Package de.jungblut.math.dense

Examples of de.jungblut.math.dense.DenseDoubleVector

102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
      this.min = min;
      this.max = max;
    }

    public DoubleVector closestPoint(DoubleVector t) {
      DoubleVector p = new DenseDoubleVector(t.getDimension());
      for (int i = 0; i < t.getDimension(); ++i) {
        if (t.get(i) <= min.get(i)) {
          p.set(i, min.get(i));
        } else if (t.get(i) >= max.get(i)) {
          p.set(i, max.get(i));
        } else {
          p.set(i, t.get(i));
        }
      }
      return p;
    }
View Full Code Here
116
117
118
119
120
121
122
123
124
125
126
127
128
129
      }
      return p;
    }

    public static HyperRectangle infiniteHyperRectangle(int dimension) {
      DoubleVector min = new DenseDoubleVector(dimension);
      DoubleVector max = new DenseDoubleVector(dimension);
      for (int i = 0; i < dimension; ++i) {
        min.set(i, Double.NEGATIVE_INFINITY);
        max.set(i, Double.POSITIVE_INFINITY);
      }

      return new HyperRectangle(min, max);
    }
View Full Code Here
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
   * @return the new n-dimensional dense vector vectorized via the hashing
   *         trick.
   */
  public static DoubleVector hashVectorize(DoubleVector inputFeature, int n,
      com.google.common.hash.HashFunction hashFunction) {
    DoubleVector dense = new DenseDoubleVector(n);
    Iterator<DoubleVectorElement> iterateNonZero = inputFeature
        .iterateNonZero();
    while (iterateNonZero.hasNext()) {
      DoubleVectorElement next = iterateNonZero.next();
      // get the hash as int
      int hash = hashFunction.hashInt(next.getIndex()).asInt();
      // abs it, as we don't want to have negative index access
      int bucket = Math.abs(hash) % n;
      // subtract 1 in case of negative values, else increment.
      // this replaces the second hash function proposed by Weinberger et al.
      dense.set(bucket, dense.get(bucket) + (hash < 0 ? -1d : 1d));
    }
    return dense;
  }
View Full Code Here
112
113
114
115
116
117
118
119
120
121
122
123
124
      // let's assume the default case ("negative") here, instead of making NPEs
      // in other areas, as the callers aren't nullsafe
      clz = 0;
    }
    if (binaryClassification) {
      return new DenseDoubleVector(new double[] { clz });
    } else {
      DoubleVector vec = outcomeDimension > 10 ? new SparseDoubleVector(
          outcomeDimension) : new DenseDoubleVector(outcomeDimension);
      vec.set(clz, 1);
      return vec;
    }
  }
View Full Code Here
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
    for (int i = 0; i < classifier.length; i++) {
      result[i] = classifier[i].predict(features);
    }
    int numPossibleOutcomes = result[0].getDimension() == 1 ? 2 : result[0]
        .getDimension();
    DoubleVector toReturn = new DenseDoubleVector(
        result[0].getDimension() == 1 ? 1 : numPossibleOutcomes);
    // now combine the results based on the rule
    switch (type) {
      case MAJORITY:
        double[] histogram = createPredictionHistogram(result,
            numPossibleOutcomes);
        if (numPossibleOutcomes == 2) {
          toReturn.set(0, ArrayUtils.maxIndex(histogram));
        } else {
          toReturn.set(ArrayUtils.maxIndex(histogram), 1d);
        }
        break;
      case PROBABILITY:
        histogram = createPredictionHistogram(result, numPossibleOutcomes);
        double histSum = 0;
        for (double d : histogram) {
          histSum += d;
        }
        if (numPossibleOutcomes == 2) {
          toReturn.set(0, histogram[1] / histSum);
        } else {
          for (int i = 0; i < histogram.length; i++) {
            toReturn.set(i, histogram[i] / histSum);
          }
        }
        break;
      case AVERAGE:
        for (int i = 0; i < result.length; i++) {
          toReturn = toReturn.add(result[i]);
        }
        toReturn = toReturn.divide(classifier.length);
        break;
      default:
        throw new UnsupportedOperationException("Type " + type
            + " isn't supported yet!");
    }
View Full Code Here
116
117
118
119
120
121
122
123
124
125
126
        System.out
            .println("Computed " + row + " / " + numDistinctClasses + "!");
      }
    }

    classPriorProbability = new DenseDoubleVector(numDistinctClasses);
    for (int i = 0; i < numDistinctClasses; i++) {
      double prior = FastMath.log(numDocumentsPerClass[i])
          - FastMath.log(numDocumentsSeen);
      classPriorProbability.set(i, prior);
    }
View Full Code Here
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
  }

  private DenseDoubleVector getProbabilityDistribution(DoubleVector document) {

    int numClasses = classPriorProbability.getLength();
    DenseDoubleVector distribution = new DenseDoubleVector(numClasses);
    // loop through all classes and get the max probable one
    for (int i = 0; i < numClasses; i++) {
      double probability = getProbabilityForClass(document, i);
      distribution.set(i, probability);
    }

    double maxProbability = distribution.max();
    double probabilitySum = 0.0d;
    // we normalize it back
    for (int i = 0; i < numClasses; i++) {
      double probability = distribution.get(i);
      double normalizedProbability = FastMath.exp(probability - maxProbability
          + classPriorProbability.get(i));
      distribution.set(i, normalizedProbability);
      probabilitySum += normalizedProbability;
    }

    // since the sum is sometimes not 1, we need to divide by the sum
    distribution = (DenseDoubleVector) distribution.divide(probabilitySum);

    return distribution;
  }
View Full Code Here
307
308
309
310
311
312
313
314
  }

  private DoubleVector computeHiddenActivations(DoubleVector input,
      DoubleMatrix theta) {
    // add the bias to the input
    DoubleVector biased = new DenseDoubleVector(1d, input.toArray());
    return activationFunction.apply(theta.multiplyVectorRow(biased));
  }
View Full Code Here
110
111
112
113
114
115
116
117
118
119
120
121
122
  /**
   * @return the transition probabilities for the states.
   */
  public DoubleVector getTransitionProbabilities(int[] stateSequence) {
    DenseDoubleVector distribution = new DenseDoubleVector(
        stateSequence.length - 1);
    for (int i = 0; i < distribution.getDimension(); i++) {
      distribution.set(i,
          transitionProbabilities.get(stateSequence[i], stateSequence[i + 1]));
    }
    return distribution;
  }
View Full Code Here
59
60
61
62
63
64
65
66
        }
      }
      nameMapping[classIndex++] = classDirString;
    }

    return new Tuple3<>(docList, new DenseDoubleVector(prediction.toArray()),
        nameMapping);
  }
View Full Code Here
TOP

Related Classes of de.jungblut.math.dense.DenseDoubleVector

  • de.jungblut.classification.bayes.MultinomialNaiveBayes
  • de.jungblut.classification.bayes.MultinomialNaiveBayesTest
  • de.jungblut.classification.eval.EvaluationSplitTest
  • de.jungblut.classification.knn.AbstractKNearestNeighbours
  • de.jungblut.classification.knn.KNearestNeighboursTest
  • de.jungblut.classification.meta.Voter
  • de.jungblut.classification.nn.MultilayerPerceptron
  • de.jungblut.classification.nn.MultiLayerPerceptronTest
  • de.jungblut.classification.nn.RBM
  • de.jungblut.classification.nn.RBMCostFunctionTest
    • 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.