Package edu.ucla.sspace.vector

Examples of edu.ucla.sspace.vector.DenseVector

190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
   
    private static IntPair pickFirstTwo(Matrix dataPoints,
                                        SimilarityFunction simFunc,
                                        int[] weights, double[] inverseSimilarities) {
  double OPT_1 = 0; // optimal 1-means cost.
  DoubleVector centerOfMass = new DenseVector(dataPoints.columns());
  double sum = 0d;
  int rows = dataPoints.rows();
        int cols = dataPoints.columns();
        double[] probs = new double[rows];
  int totalWeight = 0;

  for (int i = 0; i < rows; i++) {
            DoubleVector v = dataPoints.getRowVector(i);
            int weight = weights[i];
            // Update the center of mass for the entire solution based
            VectorMath.add(centerOfMass, new ScaledDoubleVector(v, weight));
            totalWeight += weight;
        }
        
        // Then rescale the center of mass based on the total weight
        for (int j = 0; j < cols; j++)
            centerOfMass.set(j, centerOfMass.get(j) / totalWeight);
       
       
        for (int i = 0; i < rows; i++) {
            double sim = simFunc.sim(centerOfMass, dataPoints.getRowVector(i));
            sim = invertSim(sim);
View Full Code Here
140
141
142
143
144
145
146
147
148
149
150
        clusterSizes = new int[numClusters];
        costs = new double[numClusters];

        // Initialize the clusters.
        for (int c = 0; c < numClusters; ++c)
            centroids[c] = new DenseVector(m.columns());

        // Form the cluster composite vectors, i.e. unscaled centroids.
        for (int i = 0; i < m.rows(); ++i) {
            int assignment = initialAssignments[i];
            VectorMath.add(centroids[assignment], matrix.get(i));
View Full Code Here
53
54
55
56
57
58
59
60
61
62
63
            // For any extra slots that do not have a data point, use an empty
            // vector, which will have zero similarity with any other data
            // point.
            for (int i = dataPoints.rows(); i < numCentroids; ++i)
                centers[i] = new DenseVector(dataPoints.columns());

            return centers;
        }

        // Select a subset of data points to be the new centroids.
View Full Code Here
75
76
77
78
79
80
81
82
83
84
85
        DoubleVector[] centroids = new DoubleVector[numCentroids];

        // Compute the centroid assuming that there is only one cluster to be
        // found.  This is required for computing the optimal cost of a single
        // cluster solution.
        DoubleVector singleCentroid = new DenseVector(dataPoints.columns());
        for (int i = 0; i < numDataPoints; ++i)
            VectorMath.add(singleCentroid, dataPoints.getRowVector(i));
        singleCentroid = new ScaledDoubleVector(
                singleCentroid, 1/((double)numDataPoints));
View Full Code Here
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
        /**
         * {@inheritDoc}
         */
        protected DoubleVector computeSecondEigenVector(Matrix matrix,
                                                        int vectorLength) {
            DoubleVector Rinv = new DenseVector(vectorLength);
            DoubleVector baseVector = new DenseVector(vectorLength);
            for (int i = 0; i < vectorLength; ++i) {
                Rinv.set(i, 1/Math.sqrt(rho.get(i)));
                baseVector.set(i, rho.get(i) * Rinv.get(i));
            }
    
            // Step 1, generate a random vector, v,  that is orthogonal to
            // pi*D-Inverse.
            DoubleVector v = new DenseVector(vectorLength);
            for (int i = 0; i < v.length(); ++i)
                v.set(i, Math.random());

            Matrix RinvData = (matrix instanceof SparseMatrix)
                ? new RowScaledSparseMatrix((SparseMatrix) matrix, Rinv)
                : new RowScaledMatrix(matrix, Rinv);

View Full Code Here
82
83
84
85
86
87
88
89
90
91
92
                    SparseVector sv = (SparseVector)sumVector;
                    for (int nz : sv.getNonZeroIndices())
                        centroid.set(nz, sumVector.get(nz) * d);
                }
                else {
                    centroid = new DenseVector(length);
                    for (int i = 0; i < length; ++i)
                        centroid.set(i, sumVector.get(i) * d);                   
                }
            }
        }
View Full Code Here
100
101
102
103
104
105
106
107
108
109
110
        boolean added = indices.add(index);
        assert added : "Adding duplicate indices to candidate facility";
        if (sumVector == null) {
            sumVector = (v instanceof SparseVector)
                ? new SparseHashDoubleVector(v)
                : new DenseVector(v);
        }
        else {
            VectorMath.add(sumVector, v);
            centroid = null;
        }          
View Full Code Here
217
218
219
220
221
222
223
    /**
     * {@inheritDoc}
     */
    public DoubleVector getColumnVector(int column) {
        return new DenseVector(getColumn(column));
    }
View Full Code Here
226
227
228
229
230
231
232
233
234
235
236
                // Incorporate the context into the semantic vector for the
                // focus word.  If the focus word has no semantic vector yet,
                // create a new one, as determined by the index builder.
                DoubleVector meaning = termHolographs.get(focusWord);
                if (meaning == null) {
                    meaning = new DenseVector(indexVectorSize);
                    documentVectors.put(focusWord, meaning);
                }
                updateMeaning(meaning, prevWords, nextWords);
            }
View Full Code Here
277
278
279
280
281
282
283
284
285
286
287
                               Queue<String> nextWords) {
        // Generate the semantics of the context using summation of index
        // vectors.
        if (semanticType == SemanticType.COMPOSITE ||
            semanticType == SemanticType.CONTEXT) {
            DoubleVector context = new DenseVector(indexVectorSize);

            // Sum the words prior to the focus word, skipping filtered tokens.
            for (String term: prevWords) {
                if (term.equals(IteratorFactory.EMPTY_TOKEN))
                    continue;
View Full Code Here
TOP

Related Classes of edu.ucla.sspace.vector.DenseVector

  • edu.ucla.sspace.beagle.Beagle
  • edu.ucla.sspace.clustering.Assignments
  • edu.ucla.sspace.clustering.BaseSpectralCut
  • edu.ucla.sspace.clustering.CandidateCluster
  • edu.ucla.sspace.clustering.CKVWSpectralClustering03$SpectralCut
  • edu.ucla.sspace.clustering.CKVWSpectralClustering06$SuperSpectralCut
  • edu.ucla.sspace.clustering.criterion.BaseFunction
  • edu.ucla.sspace.clustering.FastStreamingKMeans
  • edu.ucla.sspace.clustering.LinkClustering$LazySimilarityMatrix
  • edu.ucla.sspace.clustering.seeding.GeneralizedOrssSeed
    • 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.