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.spectral; 031 032import java.io.File; 033import java.io.IOException; 034 035import org.apache.log4j.Logger; 036import org.openimaj.io.IOUtils; 037 038import ch.akuhn.matrix.SparseMatrix; 039import ch.akuhn.matrix.eigenvalues.Eigenvalues; 040 041/** 042 * {@link DoubleSpectralClustering} extention which knows how to write and read its eigenvectors to disk 043 * and therefore not regenerate them when calling the underlying {@link PreparedSpectralClustering} 044 * 045 * @author Sina Samangooei (ss@ecs.soton.ac.uk) 046 * 047 */ 048public class CachedDoubleSpectralClustering extends DoubleSpectralClustering{ 049 private final static Logger logger = Logger.getLogger(CachedDoubleSpectralClustering.class); 050 private File cache; 051 052 /** 053 * @param cache location to cache the eigenvectors 054 * @param conf 055 * cluster the eigen vectors 056 */ 057 public CachedDoubleSpectralClustering(File cache, SpectralClusteringConf<double[]> conf) { 058 super(conf); 059 this.cache = cache; 060 } 061 062 protected Eigenvalues spectralCluster(SparseMatrix data) { 063 Eigenvalues eig = null; 064 if(cache.exists()){ 065 logger.debug("Loading eigenvectors from cache"); 066 try { 067 eig = IOUtils.readFromFile(cache); 068 } catch (IOException e) { 069 throw new RuntimeException(e); 070 } 071 } 072 else{ 073 // Compute the laplacian of the graph 074 logger.debug("Cache empty, recreating eigenvectors"); 075 final SparseMatrix laplacian = laplacian(data); 076 eig = laplacianEigenVectors(laplacian); 077 try { 078 logger.debug("Writing eigenvectors to cache"); 079 IOUtils.writeToFile(eig, cache); 080 } catch (IOException e) { 081 throw new RuntimeException(e); 082 } 083 } 084 085 return eig; 086 } 087 088}