Package tv.floe.metronome.deeplearning.neuralnetwork.core.learning

Examples of tv.floe.metronome.deeplearning.neuralnetwork.core.learning.AdagradLearningRate

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
        } else {
            this.connectionWeights = weights;
        }


        this.wAdagrad = new AdagradLearningRate( this.connectionWeights.numRows(), this.connectionWeights.numCols() );


        if (hBias == null) {
            // TODO: recheck if this column vector is correctly oriented
            this.hiddenBiasNeurons = new DenseMatrix(1, nHidden); //Matrix.zeros(nHidden);
            //} else if(hBias.numRows() != nHidden) {
            //throw new IllegalArgumentException("Hidden bias must have a length of " + nHidden + " length was " + hBias.numRows());
        } else {
            this.hiddenBiasNeurons = hBias;
        }

        //         this.hBiasAdaGrad = new AdaGrad(hBias.rows,hBias.columns);

        this.hBiasAdaGrad = new AdagradLearningRate( this.hiddenBiasNeurons.numRows(), this.hiddenBiasNeurons.numCols() );

        if (vBias == null) {
            this.visibleBiasNeurons = new DenseMatrix(1, nVisible); //Matrix.zeros(nVisible);
            this.visibleBiasNeurons.assign(0);

        } else if(vBias.numRows() != nVisible) {
            throw new IllegalArgumentException("Visible bias must have a length of " + nVisible + " but length was " + vBias.numRows());

        } else {
            this.visibleBiasNeurons = vBias;
        }

        // this.vBiasAdaGrad = new AdaGrad(vBias.rows,vBias.columns);

        this.vBiasAdaGrad = new AdagradLearningRate( this.visibleBiasNeurons.numRows(), this.visibleBiasNeurons.numCols() );

    }
View Full Code Here
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
            }

        }

        this.wAdagrad = new AdagradLearningRate( this.connectionWeights.numRows(), this.connectionWeights.numCols() );


        //if(this.hBias == null) {
        if ( this.hiddenBiasNeurons == null) {

            //this.hBias = Matrix.zeros(nHidden);
            this.hiddenBiasNeurons = new DenseMatrix(1, this.numberHiddenNeurons);// Matrix.zeros(nHidden);
            this.hiddenBiasNeurons.assign(0.0);
     
      /*
       * Encourage sparsity.
       * See Hinton's Practical guide to RBMs
       */
            //this.hBias.subi(4);
        }

        this.hBiasAdaGrad = new AdagradLearningRate( this.hiddenBiasNeurons.numRows(), this.hiddenBiasNeurons.numCols() );


        if (this.visibleBiasNeurons == null) {

            if (this.trainingDataset != null) {

                this.visibleBiasNeurons = new DenseMatrix(1, this.numberVisibleNeurons); // Matrix.zeros(nVisible);
                this.visibleBiasNeurons.assign(0.0);


            } else {
//        this.vBias = Matrix.zeros(nVisible);
                this.visibleBiasNeurons = new DenseMatrix(1, this.numberVisibleNeurons); // Matrix.zeros(nVisible);
                this.visibleBiasNeurons.assign(0.0);

            }

        }

        this.vBiasAdaGrad = new AdagradLearningRate( this.visibleBiasNeurons.numRows(), this.visibleBiasNeurons.numCols() );




    }
View Full Code Here
846
847
848
849
850
851
852
853
854
        this.hiddenBiasNeurons.assign(0.0);

        //  this.visibleBiasNeurons = new DenseMatrix(1, this.numberVisibleNeurons); // Matrix.zeros(nVisible);
        this.visibleBiasNeurons.assign(0.0);

        this.wAdagrad = new AdagradLearningRate( this.connectionWeights.numRows(), this.connectionWeights.numCols() );


    }
View Full Code Here
58
59
60
61
62
63
64
65
66
67
68
69
    this.connectionWeights = new DenseMatrix(nIn, nOut);
    this.connectionWeights.assign(0.0);
    this.biasTerms = new DenseMatrix(1, nOut); //Matrix.zeros(nOut);
    this.biasTerms.assign(0.0);
   
    this.adaLearningRates = new AdagradLearningRate( nIn, nOut );
   
    this.biasAdaGrad = new AdagradLearningRate( this.biasTerms.numRows(), this.biasTerms.numCols() );
 
    optimizationAlgorithm = OptimizationAlgorithm.CONJUGATE_GRADIENT;
       
  }
View Full Code Here
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
     
        this.useAdaGrad = di.readBoolean();
       
        if ( null == this.adaLearningRates ) {

          this.adaLearningRates = new AdagradLearningRate( nIn, nOut );
        }

        this.adaLearningRates.load( is );
       
        if ( null == this.biasAdaGrad ) {
       
        this.biasAdaGrad = new AdagradLearningRate( this.biasTerms.numRows(), this.biasTerms.numCols() );
         
        }
       
        this.biasAdaGrad.load( is );
       
View Full Code Here
TOP

Related Classes of tv.floe.metronome.deeplearning.neuralnetwork.core.learning.AdagradLearningRate

  • org.apache.mahout.math.DenseMatrix
  • org.apache.mahout.math.Matrix
  • tv.floe.metronome.deeplearning.neuralnetwork.core.BaseNeuralNetworkVectorized
  • tv.floe.metronome.deeplearning.neuralnetwork.core.LogisticRegression
  • java.io.DataInput
  • java.io.DataOutputStream
  • java.io.DataOutput
  • java.io.DataInputStream
    • 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.