Package com.clearnlp.classification.vector

Examples of com.clearnlp.classification.vector.SparseFeatureVector

300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
   * @param vector the string feature vector.
   * @return the sparse feature vector converted from the string feature vector.
   */
  public SparseFeatureVector toSparseFeatureVector(StringFeatureVector vector)
  {
    SparseFeatureVector sparse = new SparseFeatureVector(vector.hasWeight());
    int i, index, size = vector.size();
    ObjectIntHashMap<String> map;
    String type, value;
   
    for (i=0; i<size; i++)
    {
      type  = vector.getType(i);
      value = vector.getValue(i);
     
      if ((map = m_features.get(type)) != null && (index = map.get(value)) > 0)
      {
        if (sparse.hasWeight())
          sparse.addFeature(index, vector.getWeight(i));
        else
          sparse.addFeature(index);
      }
    }
   
    sparse.trimToSize();
    return sparse;
  }
View Full Code Here
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
 
// ================================ BUILD ================================

  public void build(int labelCutoff, int featureCutoff, int randomSeed, boolean initialize)
  {
    SparseFeatureVector vector;
    StringInstance instance;
    int label;
   
    if (initialize) init();
    buildLabels(labelCutoff);
    buildFeatures(featureCutoff);
   
    l_instances = Lists.newArrayList();
    r_shuffle = new Random(randomSeed);
    l_indices = new IntArrayList();
   
    while ((instance = i_collector.pollInstance()) != null)
    {
      if ((label = getLabelIndex(instance.getLabel())) < 0)
        continue;
     
      vector = toSparseFeatureVector(instance.getFeatureVector());
     
      if (!vector.isEmpty())
      {
        l_instances.add(new IntInstance(label, vector));
        l_indices.add(l_indices.size());
      }
    }
View Full Code Here
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
    return scores;
  }
 
  protected void updateCounts(StringModelAD model, IntInstance instance, double[] gradidents)
  {
    SparseFeatureVector x = instance.getFeatureVector();
    int i, j, len = x.size(), L = model.getLabelSize();
    double[] g = new double[L];
    double d;
   
    for (j=0; j<L; j++)
      g[j] = gradidents[j] * gradidents[j];
   
    for (i=0; i<len; i++)
    {
      d = UTMath.sq(x.getWeight(i));
     
      for (j=0; j<L; j++)
        d_gradients[model.getWeightIndex(j, x.getIndex(i))] += d * g[j];
    }
  }
 
View Full Code Here
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
    }
  }
 
  private void updateWeights(StringModelAD model, IntInstance instance, double[] gradients, int averageCount)
  {
    SparseFeatureVector x = instance.getFeatureVector();
    int i, j, len = x.size(), L = model.getLabelSize();
    int xi; double vi;
   
    for (i=0; i<len; i++)
    {
      xi = x.getIndex(i);
      vi = x.getWeight(i);
     
      for (j=0; j<L; j++)
        updateWeight(model, j, xi, gradients[j]*vi, averageCount);
    }
  }
View Full Code Here
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
    return qd;
  }
 
  private void updateWeights(StringModelAD model, IntInstance instance, int yp, int yn, double d)
  {
    SparseFeatureVector x = instance.getFeatureVector();
    int i, xi, len = x.size();
    float vi;
   
    for (i=0; i<len; i++)
    {
      xi = x.getIndex(i);
      vi = (float)(x.getWeight(i) * d);
     
      model.updateWeight(yp, xi,  vi);
      model.updateWeight(yn, xi, -vi);
    }
  }
View Full Code Here
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
    return Collections.max(ps);
  }
 
  private void updateCounts(StringModelAD model, IntInstance instance, int yp, int yn)
  {
    SparseFeatureVector x = instance.getFeatureVector();
    int i, xi, len = x.size();
    double vi;
   
    for (i=0; i<len; i++)
    {
      xi = x.getIndex(i);
      vi = UTMath.sq(x.getWeight(i));
     
      d_gradients[model.getWeightIndex(yp, xi)] += vi;
      d_gradients[model.getWeightIndex(yn, xi)] += vi;
    }
  }
View Full Code Here
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
    }
  }
 
  private void updateWeights(StringModelAD model, IntInstance instance, int yp, int yn, int averageCount)
  {
    SparseFeatureVector x = instance.getFeatureVector();
    int i, xi, len = x.size();
    double vi;
   
    for (i=0; i<len; i++)
    {
      xi = x.getIndex(i);
      vi = x.getWeight(i);

      updateWeight(model, yp, xi,  vi, averageCount);
      updateWeight(model, yn, xi, -vi, averageCount);
    }
  }
View Full Code Here
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
    LOG.info("Building:\n");
    initModelMaps();
   
    StringInstance instance;
    int y, i, size = s_instances.size();
    SparseFeatureVector x;
   
    for (i=0; i<size; i++)
    {
      instance = s_instances.get(i);
     
      if ((y = s_model.getLabelIndex(instance.getLabel())) < 0)
        continue;
     
      x = s_model.toSparseFeatureVector(instance.getFeatureVector());
     
      a_ys.add(y);
      a_xs.add(x.getIndices());
      if (b_weight)  a_vs.add(x.getWeights());
    }
   
    a_ys.trimToSize();
    a_xs.trimToSize();
    if (b_weight)  a_vs.trimToSize();
View Full Code Here
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
    LOG.info("Building:\n");
    initModelMaps();
   
    StringInstance instance;
    int y, i, size = s_instances.size();
    SparseFeatureVector x;
   
    for (i=0; i<size; i++)
    {
      instance = s_instances.get(i);
     
      if ((y = s_model.getLabelIndex(instance.getLabel())) < 0)
        continue;
     
      x = s_model.toSparseFeatureVector(instance.getFeatureVector());
     
      a_ys.add(y);
      a_xs.add(x.getIndices());
      if (b_weight)  a_vs.add(x.getWeights());
    }
   
    a_ys.trimToSize();
    a_xs.trimToSize();
    if (b_weight)  a_vs.trimToSize();
View Full Code Here
129
130
131
132
133
134
135
136
137
138
139
140
141
  static public Pair<String,SparseFeatureVector> toInstance(String line, boolean hasWeight)
  {
    String[] tmp = line.split(DELIM_COL);
    String label = tmp[0];
   
    SparseFeatureVector vector = new SparseFeatureVector(hasWeight);
    int i, size = tmp.length;
   
    for (i=1; i<size; i++)
      vector.addFeature(tmp[i]);
   
    return new Pair<String,SparseFeatureVector>(label, vector);
  }
View Full Code Here
TOP

Related Classes of com.clearnlp.classification.vector.SparseFeatureVector

  • com.carrotsearch.hppc.IntArrayList
  • com.clearnlp.classification.algorithm.AdaGradOnlineHingeLoss
  • com.clearnlp.classification.algorithm.AdaGradOnlineLogisticRegression
  • com.clearnlp.classification.algorithm.LiblinearHingeLoss
  • com.clearnlp.classification.algorithm.online.OnlineAdaGradHinge
  • com.clearnlp.classification.model.StringModel
  • com.clearnlp.classification.model.StringModelAD
  • com.clearnlp.classification.model.StringOnlineModel
  • com.clearnlp.classification.train.SparseTrainSpace
  • com.clearnlp.classification.train.StringTrainSpace
    • 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.