Package org.jblas

Examples of org.jblas.DoubleMatrix$ColumnsAsListView

98
99
100
101
102
103
104
105
106
107
108
109
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
152
153
154
155
156
157
158
  protected void gradientUpdateMiniBatch(SGDTrainConfig config, DoubleMatrix x_samples, DoubleMatrix y_samples, SGDParam curr_param) {
    int nbr_sample = x_samples.rows;
    BPParam curr_pbparam = (BPParam)curr_param;
    DoubleMatrix[] activation = new DoubleMatrix[curr_pbparam.nl];
    DoubleMatrix[] l_bias = new DoubleMatrix[curr_pbparam.nl];
    DoubleMatrix avg_hidden = null;
   
    /**
     * feedforward
     */
    activation[0] = x_samples;
    for(int i = 1; i < curr_pbparam.nl; i++) {
      activation[i] = activation[i - 1].mmul(curr_pbparam.w[i - 1].transpose()).addiRowVector(curr_pbparam.b[i - 1]);
      MathUtil.sigmod(activation[i]);
    }
    //sparsity
    if(config.isForceSparsity()) {
      avg_hidden = activation[1].columnSums().divi(nbr_sample);
    }
   
    /**
     * backward
     */
    // 1 last layer
    DoubleMatrix ai = activation[curr_pbparam.nl - 1];
    l_bias[curr_pbparam.nl - 1] = ai.sub(x_samples).muli(ai).muli(ai.neg().addi(1));
   
    //2 back
    for(int i = curr_pbparam.nl - 2; i >= 1; i--) {
      l_bias[i] = l_bias[i + 1].mmul(curr_pbparam.w[i]);
      if(config.isForceSparsity()) {
        DoubleMatrix sparsity_v = avg_hidden.dup();
        for(int k = 0; k < sparsity_v.columns; k++) {
          double roat = config.getSparsity();
          double roat_k = sparsity_v.get(0, k);
          sparsity_v.put(0, k, config.getSparsityBeta()*((1-roat)/(1-roat_k) - roat/roat_k));
        }
        l_bias[i].addiRowVector(sparsity_v);
      }
      ai = activation[i];
      l_bias[i].muli(ai).muli(ai.neg().addi(1));
    }
   
    /**
     * delta
     */
    for(int i = 0; i < curr_pbparam.w.length; i++) {
      DoubleMatrix delta_wi = l_bias[i + 1].transpose().mmul(activation[i]).divi(nbr_sample);
      if(config.isUseRegularization()) {
        //for bp, only use L2
        if(0 != config.getLamada2()) {
            delta_wi.addi(curr_pbparam.w[i].mul(config.getLamada2()));
        }
      }
      curr_pbparam.w[i].subi(delta_wi.muli(config.getLearningRate()));
    }
    for(int i = 0; i < curr_pbparam.b.length; i++) {
      DoubleMatrix delta_bi = l_bias[i + 1].columnSums().divi(nbr_sample);
      curr_pbparam.b[i].subi(delta_bi.transpose().muli(config.getLearningRate()));
    }
  }
View Full Code Here
112
113
114
115
116
117
118
119
120
121
122
123
124
   
  }

  @Override
  protected void gradientUpdateCG(SGDTrainConfig config, DoubleMatrix x_samples, DoubleMatrix y_samples, SGDParam curr_param) {
    DoubleMatrix curr_w = ((HiddenLayerParam)curr_param).w;
      DoubleMatrix curr_hbias = ((HiddenLayerParam)curr_param).hbias;
      DoubleMatrix curr_vbias = ((HiddenLayerParam)curr_param).vbias;
     
    dAOptimizer daopt = new dAOptimizer(config, x_samples, n_visible, n_hidden, curr_w, curr_hbias, curr_vbias);
        MyConjugateGradient cg = new MyConjugateGradient(daopt, config.getCgInitStepSize());
        cg.setTolerance(config.getCgTolerance());
        try {
View Full Code Here
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
            logger.error("", e);
        }
  }

  private DoubleMatrix get_corrupted_input(DoubleMatrix x, double p) {
        DoubleMatrix ret = new DoubleMatrix(x.getRows(), x.getColumns());
        for (int i = 0; i < x.getRows(); i++) {
            for (int j = 0; j < x.getColumns(); j++) {
                if (0 != x.get(i, j)) {
                    ret.put(i, j, MathUtil.binomial(1, p));
                }
            }
        }
        return ret;
    }
View Full Code Here
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
     */
    public static DoubleMatrix convertX2Matrix(List<SampleVector> samples) {
        int row = samples.size();
        int col = samples.get(0).getX().length;

        DoubleMatrix ret = new DoubleMatrix(row, col);
        row = 0;
        for (SampleVector sample: samples) {
            double[] x = sample.getX();
            for (col = 0; col < x.length; col++) {
                ret.put(row, col, x[col]);
            }
            row++;
        }
        return ret;
    }
View Full Code Here
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
     */
    public static DoubleMatrix convertY2Matrix(List<SampleVector> samples) {
        int row = samples.size();
        int col = samples.get(0).getY().length;

        DoubleMatrix ret = new DoubleMatrix(row, col);
        row = 0;
        for (SampleVector sample: samples) {
            double[] y = sample.getY();
            for (col = 0; col < y.length; col++) {
                ret.put(row, col, y[col]);
            }
            row++;
        }
        return ret;
    }
View Full Code Here
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
            return -loss;
        }

        @Override
        public void getValueGradient(double[] arg) {
            DoubleMatrix L_vbias = my_samples.sub(z);
            DoubleMatrix L_hbias = L_vbias.mmul(my_w.transpose()).muli(y).muli(y.neg().addi(1));
            DoubleMatrix delta_w = null;
            if (myConfig.isDoCorruption()) {
                delta_w = L_hbias.transpose().mmul(tilde_x).addi(y.transpose().mmul(L_vbias));
            } else {
                delta_w = L_hbias.transpose().mmul(my_samples).addi(y.transpose().mmul(L_vbias));
            }
            if (myConfig.isUseRegularization()) {
        //only L2 for autoencoder
        if (0 != myConfig.getLamada2()) {
          delta_w.subi(my_w.mul(myConfig.getLamada2()));
                }
      }
            delta_w.divi(nbr_sample);
            DoubleMatrix delta_hbias = L_hbias.columnSums().divi(nbr_sample);
            DoubleMatrix delta_vbias = L_vbias.columnSums().divi(nbr_sample);

            int idx = 0;
            for (int i = 0; i < n_hidden; i++) {
                for (int j = 0; j < n_visible; j++) {
                    arg[idx++] = delta_w.get(i, j);
                }
            }
            for (int i = 0; i < n_hidden; i++) {
                arg[idx++] = delta_hbias.get(0, i);
            }
            for (int i = 0; i < n_visible; i++) {
                arg[idx++] = delta_vbias.get(0, i);
            }
        }
View Full Code Here
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
      }
    }
  }
 
  private DoubleMatrix get_corrupted_input(DoubleMatrix x, double p) {
        DoubleMatrix ret = new DoubleMatrix(x.getRows(), x.getColumns());
        for (int i = 0; i < x.getRows(); i++) {
            for (int j = 0; j < x.getColumns(); j++) {
                if (0 != x.get(i, j)) {
                    ret.put(i, j, MathUtil.binomial(1, p));
                }
            }
        }
        return ret;
    }
View Full Code Here
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
      /**
       * backward
       */
      // 1 last layer
      DoubleMatrix ai = activation[my_bpparam.nl - 1];
      l_bias[my_bpparam.nl - 1] = ai.sub(activation[0]).muli(ai).muli(ai.neg().addi(1));
     
      //2 back(no layer0 error need)
      for(int i = my_bpparam.nl - 2; i >= 1; i--) {
        l_bias[i] = l_bias[i + 1].mmul(my_bpparam.w[i]);
        if(my_config.isForceSparsity()) {
          DoubleMatrix sparsity_v = avg_hidden.dup();
          for(int k = 0; k < sparsity_v.columns; k++) {
            double roat = my_config.getSparsity();
            double roat_k = sparsity_v.get(0, k);
            sparsity_v.put(0, k, my_config.getSparsityBeta()*((1-roat)/(1-roat_k) - roat/roat_k));
          }
          l_bias[i].addiRowVector(sparsity_v);
        }
        ai = activation[i];
        l_bias[i].muli(ai).muli(ai.neg().addi(1));
      }
     
      /**
       * delta
       */
      int idx = 0;
      for(int i = 0; i < my_bpparam.w.length; i++) {
        DoubleMatrix delta_wi = l_bias[i + 1].transpose().mmul(activation[i]).divi(nbr_samples);
        if(my_config.isUseRegularization()) {
          //for bp, only use L2
          if(0 != my_config.getLamada2()) {
              delta_wi.addi(my_bpparam.w[i].mul(my_config.getLamada2()));
          }
        }
        for(int row = 0; row < delta_wi.rows; row++) {
          for(int col = 0; col < delta_wi.columns; col++) {
            arg[idx++] = -delta_wi.get(row, col);
          }
        }
      }
      for(int i = 0; i < my_bpparam.b.length; i++) {
        DoubleMatrix delta_bi = l_bias[i + 1].columnSums().divi(nbr_samples);
        for(int row = 0; row < delta_bi.rows; row++) {
          for(int col = 0; col < delta_bi.columns; col++) {
            arg[idx++] = -delta_bi.get(row, col);
          }
        }
      }
    }
View Full Code Here
72
73
74
75
76
77
78
79
80
     * Input layer to hidden layer Sigmod output(standalone)
     * @param input Input layer matrix
     * @return Hidden layer output matrix
     */
    public final DoubleMatrix sigmod_output(DoubleMatrix input) {
        DoubleMatrix ret = input.mmul(hlparam.w.transpose()).addiRowVector(hlparam.hbias);
        MathUtil.sigmod(ret);
        return ret;
    }
View Full Code Here
193
194
195
196
197
198
199
200
201
      hlparam.vbias.addi(new_hlparam.vbias.sub(hlparam.vbias).divi(nrModelReplica));
  }
   
    @Override
    protected final double loss(List<SampleVector> samples) {
      DoubleMatrix x = MathUtil.convertX2Matrix(samples);
        DoubleMatrix reconstruct_x = reconstruct(x);
        return MatrixFunctions.powi(reconstruct_x.sub(x), 2).sum();
    }
View Full Code Here
TOP

Related Classes of org.jblas.DoubleMatrix$ColumnsAsListView

  • com.github.pmerienne.trident.ml.classification.AROWClassifier
  • com.numb3r3.common.opt.GradientChecker
  • com.numb3r3.common.opt.Norm1Tanh
  • com.numb3r3.common.opt.Sigmoid
  • com.numb3r3.common.opt.Softmax
  • com.numb3r3.common.opt.Tanh
  • org.apache.spark.mllib.recommendation.JavaALSSuite
  • org.apache.spark.mllib.regression.JavaRidgeRegressionSuite
  • org.gd.spark.opendl.downpourSGD.Backpropagation.AutoEncoder
  • org.gd.spark.opendl.downpourSGD.Backpropagation.AutoEncoder$AEOptimizer
    • 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.