@Override
public double[][] predict(List<PredictionPaper> testDocs) {
String testData = "lda/test.dat";
createLdaInputTest(testData, testDocs);
Utils.runCommand(
"lib/lda-c-dist/lda inf "
+ " lib/lda-c-dist/settings.txt "
+ "lda/final "
+ testData
+ " lda/output",
false);
double[][] gammasMatrix = Utils.readMatrix("lda/output-gamma.dat", false);
double alpha = Utils.readAlpha("lda/final.other");
for (int i = 0; i < gammasMatrix.length; i++) {
for (int j = 0; j < gammasMatrix[i].length; j++) {
gammasMatrix[i][j] -= alpha;
}
}
SimpleMatrix gammas = new SimpleMatrix(gammasMatrix);
SimpleMatrix beta = new SimpleMatrix(betaMatrix);
SimpleMatrix probabilities = gammas.mult(beta);
double[][] result = new double[probabilities.numRows()][probabilities.numCols()];
for (int row = 0; row < probabilities.numRows(); row++) {
for (int col = 0; col < probabilities.numCols(); col++) {
result[row][col] = probabilities.get(row, col);
}
}
return result;
}
public static List<Cluster> ReassignCentrids(
List<Cluster> kCentroids,
SimpleMatrix distanceMatrix,
SimpleMatrix dataSet,
int[] featureSet) {
List<Cluster> kCentroids_l = kCentroids;
int[] clusterLoc = new int[dataSet.numRows()];
for (int iRows = 0; iRows < dataSet.numRows(); iRows++) {
int clusterNo = 1;
double minvalue = distanceMatrix.get(iRows, 1);
for (int iCentroid = 0; iCentroid < kCentroids_l.size(); iCentroid++) {
// System.out.println(iRows+" "+iCentroid);
if (distanceMatrix.get(iRows, iCentroid) < minvalue) {
clusterNo = iCentroid;
minvalue = distanceMatrix.get(iRows, iCentroid);
}
}
clusterLoc[iRows] = clusterNo;
}
// Backup Centroids anc clear current centroids
for (int i = 0; i < kCentroids_l.size(); i++) {
kCentroids_l.get(i).backup();
kCentroids_l.get(i).noPoints = 0;
Arrays.fill(kCentroids_l.get(i).currPoints, -1);
Arrays.fill(kCentroids_l.get(i).intIndex, -1);
// System.out.println("Clusters Backed Up!");
}
// printKCentroids(kCentroids_l);
for (int i = 0; i < clusterLoc.length; i++) {
int insLoc = kCentroids_l.get(clusterLoc[i]).noPoints;
// System.out.println("Getting element"+dataSet.get(i, 0));
kCentroids_l.get(clusterLoc[i]).currPoints[insLoc] = (int) dataSet.get(i, 0);
kCentroids_l.get(clusterLoc[i]).intIndex[insLoc] = i;
kCentroids_l.get(clusterLoc[i]).noPoints++;
}
// System.out.println(Arrays.toString(clusterLoc));
// Now Calculate the best representative and give as new centroid values
for (int i = 0; i < kCentroids.size(); i++) {
for (int j = 0; j < featureSet.length; j++) {
double tempAvg = 0;
int[] travVector = kCentroids_l.get(i).intIndex;
for (int k = 0; k < travVector.length; k++) {
if (travVector[k] != -1) {
tempAvg = tempAvg + dataSet.get(travVector[k], featureSet[j]);
}
}
kCentroids.get(i).clusterCenter.set(0, featureSet[j], tempAvg / kCentroids.get(i).noPoints);
}
}
return kCentroids_l;
}
/**
* Finds the values of a,b,c which minimize
*
* <p>sum (a*x(+)_i + b*y(+)_i + c - x(-)_i)^2
*
* <p>See page 306
*
* @return Affine transform
*/
private SimpleMatrix computeAffineH(
List<AssociatedPair> observations, DenseMatrix64F H, DenseMatrix64F Hzero) {
SimpleMatrix A = new SimpleMatrix(observations.size(), 3);
SimpleMatrix b = new SimpleMatrix(A.numRows(), 1);
Point2D_F64 c = new Point2D_F64();
Point2D_F64 k = new Point2D_F64();
for (int i = 0; i < observations.size(); i++) {
AssociatedPair a = observations.get(i);
GeometryMath_F64.mult(Hzero, a.p1, k);
GeometryMath_F64.mult(H, a.p2, c);
A.setRow(i, 0, k.x, k.y, 1);
b.set(i, 0, c.x);
}
SimpleMatrix x = A.solve(b);
SimpleMatrix Ha = SimpleMatrix.identity(3);
Ha.setRow(0, 0, x.getMatrix().data);
return Ha;
}
// ecuatia Rosenbrock
// x(i) is coded with 8 bits
public double fitness(SimpleMatrix cromosom) {
int value = 0;
int n = cromosom.numRows();
double x1, x2;
for (int i = 0; i < n - 1; i++) {
x1 = cromosom.get(i);
x2 = cromosom.get(i + 1);
value += (100 * Math.pow(x2 - Math.pow(x1, 2), 2) + Math.pow(x1 - 1, 2));
}
return value;
}
public static SimpleMatrix compDist(
List<Cluster> kCentroids, SimpleMatrix dataSet, int[] featureSet, String distanceMetric) {
int dRows = dataSet.numRows();
int dCols = kCentroids.size();
int[] features = featureSet;
SimpleMatrix distMatrix = new SimpleMatrix(dRows, dCols);
for (int iCentroid = 0; iCentroid < dCols; iCentroid++) {
Cluster kcenter = kCentroids.get(iCentroid);
for (int iRows = 0; iRows < dRows; iRows++) {
double distTemp = 0;
for (int iFeature = 0; iFeature < features.length; iFeature++) {
double cX = kcenter.clusterCenter.get(0, features[iFeature]);
double dX = dataSet.get(iRows, features[iFeature]);
distTemp = distTemp + Math.pow(cX - dX, 2);
}
distMatrix.set(iRows, iCentroid, Math.sqrt(distTemp));
}
}
return distMatrix;
}
public static void main(String[] args) {
try {
if (args.length < 5) {
System.out.println(
"Invalid Syntax usage. \n The Syntax is KMeans inputfile distance-metric #Centroids #Iterations Tolerance featureSet crossValidation testDataSet");
}
String inputFile = args[0];
String distanceMetric = args[1];
int centroids = Integer.parseInt(args[2]);
int iterations = Integer.parseInt(args[3]);
double tolerance = Double.parseDouble(args[4]);
String inpFeatures = args[5];
String crossValidation = args[6];
String testFile = args[7];
String[] temp = inpFeatures.split(",");
int[] featureSet = new int[temp.length];
boolean hasConvered = false;
for (int i = 0; i < temp.length; i++) {
featureSet[i] = Integer.parseInt(temp[i]);
}
System.out.println("Features Considered are" + Arrays.toString(featureSet));
SimpleMatrix dataSet = new SimpleMatrix().loadCSV(inputFile);
// Cluster Parameters
int dsRows = dataSet.numRows();
int dsCol = dataSet.numCols();
// Cluster Initialization
List<Cluster> kcentroids = new ArrayList<Cluster>();
for (int i = 0; i < centroids; i++) {
int random = genRandom(0, dsRows - 1);
int[] currPoints = new int[dsRows];
int[] intIndex = new int[dsRows];
// This will have remnance of the first chosen element
Random r = new Random();
SimpleMatrix t = new SimpleMatrix().random(1, dsCol, 0, 0, r);
Cluster centers =
new Cluster(i, dataSet.extractVector(true, random), t, currPoints, intIndex);
kcentroids.add(centers);
}
// printKCentroids(kcentroids);
// dataSet.print();
SimpleMatrix distMatrix = compDist(kcentroids, dataSet, featureSet, distanceMetric);
// distMatrix.print();
// System.out.println(distMatrix.get(0, 1));
// ReassignCentrids(kcentroids, distMatrix, dataSet,featureSet);
// printKCentroids(kcentroids);
for (int k = 0; k < iterations; k++) {
System.out.println("------------------------Iteration " + k + " ------------------------");
distMatrix = compDist(kcentroids, dataSet, featureSet, distanceMetric);
// distMatrix.print();
ReassignCentrids(kcentroids, distMatrix, dataSet, featureSet);
printKCentroids(kcentroids);
hasConvered = detectConvergence(kcentroids, featureSet, tolerance);
// System.out.println(hasConvered);
if (hasConvered) {
System.out.println(
"------------------------Has Converged : Tolerance------------------------");
break;
}
}
if (!hasConvered)
System.out.println(
"------------------------Has Converged : Iterations------------------------");
if (crossValidation.equals("true")) {
SimpleMatrix test = new SimpleMatrix().loadCSV(testFile);
SimpleMatrix dist = compDist(kcentroids, test, featureSet, distanceMetric);
System.out.println("----------Associating Set Data Sets to Calculated Centroid-------");
// dist.print();
List<Cluster> kCentroids_l = kcentroids;
int[] clusterLoc = new int[test.numRows()];
for (int iRows = 0; iRows < test.numRows(); iRows++) {
int clusterNo = 1;
double minvalue = dist.get(iRows, 1);
for (int iCentroid = 0; iCentroid < kCentroids_l.size(); iCentroid++) {
// System.out.println(iRows+" "+iCentroid);
if (dist.get(iRows, iCentroid) < minvalue) {
clusterNo = iCentroid;
minvalue = dist.get(iRows, iCentroid);
}
}
clusterLoc[iRows] = clusterNo;
}
// System.out.println(Arrays.toString(clusterLoc));
for (int i = 0; i < kCentroids_l.size(); i++) {
kCentroids_l.get(i).backup();
kCentroids_l.get(i).noPoints = 0;
Arrays.fill(kCentroids_l.get(i).currPoints, -1);
Arrays.fill(kCentroids_l.get(i).intIndex, -1);
// System.out.println("Clusters Backed Up!");
}
// printKCentroids(kCentroids_l);
for (int i = 0; i < clusterLoc.length; i++) {
int insLoc = kCentroids_l.get(clusterLoc[i]).noPoints;
// System.out.println("Getting element"+dataSet.get(i, 0));
kCentroids_l.get(clusterLoc[i]).currPoints[insLoc] = (int) test.get(i, 0);
kCentroids_l.get(clusterLoc[i]).intIndex[insLoc] = i;
kCentroids_l.get(clusterLoc[i]).noPoints++;
}
printKCentroids(kCentroids_l);
}
} catch (Exception e) {
System.out.println("Unfortunately There is an error.");
e.printStackTrace();
}
}
public void backpropDerivative(
Tree tree,
List<String> words,
IdentityHashMap<Tree, SimpleMatrix> nodeVectors,
TwoDimensionalMap<String, String, SimpleMatrix> binaryW_dfs,
Map<String, SimpleMatrix> unaryW_dfs,
TwoDimensionalMap<String, String, SimpleMatrix> binaryScoreDerivatives,
Map<String, SimpleMatrix> unaryScoreDerivatives,
Map<String, SimpleMatrix> wordVectorDerivatives,
SimpleMatrix deltaUp) {
if (tree.isLeaf()) {
return;
}
if (tree.isPreTerminal()) {
if (op.trainOptions.trainWordVectors) {
String word = tree.children()[0].label().value();
word = dvModel.getVocabWord(word);
// SimpleMatrix currentVector = nodeVectors.get(tree);
// SimpleMatrix currentVectorDerivative =
// nonlinearityVectorToDerivative(currentVector);
// SimpleMatrix derivative = deltaUp.elementMult(currentVectorDerivative);
SimpleMatrix derivative = deltaUp;
wordVectorDerivatives.put(word, wordVectorDerivatives.get(word).plus(derivative));
}
return;
}
SimpleMatrix currentVector = nodeVectors.get(tree);
SimpleMatrix currentVectorDerivative =
NeuralUtils.elementwiseApplyTanhDerivative(currentVector);
SimpleMatrix scoreW = dvModel.getScoreWForNode(tree);
currentVectorDerivative = currentVectorDerivative.elementMult(scoreW.transpose());
// the delta that is used at the current nodes
SimpleMatrix deltaCurrent = deltaUp.plus(currentVectorDerivative);
SimpleMatrix W = dvModel.getWForNode(tree);
SimpleMatrix WTdelta = W.transpose().mult(deltaCurrent);
if (tree.children().length == 2) {
// TODO: RS: Change to the nice "getWForNode" setup?
String leftLabel = dvModel.basicCategory(tree.children()[0].label().value());
String rightLabel = dvModel.basicCategory(tree.children()[1].label().value());
binaryScoreDerivatives.put(
leftLabel,
rightLabel,
binaryScoreDerivatives.get(leftLabel, rightLabel).plus(currentVector.transpose()));
SimpleMatrix leftVector = nodeVectors.get(tree.children()[0]);
SimpleMatrix rightVector = nodeVectors.get(tree.children()[1]);
SimpleMatrix childrenVector = NeuralUtils.concatenateWithBias(leftVector, rightVector);
if (op.trainOptions.useContextWords) {
childrenVector = concatenateContextWords(childrenVector, tree.getSpan(), words);
}
SimpleMatrix W_df = deltaCurrent.mult(childrenVector.transpose());
binaryW_dfs.put(leftLabel, rightLabel, binaryW_dfs.get(leftLabel, rightLabel).plus(W_df));
// and then recurse
SimpleMatrix leftDerivative = NeuralUtils.elementwiseApplyTanhDerivative(leftVector);
SimpleMatrix rightDerivative = NeuralUtils.elementwiseApplyTanhDerivative(rightVector);
SimpleMatrix leftWTDelta = WTdelta.extractMatrix(0, deltaCurrent.numRows(), 0, 1);
SimpleMatrix rightWTDelta =
WTdelta.extractMatrix(deltaCurrent.numRows(), deltaCurrent.numRows() * 2, 0, 1);
backpropDerivative(
tree.children()[0],
words,
nodeVectors,
binaryW_dfs,
unaryW_dfs,
binaryScoreDerivatives,
unaryScoreDerivatives,
wordVectorDerivatives,
leftDerivative.elementMult(leftWTDelta));
backpropDerivative(
tree.children()[1],
words,
nodeVectors,
binaryW_dfs,
unaryW_dfs,
binaryScoreDerivatives,
unaryScoreDerivatives,
wordVectorDerivatives,
rightDerivative.elementMult(rightWTDelta));
} else if (tree.children().length == 1) {
String childLabel = dvModel.basicCategory(tree.children()[0].label().value());
unaryScoreDerivatives.put(
childLabel, unaryScoreDerivatives.get(childLabel).plus(currentVector.transpose()));
SimpleMatrix childVector = nodeVectors.get(tree.children()[0]);
SimpleMatrix childVectorWithBias = NeuralUtils.concatenateWithBias(childVector);
if (op.trainOptions.useContextWords) {
childVectorWithBias = concatenateContextWords(childVectorWithBias, tree.getSpan(), words);
}
SimpleMatrix W_df = deltaCurrent.mult(childVectorWithBias.transpose());
// System.out.println("unary backprop derivative for " + childLabel);
// System.out.println("Old transform:");
// System.out.println(unaryW_dfs.get(childLabel));
// System.out.println(" Delta:");
// System.out.println(W_df.scale(scale));
unaryW_dfs.put(childLabel, unaryW_dfs.get(childLabel).plus(W_df));
// and then recurse
SimpleMatrix childDerivative = NeuralUtils.elementwiseApplyTanhDerivative(childVector);
// SimpleMatrix childDerivative = childVector;
SimpleMatrix childWTDelta = WTdelta.extractMatrix(0, deltaCurrent.numRows(), 0, 1);
backpropDerivative(
tree.children()[0],
words,
nodeVectors,
binaryW_dfs,
unaryW_dfs,
binaryScoreDerivatives,
unaryScoreDerivatives,
wordVectorDerivatives,
childDerivative.elementMult(childWTDelta));
}
}
