public LinearClassifier batchTrain(ArrayList<ClassificationInstance> trainingData) {
LinearClassifier w = new LinearClassifier(xAlphabet, yAlphabet, fxy);
LinearClassifier theta = null;
if (performAveraging) theta = new LinearClassifier(xAlphabet, yAlphabet, fxy);
for (int iter = 0; iter < numIterations; iter++) {
for (ClassificationInstance inst : trainingData) {
int yhat = w.label(inst.x);
if (yhat != inst.y) {
StaticUtils.plusEquals(w.w, fxy.apply(inst.x, inst.y));
StaticUtils.plusEquals(w.w, fxy.apply(inst.x, yhat), -1);
}
if (performAveraging) StaticUtils.plusEquals(theta.w, w.w, 1);
}
}
if (performAveraging) return theta;
return w;
}
public static void main(String[] args) throws IOException {
// read in the data.
ArrayList<ClassificationInstance> allData =
(new InternetAdReader(new Alphabet(), new Alphabet())).readFile(args[0]);
StaticUtils.shuffle(allData, 0);
// randomly split data into training and testing part
ArrayList<ClassificationInstance>[] tmp = StaticUtils.split(allData, 200);
ArrayList<ClassificationInstance> train = tmp[0];
ArrayList<ClassificationInstance> test = tmp[1];
Alphabet xA = allData.get(0).xAlphabet;
Alphabet yA = allData.get(0).yAlphabet;
System.out.println("num Features = " + allData.get(0).xAlphabet.size());
LinearClassifier h;
h = trainAdaBoost(50, train, xA, yA);
System.out.println("Boost Train Accuracy = " + StaticUtils.computeAccuracy(h, train));
System.out.println("Boost Test Accuracy = " + StaticUtils.computeAccuracy(h, test));
h = trainMaxEnt(train, xA, yA);
// print out accuracy
System.out.println("MaxEnt Train Accuracy = " + StaticUtils.computeAccuracy(h, train));
System.out.println("MaxEnt Test Accuracy = " + StaticUtils.computeAccuracy(h, test));
h = trainNaivBayes(train, xA, yA);
// print out accuracy
System.out.println("NaiveB Train Accuracy = " + StaticUtils.computeAccuracy(h, train));
System.out.println("NaiveB Test Accuracy = " + StaticUtils.computeAccuracy(h, test));
h = trainPerceptron(false, 20, train, xA, yA);
System.out.println("Percep Train Accuracy = " + StaticUtils.computeAccuracy(h, train));
System.out.println("Percep Test Accuracy = " + StaticUtils.computeAccuracy(h, test));
h = trainPerceptron(true, 20, train, xA, yA);
System.out.println("AvgPer Train Accuracy = " + StaticUtils.computeAccuracy(h, train));
System.out.println("AvgPer Test Accuracy = " + StaticUtils.computeAccuracy(h, test));
}