private static void ensembleLearningDemo() {
System.out.println(Util.ntimes("*", 100));
System.out.println(
"\n Ensemble Decision Demo - Weak Learners co operating to give Superior decisions ");
System.out.println(Util.ntimes("*", 100));
try {
DataSet ds = DataSetFactory.getRestaurantDataSet();
List stumps = DecisionTree.getStumpsFor(ds, "Yes", "No");
List<Learner> learners = new ArrayList<Learner>();
System.out.println("\nStump Learners vote to decide in this algorithm");
for (Object stump : stumps) {
DecisionTree sl = (DecisionTree) stump;
StumpLearner stumpLearner = new StumpLearner(sl, "No");
learners.add(stumpLearner);
}
AdaBoostLearner learner = new AdaBoostLearner(learners, ds);
learner.train(ds);
int[] result = learner.test(ds);
System.out.println(
"\nThis Ensemble Learner classifies the data set with "
+ result[0]
+ " successes"
+ " and "
+ result[1]
+ " failures");
System.out.println("\n");
} catch (Exception e) {
}
}
private static void decisionListDemo() {
try {
System.out.println(Util.ntimes("*", 100));
System.out.println(
"DecisionList Demo - Inducing a DecisionList from the Restaurant DataSet\n ");
System.out.println(Util.ntimes("*", 100));
DataSet ds = DataSetFactory.getRestaurantDataSet();
DecisionListLearner learner = new DecisionListLearner("Yes", "No", new DLTestFactory());
learner.train(ds);
System.out.println("The Induced DecisionList is");
System.out.println(learner.getDecisionList());
int[] result = learner.test(ds);
System.out.println(
"\nThis Decision List classifies the data set with "
+ result[0]
+ " successes"
+ " and "
+ result[1]
+ " failures");
System.out.println("\n");
} catch (Exception e) {
System.out.println("Decision ListDemo Failed");
}
}
private static void backPropogationDemo() {
try {
System.out.println(Util.ntimes("*", 100));
System.out.println(
"\n BackpropagationDemo - Running BackProp on Iris data Set with 10 epochs of learning ");
System.out.println(Util.ntimes("*", 100));
DataSet irisDataSet = DataSetFactory.getIrisDataSet();
Numerizer numerizer = new IrisDataSetNumerizer();
NNDataSet innds = new IrisNNDataSet();
innds.createExamplesFromDataSet(irisDataSet, numerizer);
NNConfig config = new NNConfig();
config.setConfig(FeedForwardNeuralNetwork.NUMBER_OF_INPUTS, 4);
config.setConfig(FeedForwardNeuralNetwork.NUMBER_OF_OUTPUTS, 3);
config.setConfig(FeedForwardNeuralNetwork.NUMBER_OF_HIDDEN_NEURONS, 6);
config.setConfig(FeedForwardNeuralNetwork.LOWER_LIMIT_WEIGHTS, -2.0);
config.setConfig(FeedForwardNeuralNetwork.UPPER_LIMIT_WEIGHTS, 2.0);
FeedForwardNeuralNetwork ffnn = new FeedForwardNeuralNetwork(config);
ffnn.setTrainingScheme(new BackPropLearning(0.1, 0.9));
ffnn.trainOn(innds, 10);
innds.refreshDataset();
int[] result = ffnn.testOnDataSet(innds);
System.out.println(result[0] + " right, " + result[1] + " wrong");
} catch (Exception e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
private static void perceptronDemo() {
try {
System.out.println(Util.ntimes("*", 100));
System.out.println(
"\n Perceptron Demo - Running Perceptron on Iris data Set with 10 epochs of learning ");
System.out.println(Util.ntimes("*", 100));
DataSet irisDataSet = DataSetFactory.getIrisDataSet();
Numerizer numerizer = new IrisDataSetNumerizer();
NNDataSet innds = new IrisNNDataSet();
innds.createExamplesFromDataSet(irisDataSet, numerizer);
Perceptron perc = new Perceptron(3, 4);
perc.trainOn(innds, 10);
innds.refreshDataset();
int[] result = perc.testOnDataSet(innds);
System.out.println(result[0] + " right, " + result[1] + " wrong");
} catch (Exception e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
// function SIMULATED-ANNEALING(problem, schedule) returns a solution state
// inputs: problem, a problem
// schedule, a mapping from time to "temperature"
public List<String> search(Problem p) throws Exception {
// local variables: current, a node
// next, a node
// T, a "temperature" controlling the probability of downward steps
clearInstrumentation();
outcome = SearchOutcome.FAILURE;
lastState = null;
// current <- MAKE-NODE(INITIAL-STATE[problem])
Node current = new Node(p.getInitialState());
Node next = null;
List<String> ret = new ArrayList<String>();
// for t <- 1 to INFINITY do
int timeStep = 0;
while (true) {
// temperature <- schedule[t]
double temperature = scheduler.getTemp(timeStep);
timeStep++;
// if temperature = 0 then return current
if (temperature == 0.0) {
if (p.isGoalState(current.getState())) {
outcome = SearchOutcome.SOLUTION_FOUND;
}
ret = SearchUtils.actionsFromNodes(current.getPathFromRoot());
lastState = current.getState();
break;
}
List<Node> children = expandNode(current, p);
if (children.size() > 0) {
// next <- a randomly selected successor of current
next = Util.selectRandomlyFromList(children);
// /\E <- VALUE[next] - VALUE[current]
int deltaE = getValue(p, next) - getValue(p, current);
if (shouldAccept(temperature, deltaE)) {
current = next;
}
}
}
return ret;
}
