/*
* RapidMiner
*
* Copyright (C) 2001-2008 by Rapid-I and the contributors
*
* Complete list of developers available at our web site:
*
* http://rapid-i.com
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero 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 Affero General Public License for more details.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see http://www.gnu.org/licenses/.
*/
package com.rapidminer.operator.learner.functions.kernel.evosvm;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.List;
import com.rapidminer.datatable.DataTable;
import com.rapidminer.datatable.SimpleDataTable;
import com.rapidminer.datatable.SimpleDataTableRow;
import com.rapidminer.example.Attribute;
import com.rapidminer.example.Example;
import com.rapidminer.example.ExampleSet;
import com.rapidminer.gui.plotter.SimplePlotterDialog;
import com.rapidminer.operator.learner.functions.kernel.SupportVector;
import com.rapidminer.operator.learner.functions.kernel.functions.Kernel;
import com.rapidminer.operator.performance.EstimatedPerformance;
import com.rapidminer.operator.performance.PerformanceVector;
import com.rapidminer.tools.RandomGenerator;
import com.rapidminer.tools.math.optimization.ec.pso.PSOOptimization;
/**
* PSO approach for SVM optimization. Currently only classification problems are
* supported.
*
* @author Ingo Mierswa
* @version $Id: PSOSVMOptimization.java,v 1.3 2006/04/05 08:57:26 ingomierswa
* Exp $
*/
public class PSOSVMOptimization extends PSOOptimization {
/** Number smaller than this number are regarded as zero. */
private final static double IS_ZERO = 1e-10;
/** The training example set. */
private ExampleSet exampleSet;
/** The used kernel function. */
private Kernel kernel;
/** This parameter indicates the weight of errors. */
private double c;
/** This parameter indicates the weight of errors for regression. */
private double epsilon = 0.0d;
/** The label values. */
private double[] ys;
/** This function is to maximize. */
private OptimizationFunction optimizationFunction;
/** Indicates if a convergence plot should be drawn. */
private boolean showPlot = false;
private DataTable dataTable;
private SimplePlotterDialog plotter;
/** Creates a new evolutionary SVM optimization. */
public PSOSVMOptimization(ExampleSet exampleSet, // training data
Kernel kernel, double c, // double epsilon, // SVM paras
int maxIterations, int generationsWithoutImprovement, // convergence
// paras
int popSize, double inertiaWeight, double localWeight, double globalWeight, boolean dynamicInertiaWeight, boolean showPlot, RandomGenerator random) {
super(popSize < 1 ? exampleSet.size() : popSize, exampleSet.size(), maxIterations, generationsWithoutImprovement, inertiaWeight, localWeight, globalWeight, 0.0d, 1.0d, dynamicInertiaWeight, random);
this.exampleSet = exampleSet;
this.kernel = kernel;
this.kernel.init(exampleSet);
this.c = c;
if (this.c <= 0.0d) {
this.c = 0.0d;
for (int i = 0; i < exampleSet.size(); i++) {
this.c += kernel.getDistance(i, i);
}
this.c = exampleSet.size() / this.c;
exampleSet.getLog().log("Determine probably good value for C: set to " + this.c);
}
setMinValue(0.0d);
setMaxValue(this.c);
// label values
this.ys = new double[exampleSet.size()];
Iterator<Example> reader = exampleSet.iterator();
int index = 0;
Attribute label = exampleSet.getAttributes().getLabel();
boolean regression = !label.isNominal() && label.getMapping().size() != 2;
while (reader.hasNext()) {
Example example = reader.next();
if (!regression)
ys[index++] = example.getLabel() == label.getMapping().getPositiveIndex() ? 1.0d : -1.0d;
else
ys[index++] = example.getLabel();
}
// optimization function
if (!regression)
this.optimizationFunction = new ClassificationOptimizationFunction(false);
else
this.optimizationFunction = new RegressionOptimizationFunction(epsilon);
// plotter
this.showPlot = showPlot;
if (showPlot) {
dataTable = new SimpleDataTable("Fitness vs. Generations", new String[] { "Generations", "Best Fitness", "Current Fitness" });
plotter = new SimplePlotterDialog(dataTable, false);
plotter.setXAxis(0);
plotter.plotColumn(1, true);
plotter.plotColumn(2, true);
plotter.setVisible(true);
}
}
public void nextIteration() {
if (showPlot)
dataTable.add(new SimpleDataTableRow(new double[] { getGeneration(), getBestFitnessEver(), getBestFitnessInGeneration() }));
}
/** Evaluates the individuals of the given population. */
public PerformanceVector evaluateIndividual(double[] individual) {
double fitness = optimizationFunction.getFitness(individual, ys, kernel)[0];
PerformanceVector result = new PerformanceVector();
result.addCriterion(new EstimatedPerformance("SVMOptValue", fitness, 1, false));
return result;
}
/**
* Returns a model containing all support vectors, i.e. the examples with
* non-zerp alphas.
*/
public EvoSVMModel getModel(double[] alphas) {
if (showPlot) {
plotter.dispose();
}
// calculate support vectors
Iterator<Example> reader = exampleSet.iterator();
List<SupportVector> supportVectors = new ArrayList<SupportVector>();
int index = 0;
while (reader.hasNext()) {
double currentAlpha = alphas[index];
Example currentExample = reader.next();
if (currentAlpha != 0.0d) {
double[] x = new double[exampleSet.getAttributes().size()];
int a = 0;
for (Attribute attribute : exampleSet.getAttributes())
x[a++] = currentExample.getValue(attribute);
supportVectors.add(new SupportVector(x, ys[index], currentAlpha));
}
index++;
}
// calculate all sum values
double[] sum = new double[exampleSet.size()];
reader = exampleSet.iterator();
index = 0;
// double maxNeg = Double.NEGATIVE_INFINITY;
// double minPos = Double.POSITIVE_INFINITY;
while (reader.hasNext()) {
Example current = reader.next();
double[] x = new double[exampleSet.getAttributes().size()];
int a = 0;
for (Attribute attribute : exampleSet.getAttributes())
x[a++] = current.getValue(attribute);
sum[index] = kernel.getSum(supportVectors, x);
// if ((ys[index] < 0) && (sum[index] > maxNeg))
// maxNeg = sum[index];
// if ((ys[index] > 0) && (sum[index] < minPos))
// minPos = sum[index];
index++;
}
// return new EvoSVMModel(exampleSet.getLabel(), supportVectors, kernel,
// (double)(-maxNeg - minPos) / 2.0d);
// calculate b (from Stefan's mySVM code)
double bSum = 0.0d;
int bCounter = 0;
for (int i = 0; i < alphas.length; i++) {
if ((ys[i] * alphas[i] - c < -IS_ZERO) && (ys[i] * alphas[i] > IS_ZERO)) {
bSum += ys[i] - sum[i] - epsilon;
bCounter++;
} else if ((ys[i] * alphas[i] + c > IS_ZERO) && (ys[i] * alphas[i] < -IS_ZERO)) {
bSum += ys[i] - sum[i] - epsilon;
bCounter++;
}
}
if (bCounter == 0) {
// unlikely
bSum = 0.0d;
for (int i = 0; i < alphas.length; i++) {
if ((ys[i] * alphas[i] < IS_ZERO) && (ys[i] * alphas[i] > -IS_ZERO)) {
bSum += ys[i] - sum[i];
bCounter++;
}
}
if (bCounter == 0) {
// even unlikelier
bSum = 0.0d;
for (int i = 0; i < alphas.length; i++) {
bSum += ys[i] - sum[i];
bCounter++;
}
}
}
return new EvoSVMModel(exampleSet, supportVectors, kernel, (bSum / bCounter));
}
}