Package de.jungblut.math.sparse

Examples of de.jungblut.math.sparse.SparseDoubleRowMatrix

61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
        "Features and Outcomes need to match in length!");
    DoubleMatrix x = null;
    DoubleMatrix y = null;
    // add the bias
    if (features[0].isSparse()) {
      x = new SparseDoubleRowMatrix(DenseDoubleVector.ones(features.length),
          new SparseDoubleRowMatrix(features));
    } else {
      x = new DenseDoubleMatrix(DenseDoubleVector.ones(features.length),
          new DenseDoubleMatrix(features));
    }
    if (outcome[0].isSparse()) {
      y = new SparseDoubleRowMatrix(outcome);
    } else {
      y = new DenseDoubleMatrix(outcome);
    }
    // transpose y to get a faster lookup in the cost function
    y = y.transpose();
View Full Code Here
136
137
138
139
140
141
142
143
144
145
146
147
        DoubleVector[] outcomeSubArray = ArrayUtils.subArray(outcomeMatrix,
            start, end);
        outcomeMat = new DenseDoubleMatrix(outcomeSubArray);
      }
      DenseDoubleVector bias = DenseDoubleVector.ones(featureSubArray.length);
      DoubleMatrix featureMatrix = sparse ? new SparseDoubleRowMatrix(
          featureSubArray) : new DenseDoubleMatrix(featureSubArray);
      DoubleMatrix featuresWithBias = sparse ? new SparseDoubleRowMatrix(bias,
          featureMatrix) : new DenseDoubleMatrix(bias, featureMatrix);
      batches.add(new Tuple<>(featuresWithBias, outcomeMat));
    }
  }
View Full Code Here
48
49
50
51
52
53
54
   */
  public SparseMultiLabelRegression(int epochs, double alpha, int numFeatures,
      int numOutcomes) {
    this.epochs = epochs;
    this.alpha = alpha;
    this.weights = new SparseDoubleRowMatrix(numFeatures, numOutcomes);
  }
View Full Code Here
64
65
66
67
68
69
70
71
72
73
74
75
        DoubleVector feature = tuple.getFirst();
        DoubleVector outcome = tuple.getSecond();
        DoubleVector z1 = theta.multiplyVectorColumn(feature);
        DoubleVector activations = SIGMOID.apply(z1);
        double loss = LOSS.calculateError(
            new SparseDoubleRowMatrix(Arrays.asList(outcome)),
            new SparseDoubleRowMatrix(Arrays.asList(activations)));
        lossSum += loss;
        DoubleVector activationDifference = activations.subtract(outcome);
        // update theta by a smarter sparsity algorithm
        Iterator<DoubleVectorElement> featureIterator = feature
            .iterateNonZero();
View Full Code Here
77
78
79
80
81
82
83
84
85
86
87
    int numDistinctClasses = first.getSecond().getDimension();
    // respect the binary case
    numDistinctClasses = numDistinctClasses == 1 ? 2 : numDistinctClasses;
    // sparse row representations, so every class has the features as a hashset
    // of values. This gives good compression for many class problems.
    probabilityMatrix = new SparseDoubleRowMatrix(numDistinctClasses, first
        .getFirst().getDimension());

    int[] tokenPerClass = new int[numDistinctClasses];
    int[] numDocumentsPerClass = new int[numDistinctClasses];
View Full Code Here
31
32
33
34
35
36
37
   */
  private final DoubleMatrix transitionProbabilities;
  private final int numStates;

  private MarkovChain(int numStates) {
    this(numStates, new SparseDoubleRowMatrix(numStates, numStates));
  }
View Full Code Here
101
102
103
104
105
106
107
108
109
110
    return newInstance;
  }

  protected DoubleMatrix newInstance(DoubleMatrix mat) {
    if (mat.isSparse()) {
      return new SparseDoubleRowMatrix(mat.getRowCount(), mat.getColumnCount());
    } else {
      return new DenseDoubleMatrix(mat.getRowCount(), mat.getColumnCount());
    }
  }
View Full Code Here
131
132
133
134
135
136
137
138
   * @return a matrix containing the predicted hidden state on each row vector.
   */
  public DoubleMatrix decode(DoubleVector[] observationSequence,
      DoubleVector[] featuresPerHiddenState) {
    return ViterbiUtils.decode(emissionProbabilityMatrix,
        new SparseDoubleRowMatrix(observationSequence),
        new SparseDoubleRowMatrix(featuresPerHiddenState), numHiddenStates);
  }
View Full Code Here
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
   * @return a new sparse matrix from the stream.
   * @throws IOException in case of an IO error.
   */
  public static SparseDoubleRowMatrix readSparseMatrix(DataInput in)
      throws IOException {
    SparseDoubleRowMatrix mat = new SparseDoubleRowMatrix(in.readInt(),
        in.readInt());

    final int numRowIndices = in.readInt();
    for (int i = 0; i < numRowIndices; i++) {
      final int rowIndex = in.readInt();
      final int numColumns = in.readInt();
      DoubleVector row = new SparseDoubleVector(mat.getColumnCount());
      for (int j = 0; j < numColumns; j++) {
        row.set(in.readInt(), in.readDouble());
      }
      mat.setRowVector(rowIndex, row);
    }

    return mat;
  }
View Full Code Here
7
8
9
10
11
12
13
14
15
16
17
18
19
20
public class HammingLossFunctionTest {

  @Test
  public void testSingleHammingLoss() {
    SparseDoubleRowMatrix groundTruth = new SparseDoubleRowMatrix(
        new double[][] { { 1, 1, 1, 1 } });

    SparseDoubleRowMatrix hypo = new SparseDoubleRowMatrix(new double[][] { {
        0, 1, 1, 1 } });

    HammingLossFunction fnc = new HammingLossFunction(0.5d);

    double err = fnc.calculateError(groundTruth, hypo);
View Full Code Here
TOP

Related Classes of de.jungblut.math.sparse.SparseDoubleRowMatrix

  • de.jungblut.classification.bayes.MultinomialNaiveBayes
  • de.jungblut.classification.regression.LogisticRegression
  • de.jungblut.classification.regression.SparseMultiLabelRegression
  • de.jungblut.math.activation.AbstractActivationFunction
  • de.jungblut.math.minimize.AbstractMiniBatchCostFunction
  • de.jungblut.math.squashing.HammingLossFunctionTest
  • de.jungblut.ner.MaxEntMarkovModel
  • de.jungblut.nlp.HMM
  • de.jungblut.nlp.MarkovChain
  • de.jungblut.writable.MatrixWritable
    • Copyright © 2018 www.massapicom. All rights reserved.
    All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.