_type = svm_parameter.LINEAR;
}else if(kernelType.compareTo("POLY")==0){
SVMparam.kernel_type = svm_parameter.POLY;
}else if(kernelType.compareTo("RBF")==0){
SVMparam.kernel_type = svm_parameter.RBF;
}else if(kernelType.compareTo("SIGMOID")==0){
SVMparam.kernel_type = svm_parameter.SIGMOID;
}
SVMparam.svm_type = svm_parameter.C_SVC;
SVMp = new svm_problem();
SVMp.l = datosTrain.length;
SVMp.y = new double[SVMp.l];
SVMp.x = new svm_node[SVMp.l][datosTrain[0].length+1];
for(i=0;i<SVMp.l;i++)
for(j=0;j<Attributes.getInputNumAttributes()+1;j++)
SVMp.x[i][j] = new svm_node();
for (i=0; i<datosTrain.length; i++) {
SVMp.y[i] = clasesTrain[i];
for (j=0; j < Attributes.getInputNumAttributes(); j++){
SVMp.x[i][j].index = j;
SVMp.x[i][j].value = datosTrain[i][j];
}
//end of instance
SVMp.x[i][Attributes.getInputNumAttributes()].index = -1;
}
if(svm.svm_check_parameter(SVMp, SVMparam)!=null){
System.err.print("SVM parameter error in training: ");
System.err.println(svm.svm_check_parameter(SVMp, SVMparam));
System.exit(-1);
}
//Train the SVM
svr = svm.svm_train(SVMp, SVMparam);
exTmp = new double[datosTrain[0].length];
marcas = new boolean[datosTrain.length];
Arrays.fill(marcas, false);
nSel = 0;
for (i=0; i<svr.getSV().length; i++) {
for (j=0; j<svr.getSV()[i].length-1; j++) {
exTmp[j] = svr.getSV()[i][j].value;
}
coincide = false;
for (j=0; j<datosTrain.length && !coincide; j++) {
igual = true;
for (l=0; l<datosTrain[j].length && igual; l++) {
if (exTmp[l] != datosTrain[j][l]) {
igual = false;
}
}
if (igual) {
marcas[j] = true;
nSel++;
coincide = true;
}
}
}
/*Building an instance vector with distances to the nearest enemy*/
orden = new Referencia[datosTrain.length];
for (i=0; i<datosTrain.length; i++) {
bestD = Double.POSITIVE_INFINITY;
for (j=0; j<datosTrain.length; j++) {
if (clasesTrain[i] != clasesTrain[j]) {
dist = KNN.distancia(datosTrain[i], realTrain[i], nominalTrain[i], nulosTrain[i], datosTrain[j], realTrain[j], nominalTrain[j], nulosTrain[j], distanceEu);
if (dist < bestD)
bestD = dist;
}
}
orden[i] = new Referencia (i, bestD);
}
/*Sort the previos vector*/
Arrays.sort(orden);
/*Inicialization of data structures of neighbors and associates*/
distTemp = new double[k+1];
vecinosTemp = new int[k+1];
vecinos = new int[datosTrain.length][k+1];
asociados = new Vector[datosTrain.length];
for (i=0; i<datosTrain.length; i++)
asociados[i] = new Vector ();
/*Body of the DROP2 algorithm. It calculates, for each instance, a set of associates instances
and look if the deletion of the main instance produces a change of accuracy in those associates*/
for (i=0; i<datosTrain.length; i++) {
/*Calculate the k+1 nearest neighbors of each instance*/
KNN.evaluacionKNN2(k+1, datosTrain, realTrain, nominalTrain, nulosTrain, clasesTrain, datosTrain[i], realTrain[i], nominalTrain[i], nulosTrain[i], nClases, distanceEu, vecinos[i]);
for (j=0; j<vecinos[i].length; j++) {
asociados[vecinos[i][j]].addElement (new Referencia (i,0));
}
}
/*Check if deleting or not the instances considering the WITH and WITHOUT sets*/
for (o=0; o<datosTrain.length; o++){
i = orden[o].entero;
if (marcas[i]) {
aciertosSin = 0;
marcas[i] = false;
nSel--;
/*Construction of S set from the temporal flags*/
conjS = new double[nSel][datosTrain[0].length];
conjR = new double[nSel][datosTrain[0].length];
conjN = new int[nSel][datosTrain[0].length];
conjM = new boolean[nSel][datosTrain[0].length];
clasesS = new int[nSel];
for (m=0, l=0; m<datosTrain.length; m++) {
if (marcas[m]) { //the instance will be evaluated
for (j=0; j<datosTrain[0].length; j++) {
conjS[l][j] = datosTrain[m][j];
conjR[l][j] = realTrain[m][j];
conjN[l][j] = nominalTrain[m][j];
conjM[l][j] = nulosTrain[m][j];
}
clasesS[l] = clasesTrain[m];
l++;
}
}
marcas[i] = true;
nSel++;
/*Evaluation of associates without the instance in T*/
for (j=0; j<k+1; j++) {
claseObt = KNN.evaluacionKNN2(k, conjS, conjR, conjN, conjM, clasesS, datosTrain[vecinos[i][j]], realTrain[vecinos[i][j]], nominalTrain[vecinos[i][j]], nulosTrain[vecinos[i][j]], nClases, distanceEu);
if (claseObt == clasesTrain[vecinos[i][j]]) //it classify it correctly
aciertosSin++;
}
mayoria = (k+1) / 2;
if (aciertosSin > mayoria) {
/*Delete P of S*/
marcas[i] = false;
nSel--;
/*For each associate of P, search a new nearest neighbor*/
for (j=0; j<asociados[i].size(); j++) {
for (l=0; l<k+1; l++) {
vecinosTemp[l] = vecinos[((Referencia)(asociados[i].elementAt(j))).entero][l];
vecinos[((Referencia)(asociados[i].elementAt(j))).entero][l] = -1;
distTemp[l] = Double.POSITIVE_INFINITY;
}
for (l=0; l<datosTrain.length; l++) {
if (marcas[l]) { //is in S
dist = KNN.distancia(datosTrain[((Referencia)(asociados[i].elementAt(j))).entero], realTrain[((Referencia)(asociados[i].elementAt(j))).entero], nominalTrain[((Referencia)(asociados[i].elementAt(j))).entero], nulosTrain[((Referencia)(asociados[i].elementAt(j))).entero], datosTrain[l], realTrain[l], nominalTrain[l], nulosTrain[l], distanceEu);
parar = false;
/*Get the nearest neighbors in this situation again*/
for (m=0; m<(k+1) && !parar; m++) {
if (dist < distTemp[m]) {
parar = true;
for (n=m+1; n<k+1; n++) {
distTemp[n] = distTemp[n-1];
vecinos[((Referencia)(asociados[i].elementAt(j))).entero][n] = vecinos[((Referencia)(asociados[i].elementAt(j))).entero][n-1];
}
distTemp[m] = dist;
vecinos[((Referencia)(asociados[i].elementAt(j))).entero][m] = l;
}
}
}
}
/*Add to the list of associates of the new neighbor this instance*/
for (l=0; l<k+1; l++) {
parar = false;
for (m=0; m<asociados[vecinosTemp[l]].size() && !parar; m++) {
if (((Referencia)(asociados[vecinosTemp[l]].elementAt(m))).entero == ((Referencia)(asociados[i].elementAt(j))).entero
&& vecinosTemp[l] != i) {
asociados[vecinosTemp[l]].removeElementAt(m);
parar = true;
}
}
}
for (l=0; l<k+1; l++) {
asociados[vecinos[((Referencia)(asociados[i].elementAt(j))).entero][l]].addElement(new Referencia (((Referencia)(asociados[i].elementAt(j))).entero,0));
}
}
}
}
}
/*Construction of S set from the temporal flags*/
conjS = new double[nSel][datosTrain[0].length];
conjR = new double[nSel][datosTrain[0].length];
conjN = new int[nSel][datosTrain[0].length];
conjM = new boolean[nSel][datosTrain[0].length];
clasesS = new int[nSel];
for (m=0, l=0; m<datosTrain.length; m++) {
if (marcas[m]) { //the instance will be evaluated
for (j=0; j<datosTrain[0].length; j++) {
conjS[l][j] = datosTrain[m][j];
conjR[l][j] = realTrain[m][j];
conjN[l][j] = nominalTrain[m][j];
conjM[l][j] = nulosTrain[m][j];
}
clasesS[l] = clasesTrain[m];
l++;
}
}
System.out.println("SVBPS "+ relation + " " + (double)(System.currentTimeMillis()-tiempo)/1000.0 + "s");
OutputIS.escribeSalida(ficheroSalida[0], conjR, conjN, conjM, clasesS, entradas, salida, nEntradas, relation);
OutputIS.escribeSalida(ficheroSalida[1], test, entradas, salida, nEntradas, relation);
}
public void leerConfiguracion (String ficheroScript) {
String fichero, linea, token;
StringTokenizer lineasFichero, tokens;
byte line[];
int i, j;
ficheroSalida = new String[2];
fichero = Fichero.leeFichero (ficheroScript);
lineasFichero = new StringTokenizer (fichero,"\n\r");
lineasFichero.nextToken();
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
token = tokens.nextToken();
/*Getting the names of the training and test files*/
line = token.getBytes();
for (i=0; line[i]!='\"'; i++);
i++;
for (j=i; line[j]!='\"'; j++);
ficheroTraining = new String (line,i,j-i);
for (i=j+1; line[i]!='\"'; i++);
i++;
for (j=i; line[j]!='\"'; j++);
ficheroTest = new String (line,i,j-i);
/*Getting the path and base name of the results files*/
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
token = tokens.nextToken();
/*Getting the names of output files*/
line = token.getBytes();
for (i=0; line[i]!='\"'; i++);
i++;
for (j=i; line[j]!='\"'; j++);
ficheroSalida[0] = new String (line,i,j-i);
for (i=j+1; line[i]!='\"'; i++);
i++;
for (j=i; line[j]!='\"'; j++);
ficheroSalida[1] = new String (line,i,j-i);
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
kernelType = tokens.nextToken().substring(1);
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
C = Double.parseDouble(tokens.nextToken().substring(1));
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
eps = Double.parseDouble(tokens.nextToken().substring(1));
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
degree = Integer.parseInt(tokens.nextToken().substring(1));
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
gamma = Double.parseDouble(tokens.nextToken().substring(1));
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
nu = Double.parseDouble(tokens.nextToken().substring(1));
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
p = Double.parseDouble(tokens.nextToken().substring(1));
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
shrinking = Integer.parseInt(tokens.nextToken().substring(1));
/*Getting the number of neighbors*/
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
k = Integer.parseInt(tokens.nextToken().substring(1));
/*Getting the type of distance function*/
linea = lineasFichero.nextToken();
tokens = new StringTokenizer (linea, "=");
tokens.nextToken();
distanceEu = tokens.nextToken().substring(1).equalsIgnoreCase("Euclidean")?true:false;
}
}