weka.classifiers.neural.multilayerperceptron.algorithm
Class BoldDriverBackPropagationAlgorithm

java.lang.Object
  extended by weka.classifiers.neural.common.CommonNeuralAlgorithmAncestor
      extended by weka.classifiers.neural.multilayerperceptron.algorithm.BackPropagationAlgorithm
          extended by weka.classifiers.neural.multilayerperceptron.algorithm.BoldDriverBackPropagationAlgorithm
All Implemented Interfaces:
java.io.Serializable, BatchTrainableNeuralModel, NeuralModel

public class BoldDriverBackPropagationAlgorithm
extends BackPropagationAlgorithm

Date: 31/05/2004 File: BoldDriverBackPropagationAlgorithm.java Adaptive learning rate algorithm, Bold Driver (Voglís Method)

Author:
Jason Brownlee
See Also:
Serialized Form

Field Summary
static double ERROR_INCREASE_SCALE
           
 
Constructor Summary
BoldDriverBackPropagationAlgorithm(TransferFunction aTransferFunction, RandomWrapper aRand, LearningRateKernel aLearningRateKernel, double aMomentum, double aWeightDecay, double aBiasValue, int[] aHiddenLayersTopology, weka.core.Instances aTrainingInstances, double increase, double decrease)
           
 
Method Summary
 void finishedEpoch(weka.core.Instances instances, double aLearningRate)
           
 double getLearningRate(int epochNumber)
           
 java.lang.String getModelInformation()
           
 
Methods inherited from class weka.classifiers.neural.multilayerperceptron.algorithm.BackPropagationAlgorithm
applyWeightDeltas, calculateWeightErrors, getAllWeights, getNetworkOutputs, getNumOutputNeurons, startingEpoch, updateModel
 
Methods inherited from class weka.classifiers.neural.common.CommonNeuralAlgorithmAncestor
getDistributionForInstance
 
Methods inherited from class java.lang.Object
equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

ERROR_INCREASE_SCALE

public static final double ERROR_INCREASE_SCALE
See Also:
Constant Field Values
Constructor Detail

BoldDriverBackPropagationAlgorithm

public BoldDriverBackPropagationAlgorithm(TransferFunction aTransferFunction,
                                          RandomWrapper aRand,
                                          LearningRateKernel aLearningRateKernel,
                                          double aMomentum,
                                          double aWeightDecay,
                                          double aBiasValue,
                                          int[] aHiddenLayersTopology,
                                          weka.core.Instances aTrainingInstances,
                                          double increase,
                                          double decrease)
Method Detail

getModelInformation

public java.lang.String getModelInformation()
Specified by:
getModelInformation in interface NeuralModel
Overrides:
getModelInformation in class BackPropagationAlgorithm

getLearningRate

public double getLearningRate(int epochNumber)
Specified by:
getLearningRate in interface NeuralModel
Overrides:
getLearningRate in class BackPropagationAlgorithm

finishedEpoch

public void finishedEpoch(weka.core.Instances instances,
                          double aLearningRate)
Specified by:
finishedEpoch in interface NeuralModel
Overrides:
finishedEpoch in class BackPropagationAlgorithm