001 /*
002 * Created on Aug 6, 2005
003 *
004 */
005 package aima.learning.demos;
006
007 import java.util.ArrayList;
008 import java.util.List;
009
010 import aima.learning.framework.DataSet;
011 import aima.learning.framework.DataSetFactory;
012 import aima.learning.framework.Learner;
013 import aima.learning.inductive.DLTestFactory;
014 import aima.learning.inductive.DecisionTree;
015 import aima.learning.learners.AdaBoostLearner;
016 import aima.learning.learners.DecisionListLearner;
017 import aima.learning.learners.DecisionTreeLearner;
018 import aima.learning.learners.StumpLearner;
019 import aima.learning.neural.BackPropLearning;
020 import aima.learning.neural.FeedForwardNeuralNetwork;
021 import aima.learning.neural.IrisDataSetNumerizer;
022 import aima.learning.neural.IrisNNDataSet;
023 import aima.learning.neural.NNConfig;
024 import aima.learning.neural.NNDataSet;
025 import aima.learning.neural.Numerizer;
026 import aima.learning.neural.Perceptron;
027 import aima.util.Util;
028
029 public class LearningDemo {
030 public static void main(String[] args) {
031
032 // For Reinforcement Learning Demos see Probability Demo
033 decisionTreeDemo();
034 decisionListDemo();
035 ensembleLearningDemo();
036 perceptronDemo();
037 backPropogationDemo();
038 }
039
040 private static void decisionTreeDemo() {
041 System.out.println(Util.ntimes("*", 100));
042 System.out
043 .println("\nDecisionTree Demo - Inducing a DecisionList from the Restaurant DataSet\n ");
044 System.out.println(Util.ntimes("*", 100));
045 try {
046 DataSet ds = DataSetFactory.getRestaurantDataSet();
047 DecisionTreeLearner learner = new DecisionTreeLearner();
048 learner.train(ds);
049 System.out.println("The Induced Decision Tree is ");
050 System.out.println(learner.getDecisionTree());
051 int[] result = learner.test(ds);
052
053 System.out
054 .println("\nThis Decision Tree classifies the data set with "
055 + result[0]
056 + " successes"
057 + " and "
058 + result[1]
059 + " failures");
060 System.out.println("\n");
061 } catch (Exception e) {
062 System.out.println("Decision Tree Demo Failed ");
063 e.printStackTrace();
064 }
065
066 }
067
068 private static void decisionListDemo() {
069 try {
070 System.out.println(Util.ntimes("*", 100));
071 System.out
072 .println("DecisionList Demo - Inducing a DecisionList from the Restaurant DataSet\n ");
073 System.out.println(Util.ntimes("*", 100));
074 DataSet ds = DataSetFactory.getRestaurantDataSet();
075 DecisionListLearner learner = new DecisionListLearner("Yes", "No",
076 new DLTestFactory());
077 learner.train(ds);
078 System.out.println("The Induced DecisionList is");
079 System.out.println(learner.getDecisionList());
080 int[] result = learner.test(ds);
081
082 System.out
083 .println("\nThis Decision List classifies the data set with "
084 + result[0]
085 + " successes"
086 + " and "
087 + result[1]
088 + " failures");
089 System.out.println("\n");
090
091 } catch (Exception e) {
092 System.out.println("Decision ListDemo Failed");
093 }
094 }
095
096 private static void ensembleLearningDemo() {
097 System.out.println(Util.ntimes("*", 100));
098 System.out
099 .println("\n Ensemble Decision Demo - Weak Learners co operating to give Superior decisions ");
100 System.out.println(Util.ntimes("*", 100));
101 try {
102 DataSet ds = DataSetFactory.getRestaurantDataSet();
103 List stumps = DecisionTree.getStumpsFor(ds, "Yes", "No");
104 List<Learner> learners = new ArrayList<Learner>();
105
106 System.out
107 .println("\nStump Learners vote to decide in this algorithm");
108 for (Object stump : stumps) {
109 DecisionTree sl = (DecisionTree) stump;
110 StumpLearner stumpLearner = new StumpLearner(sl, "No");
111 learners.add(stumpLearner);
112 }
113 AdaBoostLearner learner = new AdaBoostLearner(learners, ds);
114 learner.train(ds);
115 int[] result = learner.test(ds);
116 System.out
117 .println("\nThis Ensemble Learner classifies the data set with "
118 + result[0]
119 + " successes"
120 + " and "
121 + result[1]
122 + " failures");
123 System.out.println("\n");
124
125 } catch (Exception e) {
126
127 }
128
129 }
130
131 private static void perceptronDemo() {
132 try {
133 System.out.println(Util.ntimes("*", 100));
134 System.out
135 .println("\n Perceptron Demo - Running Perceptron on Iris data Set with 10 epochs of learning ");
136 System.out.println(Util.ntimes("*", 100));
137 DataSet irisDataSet = DataSetFactory.getIrisDataSet();
138 Numerizer numerizer = new IrisDataSetNumerizer();
139 NNDataSet innds = new IrisNNDataSet();
140
141 innds.createExamplesFromDataSet(irisDataSet, numerizer);
142
143 Perceptron perc = new Perceptron(3, 4);
144
145 perc.trainOn(innds, 10);
146
147 innds.refreshDataset();
148 int[] result = perc.testOnDataSet(innds);
149 System.out.println(result[0] + " right, " + result[1] + " wrong");
150 } catch (Exception e) {
151 // TODO Auto-generated catch block
152 e.printStackTrace();
153 }
154
155 }
156
157 private static void backPropogationDemo() {
158 try {
159 System.out.println(Util.ntimes("*", 100));
160 System.out
161 .println("\n BackpropagationDemo - Running BackProp on Iris data Set with 10 epochs of learning ");
162 System.out.println(Util.ntimes("*", 100));
163
164 DataSet irisDataSet = DataSetFactory.getIrisDataSet();
165 Numerizer numerizer = new IrisDataSetNumerizer();
166 NNDataSet innds = new IrisNNDataSet();
167
168 innds.createExamplesFromDataSet(irisDataSet, numerizer);
169
170 NNConfig config = new NNConfig();
171 config.setConfig(FeedForwardNeuralNetwork.NUMBER_OF_INPUTS, 4);
172 config.setConfig(FeedForwardNeuralNetwork.NUMBER_OF_OUTPUTS, 3);
173 config.setConfig(FeedForwardNeuralNetwork.NUMBER_OF_HIDDEN_NEURONS,
174 6);
175 config
176 .setConfig(FeedForwardNeuralNetwork.LOWER_LIMIT_WEIGHTS,
177 -2.0);
178 config.setConfig(FeedForwardNeuralNetwork.UPPER_LIMIT_WEIGHTS, 2.0);
179
180 FeedForwardNeuralNetwork ffnn = new FeedForwardNeuralNetwork(config);
181 ffnn.setTrainingScheme(new BackPropLearning(0.1, 0.9));
182
183 ffnn.trainOn(innds, 10);
184
185 innds.refreshDataset();
186 int[] result = ffnn.testOnDataSet(innds);
187 System.out.println(result[0] + " right, " + result[1] + " wrong");
188 } catch (Exception e) {
189 // TODO Auto-generated catch block
190 e.printStackTrace();
191 }
192 }
193
194 }