| 1 | // Copyright 2015 Georg-August-Universität Göttingen, Germany |
|---|
| 2 | // |
|---|
| 3 | // Licensed under the Apache License, Version 2.0 (the "License"); |
|---|
| 4 | // you may not use this file except in compliance with the License. |
|---|
| 5 | // You may obtain a copy of the License at |
|---|
| 6 | // |
|---|
| 7 | // http://www.apache.org/licenses/LICENSE-2.0 |
|---|
| 8 | // |
|---|
| 9 | // Unless required by applicable law or agreed to in writing, software |
|---|
| 10 | // distributed under the License is distributed on an "AS IS" BASIS, |
|---|
| 11 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
|---|
| 12 | // See the License for the specific language governing permissions and |
|---|
| 13 | // limitations under the License. |
|---|
| 14 | |
|---|
| 15 | package de.ugoe.cs.cpdp.training; |
|---|
| 16 | |
|---|
| 17 | import java.util.LinkedList; |
|---|
| 18 | import java.util.List; |
|---|
| 19 | |
|---|
| 20 | import de.ugoe.cs.cpdp.util.WekaUtils; |
|---|
| 21 | import weka.classifiers.AbstractClassifier; |
|---|
| 22 | import weka.classifiers.Classifier; |
|---|
| 23 | import weka.core.Instance; |
|---|
| 24 | import weka.core.Instances; |
|---|
| 25 | |
|---|
| 26 | /** |
|---|
| 27 | * <p> |
|---|
| 28 | * Implements training following the LASER classification scheme. |
|---|
| 29 | * </p> |
|---|
| 30 | * |
|---|
| 31 | * @author Steffen Herbold |
|---|
| 32 | */ |
|---|
| 33 | public class WekaLASERTraining extends WekaBaseTraining implements ITrainingStrategy { |
|---|
| 34 | |
|---|
| 35 | /** |
|---|
| 36 | * Internal classifier used for LASER. |
|---|
| 37 | */ |
|---|
| 38 | private final LASERClassifier internalClassifier = new LASERClassifier(); |
|---|
| 39 | |
|---|
| 40 | /* |
|---|
| 41 | * (non-Javadoc) |
|---|
| 42 | * |
|---|
| 43 | * @see de.ugoe.cs.cpdp.training.WekaBaseTraining#getClassifier() |
|---|
| 44 | */ |
|---|
| 45 | @Override |
|---|
| 46 | public Classifier getClassifier() { |
|---|
| 47 | return internalClassifier; |
|---|
| 48 | } |
|---|
| 49 | |
|---|
| 50 | /* |
|---|
| 51 | * (non-Javadoc) |
|---|
| 52 | * |
|---|
| 53 | * @see de.ugoe.cs.cpdp.training.ITrainingStrategy#apply(weka.core.Instances) |
|---|
| 54 | */ |
|---|
| 55 | @Override |
|---|
| 56 | public void apply(Instances traindata) { |
|---|
| 57 | try { |
|---|
| 58 | internalClassifier.buildClassifier(traindata); |
|---|
| 59 | } |
|---|
| 60 | catch (Exception e) { |
|---|
| 61 | throw new RuntimeException(e); |
|---|
| 62 | } |
|---|
| 63 | } |
|---|
| 64 | |
|---|
| 65 | /** |
|---|
| 66 | * <p> |
|---|
| 67 | * Internal helper class that defines the laser classifier. |
|---|
| 68 | * </p> |
|---|
| 69 | * |
|---|
| 70 | * @author Steffen Herbold |
|---|
| 71 | */ |
|---|
| 72 | public class LASERClassifier extends AbstractClassifier { |
|---|
| 73 | |
|---|
| 74 | /** |
|---|
| 75 | * Default serial ID. |
|---|
| 76 | */ |
|---|
| 77 | private static final long serialVersionUID = 1L; |
|---|
| 78 | |
|---|
| 79 | /** |
|---|
| 80 | * Internal reference to the classifier. |
|---|
| 81 | */ |
|---|
| 82 | private Classifier laserClassifier = null; |
|---|
| 83 | |
|---|
| 84 | /** |
|---|
| 85 | * Internal storage of the training data required for NN analysis. |
|---|
| 86 | */ |
|---|
| 87 | private Instances traindata = null; |
|---|
| 88 | |
|---|
| 89 | /* |
|---|
| 90 | * (non-Javadoc) |
|---|
| 91 | * |
|---|
| 92 | * @see weka.classifiers.AbstractClassifier#classifyInstance(weka.core.Instance) |
|---|
| 93 | */ |
|---|
| 94 | @Override |
|---|
| 95 | public double classifyInstance(Instance instance) throws Exception { |
|---|
| 96 | List<Integer> closestInstances = new LinkedList<>(); |
|---|
| 97 | double minDistance = Double.MAX_VALUE; |
|---|
| 98 | for (int i = 0; i < traindata.size(); i++) { |
|---|
| 99 | double distance = WekaUtils.hammingDistance(instance, traindata.get(i)); |
|---|
| 100 | if (distance < minDistance) { |
|---|
| 101 | minDistance = distance; |
|---|
| 102 | } |
|---|
| 103 | } |
|---|
| 104 | for (int i = 0; i < traindata.size(); i++) { |
|---|
| 105 | double distance = WekaUtils.hammingDistance(instance, traindata.get(i)); |
|---|
| 106 | if (distance <= minDistance) { |
|---|
| 107 | closestInstances.add(i); |
|---|
| 108 | } |
|---|
| 109 | } |
|---|
| 110 | if (closestInstances.size() == 1) { |
|---|
| 111 | int closestIndex = closestInstances.get(0); |
|---|
| 112 | Instance closestTrainingInstance = traindata.get(closestIndex); |
|---|
| 113 | List<Integer> closestToTrainingInstance = new LinkedList<>(); |
|---|
| 114 | double minTrainingDistance = Double.MAX_VALUE; |
|---|
| 115 | for (int i = 0; i < traindata.size(); i++) { |
|---|
| 116 | if (closestIndex != i) { |
|---|
| 117 | double distance = |
|---|
| 118 | WekaUtils.hammingDistance(closestTrainingInstance, traindata.get(i)); |
|---|
| 119 | if (distance < minTrainingDistance) { |
|---|
| 120 | minTrainingDistance = distance; |
|---|
| 121 | } |
|---|
| 122 | } |
|---|
| 123 | } |
|---|
| 124 | for (int i = 0; i < traindata.size(); i++) { |
|---|
| 125 | if (closestIndex != i) { |
|---|
| 126 | double distance = |
|---|
| 127 | WekaUtils.hammingDistance(closestTrainingInstance, traindata.get(i)); |
|---|
| 128 | if (distance <= minTrainingDistance) { |
|---|
| 129 | closestToTrainingInstance.add(i); |
|---|
| 130 | } |
|---|
| 131 | } |
|---|
| 132 | } |
|---|
| 133 | if (closestToTrainingInstance.size() == 1) { |
|---|
| 134 | return laserClassifier.classifyInstance(instance); |
|---|
| 135 | } |
|---|
| 136 | else { |
|---|
| 137 | double label = Double.NaN; |
|---|
| 138 | boolean allEqual = true; |
|---|
| 139 | for (Integer index : closestToTrainingInstance) { |
|---|
| 140 | if (Double.isNaN(label)) { |
|---|
| 141 | label = traindata.get(index).classValue(); |
|---|
| 142 | } |
|---|
| 143 | else if (label != traindata.get(index).classValue()) { |
|---|
| 144 | allEqual = false; |
|---|
| 145 | break; |
|---|
| 146 | } |
|---|
| 147 | } |
|---|
| 148 | if (allEqual) { |
|---|
| 149 | return label; |
|---|
| 150 | } |
|---|
| 151 | else { |
|---|
| 152 | return laserClassifier.classifyInstance(instance); |
|---|
| 153 | } |
|---|
| 154 | } |
|---|
| 155 | } |
|---|
| 156 | else { |
|---|
| 157 | double label = Double.NaN; |
|---|
| 158 | boolean allEqual = true; |
|---|
| 159 | for (Integer index : closestInstances) { |
|---|
| 160 | if (Double.isNaN(label)) { |
|---|
| 161 | label = traindata.get(index).classValue(); |
|---|
| 162 | } |
|---|
| 163 | else if (label != traindata.get(index).classValue()) { |
|---|
| 164 | allEqual = false; |
|---|
| 165 | break; |
|---|
| 166 | } |
|---|
| 167 | } |
|---|
| 168 | if (allEqual) { |
|---|
| 169 | return label; |
|---|
| 170 | } |
|---|
| 171 | else { |
|---|
| 172 | return laserClassifier.classifyInstance(instance); |
|---|
| 173 | } |
|---|
| 174 | } |
|---|
| 175 | } |
|---|
| 176 | |
|---|
| 177 | /* |
|---|
| 178 | * (non-Javadoc) |
|---|
| 179 | * |
|---|
| 180 | * @see weka.classifiers.Classifier#buildClassifier(weka.core.Instances) |
|---|
| 181 | */ |
|---|
| 182 | @Override |
|---|
| 183 | public void buildClassifier(Instances traindata) throws Exception { |
|---|
| 184 | this.traindata = new Instances(traindata); |
|---|
| 185 | laserClassifier = setupClassifier(); |
|---|
| 186 | laserClassifier.buildClassifier(traindata); |
|---|
| 187 | } |
|---|
| 188 | } |
|---|
| 189 | } |
|---|