001/**
002 * Copyright (c) 2011, The University of Southampton and the individual contributors.
003 * All rights reserved.
004 *
005 * Redistribution and use in source and binary forms, with or without modification,
006 * are permitted provided that the following conditions are met:
007 *
008 *   *  Redistributions of source code must retain the above copyright notice,
009 *      this list of conditions and the following disclaimer.
010 *
011 *   *  Redistributions in binary form must reproduce the above copyright notice,
012 *      this list of conditions and the following disclaimer in the documentation
013 *      and/or other materials provided with the distribution.
014 *
015 *   *  Neither the name of the University of Southampton nor the names of its
016 *      contributors may be used to endorse or promote products derived from this
017 *      software without specific prior written permission.
018 *
019 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
020 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
021 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
022 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
023 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
024 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
025 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
026 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
027 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
028 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
029 */
030package org.openimaj.ml.clustering.kmeans;
031
032import java.io.IOException;
033
034import org.openimaj.data.DataSource;
035
036import com.rits.cloning.Cloner;
037
038/**
039 * Initialisation for K-Means clustering. Given a data source of samples and a
040 * set of clusters to fill, implementations of this class should initialise the
041 * KMeans algorithm.
042 *
043 * A default RANDOM implementation is provided which uses
044 * {@link DataSource#getRandomRows}
045 *
046 * @author Jonathon Hare (jsh2@ecs.soton.ac.uk)
047 * @author Sina Samangooei (ss@ecs.soton.ac.uk)
048 *
049 * @param <T>
050 *            Type of object being clustered
051 */
052public abstract class FeatureVectorKMeansInit<T> {
053        /**
054         * Initialise the centroids based on the given data.
055         *
056         * @param bds
057         *            the data source of samples
058         * @param clusters
059         *            the clusters to init
060         * @throws IOException
061         *             problem reading samples
062         */
063        public abstract void initKMeans(DataSource<T> bds, T[] clusters) throws IOException;
064
065        /**
066         * Simple kmeans initialized on randomly selected samples.
067         *
068         * @author Jonathon Hare (jsh2@ecs.soton.ac.uk)
069         * @author Sina Samangooei (ss@ecs.soton.ac.uk)
070         *
071         * @param <T>
072         *            Type of object being clustered
073         */
074        public static class RANDOM<T> extends FeatureVectorKMeansInit<T> {
075                @Override
076                public void initKMeans(DataSource<T> bds, T[] clusters) throws IOException {
077                        bds.getRandomRows(clusters);
078
079                        final Cloner cloner = new Cloner();
080                        for (int i = 0; i < clusters.length; i++) {
081                                clusters[i] = cloner.deepClone(clusters[i]);
082                        }
083                }
084        }
085}