/*
* RapidMiner
*
* Copyright (C) 2001-2011 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.clustering.clusterer;
import java.util.Arrays;
import java.util.LinkedList;
import java.util.List;
import com.rapidminer.example.Attribute;
import com.rapidminer.example.Example;
import com.rapidminer.example.ExampleSet;
import com.rapidminer.example.Tools;
import com.rapidminer.example.table.AttributeFactory;
import com.rapidminer.operator.OperatorCapability;
import com.rapidminer.operator.OperatorDescription;
import com.rapidminer.operator.OperatorException;
import com.rapidminer.operator.clustering.ClusterModel;
import com.rapidminer.operator.learner.CapabilityProvider;
import com.rapidminer.operator.learner.functions.kernel.jmysvm.examples.SVMExample;
import com.rapidminer.operator.learner.functions.kernel.jmysvm.kernel.Kernel;
import com.rapidminer.operator.learner.functions.kernel.jmysvm.kernel.KernelDot;
import com.rapidminer.operator.learner.functions.kernel.jmysvm.kernel.KernelNeural;
import com.rapidminer.operator.learner.functions.kernel.jmysvm.kernel.KernelPolynomial;
import com.rapidminer.operator.learner.functions.kernel.jmysvm.kernel.KernelRadial;
import com.rapidminer.parameter.ParameterType;
import com.rapidminer.parameter.ParameterTypeCategory;
import com.rapidminer.parameter.ParameterTypeDouble;
import com.rapidminer.parameter.ParameterTypeInt;
import com.rapidminer.parameter.UndefinedParameterError;
import com.rapidminer.parameter.conditions.EqualTypeCondition;
import com.rapidminer.tools.Ontology;
/**
* An implementation of Support Vector Clustering based on {@rapidminer.cite BenHur/etal/2001a}.
* This operator will create a cluster attribute if not present yet.
*
* @author Stefan Rueping, Ingo Mierswa, Michael Wurst, Sebastian Land
*/
public class SVClustering extends RMAbstractClusterer implements CapabilityProvider {
public static final String MIN_PTS_NAME = "min_pts";
/** The parameter name for "The SVM kernel type" */
public static final String PARAMETER_KERNEL_TYPE = "kernel_type";
/** The parameter name for "The SVM kernel parameter gamma (radial)." */
public static final String PARAMETER_KERNEL_GAMMA = "kernel_gamma";
/** The parameter name for "The SVM kernel parameter degree (polynomial)." */
public static final String PARAMETER_KERNEL_DEGREE = "kernel_degree";
/** The parameter name for "The SVM kernel parameter a (neural)." */
public static final String PARAMETER_KERNEL_A = "kernel_a";
/** The parameter name for "The SVM kernel parameter b (neural)." */
public static final String PARAMETER_KERNEL_B = "kernel_b";
/** The parameter name for "Size of the cache for kernel evaluations im MB " */
public static final String PARAMETER_KERNEL_CACHE = "kernel_cache";
/** The parameter name for "Precision on the KKT conditions" */
public static final String PARAMETER_CONVERGENCE_EPSILON = "convergence_epsilon";
/** The parameter name for "Stop after this many iterations" */
public static final String PARAMETER_MAX_ITERATIONS = "max_iterations";
/** The parameter name for "The fraction of allowed outliers." */
public static final String PARAMETER_P = "p";
/** The parameter name for "Use this radius instead of the calculated one (-1 for calculated radius)." */
public static final String PARAMETER_R = "r";
/** The parameter name for "The number of virtual sample points to check for neighborship." */
public static final String PARAMETER_NUMBER_SAMPLE_POINTS = "number_sample_points";
/** The kernels which can be used from RapidMiner for the mySVM / myKLR. */
private static final String[] KERNEL_TYPES = {
"dot", "radial", "polynomial", "neural"
};
/** Indicates a linear kernel. */
public static final int KERNEL_DOT = 0;
/** Indicates a rbf kernel. */
public static final int KERNEL_RADIAL = 1;
/** Indicates a polynomial kernel. */
public static final int KERNEL_POLYNOMIAL = 2;
/** Indicates a neural net kernel. */
public static final int KERNEL_NEURAL = 3;
protected static final int UNASSIGNED = -1;
public static final int NOISE = 0;
public static final String NOISE_CLUSTER_DESCRIPTION = "Outliers";
public SVClustering(OperatorDescription description) {
super(description);
}
@Override
public ClusterModel generateClusterModel(ExampleSet exampleSet) throws OperatorException {
// checking and creating ids if necessary
Tools.checkAndCreateIds(exampleSet);
// additional checks
Tools.onlyNonMissingValues(exampleSet, "SVClustering");
Tools.isNonEmpty(exampleSet);
// creating kernel
int kernelType = getParameterAsInt(PARAMETER_KERNEL_TYPE);
int cacheSize = getParameterAsInt(PARAMETER_KERNEL_CACHE);
Kernel kernel = createKernel(kernelType);
if (kernelType == KERNEL_RADIAL)
((KernelRadial) kernel).setGamma(getParameterAsDouble(PARAMETER_KERNEL_GAMMA));
else if (kernelType == KERNEL_POLYNOMIAL)
((KernelPolynomial) kernel).setDegree(getParameterAsInt(PARAMETER_KERNEL_DEGREE));
else if (kernelType == KERNEL_NEURAL)
((KernelNeural) kernel).setParameters(getParameterAsDouble(PARAMETER_KERNEL_A), getParameterAsDouble(PARAMETER_KERNEL_B));
SVCExampleSet svmExamples = new SVCExampleSet(exampleSet, false);
kernel.init(svmExamples, cacheSize);
// creating kernel using SVClusteringAlgorithm
SVClusteringAlgorithm clustering = new SVClusteringAlgorithm(this, kernel, svmExamples);
clustering.train();
// doing neighborhood search for density estimation
int nextClusterId = 0;
int minPts = getParameterAsInt(MIN_PTS_NAME);
int[] clusterAssignments = new int[exampleSet.size()];
Arrays.fill(clusterAssignments, UNASSIGNED);
int i = 0;
for (Example example: exampleSet) {
if (clusterAssignments[i] == UNASSIGNED) {
LinkedList<Integer> neighbours = getNeighbours(exampleSet, example, i, clusterAssignments, clustering);
if (neighbours.size() >= minPts) {
nextClusterId++;
clusterAssignments[i] = nextClusterId;
for (int exampleIndex: neighbours) {
clusterAssignments[exampleIndex] = nextClusterId;
}
while (neighbours.size() > 0) {
// Take the first index from the queue and fetch indexed example
int index = neighbours.poll().intValue();
Example neighbourExample = exampleSet.getExample(index);
// Find its neighbours and if the density is sufficient
// recurse through them and
// assign it the current cluster id
LinkedList<Integer> neighboursRecursive = getNeighbours(exampleSet, neighbourExample, index, clusterAssignments, clustering);
if (neighboursRecursive.size() >= minPts) {
for (int recursiveIndex: neighboursRecursive) {
// If already identified as noise, just assign
// it, if its unclassified, add to queue
if (clusterAssignments[recursiveIndex] == UNASSIGNED) {
neighbours.add(recursiveIndex);
}
clusterAssignments[recursiveIndex] = nextClusterId;
}
}
}
} else {
clusterAssignments[i] = NOISE;
}
}
i++;
}
ClusterModel model = new ClusterModel(exampleSet, nextClusterId + 1, getParameterAsBoolean(RMAbstractClusterer.PARAMETER_ADD_AS_LABEL), getParameterAsBoolean(RMAbstractClusterer.PARAMETER_REMOVE_UNLABELED));
model.setClusterAssignments(clusterAssignments, exampleSet);
if (addsClusterAttribute()) {
Attribute cluster = AttributeFactory.createAttribute("cluster", Ontology.NOMINAL);
exampleSet.getExampleTable().addAttribute(cluster);
exampleSet.getAttributes().setCluster(cluster);
i = 0;
for (Example example: exampleSet) {
if (clusterAssignments[i] == NOISE)
example.setValue(cluster, "noise");
else
example.setValue(cluster, "cluster_" + clusterAssignments[i]);
i++;
}
}
return model;
}
protected LinkedList<Integer> getNeighbours(ExampleSet exampleSet, Example centroid, int centroidIndex, final int[] assignments, SVClusteringAlgorithm clustering) throws UndefinedParameterError {
LinkedList<Integer> neighbors = new LinkedList<Integer>();
double paramR = getParameterAsDouble(PARAMETER_R);
double maxRadius = paramR < 0 ? clustering.getR() : paramR;
int numSamplePoints = getParameterAsInt(PARAMETER_NUMBER_SAMPLE_POINTS);
int i = 0;
for(Example example: exampleSet) {
if (i != centroidIndex && assignments[i] == UNASSIGNED) {
double[] directions = new double[centroid.getAttributes().size()];
int x = 0;
for (Attribute attribute : centroid.getAttributes()) {
directions[x++] = example.getValue(attribute) - centroid.getValue(attribute);
}
boolean addAsNeighbor = true;
for (int j = 0; j < numSamplePoints; j++) {
if (addAsNeighbor) {
double[] virtualExample = new double[directions.length];
x = 0;
for (Attribute attribute : centroid.getAttributes()) {
virtualExample[x] = centroid.getValue(attribute) + (j + 1) * directions[x] / (numSamplePoints + 1);
x++;
}
SVMExample svmExample = new SVMExample(virtualExample);
double currentRadius = clustering.predict(svmExample);
if (currentRadius > maxRadius) {
addAsNeighbor = false;
break;
}
} else {
break;
}
}
if (addAsNeighbor)
neighbors.add(i);
}
i++;
}
return neighbors;
}
/**
* Creates a new kernel of the given type. The kernel type has to be one out of KERNEL_DOT, KERNEL_RADIAL, KERNEL_POLYNOMIAL, or KERNEL_NEURAL.
*/
public static Kernel createKernel(int kernelType) {
switch (kernelType) {
case KERNEL_RADIAL:
return new KernelRadial();
case KERNEL_POLYNOMIAL:
return new KernelPolynomial();
case KERNEL_NEURAL:
return new KernelNeural();
default:
return new KernelDot();
}
}
@Override
public boolean supportsCapability(OperatorCapability capability) {
switch (capability) {
case BINOMINAL_ATTRIBUTES:
case POLYNOMINAL_ATTRIBUTES:
return false;
default:
return true;
}
}
@Override
public List<ParameterType> getParameterTypes() {
List<ParameterType> types = super.getParameterTypes();
ParameterType type;
type = new ParameterTypeInt(MIN_PTS_NAME, "The minimal number of points in each cluster.", 0, Integer.MAX_VALUE, 2);
type.setExpert(false);
types.add(type);
type = new ParameterTypeCategory(PARAMETER_KERNEL_TYPE, "The SVM kernel type", KERNEL_TYPES, KERNEL_RADIAL);
type.setExpert(false);
types.add(type);
type = new ParameterTypeDouble(PARAMETER_KERNEL_GAMMA, "The SVM kernel parameter gamma (radial).", 0.0d, Double.POSITIVE_INFINITY, 1.0d);
type.registerDependencyCondition(new EqualTypeCondition(this, PARAMETER_KERNEL_TYPE, KERNEL_TYPES, true, KERNEL_RADIAL));
type.setExpert(false);
types.add(type);
type = new ParameterTypeInt(PARAMETER_KERNEL_DEGREE, "The SVM kernel parameter degree (polynomial).", 0, Integer.MAX_VALUE, 2);
type.registerDependencyCondition(new EqualTypeCondition(this, PARAMETER_KERNEL_TYPE, KERNEL_TYPES, true, KERNEL_POLYNOMIAL));
type.setExpert(false);
types.add(type);
type = new ParameterTypeDouble(PARAMETER_KERNEL_A, "The SVM kernel parameter a (neural).", Double.NEGATIVE_INFINITY, Double.POSITIVE_INFINITY, 1.0d);
type.registerDependencyCondition(new EqualTypeCondition(this, PARAMETER_KERNEL_TYPE, KERNEL_TYPES, true, KERNEL_NEURAL));
type.setExpert(false);
types.add(type);
type = new ParameterTypeDouble(PARAMETER_KERNEL_B, "The SVM kernel parameter b (neural).", Double.NEGATIVE_INFINITY, Double.POSITIVE_INFINITY, 0.0d);
type.registerDependencyCondition(new EqualTypeCondition(this, PARAMETER_KERNEL_TYPE, KERNEL_TYPES, true, KERNEL_NEURAL));
type.setExpert(false);
types.add(type);
types.add(new ParameterTypeInt(PARAMETER_KERNEL_CACHE, "Size of the cache for kernel evaluations im MB ", 0, Integer.MAX_VALUE, 200));
type = new ParameterTypeDouble(PARAMETER_CONVERGENCE_EPSILON, "Precision on the KKT conditions", 0.0d, Double.POSITIVE_INFINITY, 1e-3);
types.add(type);
types.add(new ParameterTypeInt(PARAMETER_MAX_ITERATIONS, "Stop after this many iterations", 1, Integer.MAX_VALUE, 100000));
type = new ParameterTypeDouble(PARAMETER_P, "The fraction of allowed outliers.", 0, 1, 0.0d);
types.add(type);
type = new ParameterTypeDouble(PARAMETER_R, "Use this radius instead of the calculated one (-1 for calculated radius).", -1.0d, Double.POSITIVE_INFINITY, -1.0d);
types.add(type);
types.add(new ParameterTypeInt(PARAMETER_NUMBER_SAMPLE_POINTS, "The number of virtual sample points to check for neighborship.", 1, Integer.MAX_VALUE, 20));
return types;
}
}