org.neuroph.nnet.learning

Class ResilientPropagation

java.lang.Object

org.neuroph.core.learning.LearningRule

org.neuroph.core.learning.IterativeLearning

org.neuroph.core.learning.SupervisedLearning

org.neuroph.nnet.learning.LMS

org.neuroph.nnet.learning.BackPropagation

org.neuroph.nnet.learning.ResilientPropagation

All Implemented Interfaces:

Serializable

public class ResilientPropagation
extends BackPropagation

Resilient Propagation learning rule used for Multi Layer Perceptron neural networks. Its one of the most efficent learning rules for this type of networks, and it does not require setting of learning rule parameter.

Author:

Borislav Markov, Zoran Sevarac

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	ResilientPropagation.ResilientWeightTrainingtData

Field Summary

Fields inherited from class org.neuroph.core.learning.SupervisedLearning

maxError, previousEpochError

Fields inherited from class org.neuroph.core.learning.IterativeLearning

currentIteration, learningRate, stopConditions

Fields inherited from class org.neuroph.core.learning.LearningRule

listeners, neuralNetwork, trainingSet

Constructor Summary

Constructors

Constructor and Description
ResilientPropagation()

Method Summary

Modifier and Type	Method and Description
void	calculateWeightChanges(Neuron neuron) Calculate and sum gradients for each neuron's weight, the actual weight update is done in batch mode.
protected void	doBatchWeightsUpdate() This method updates network weights in batch mode - use accumulated weights change stored in Weight.deltaWeight It is executed after each learning epoch, only if learning is done in batch mode.
double	getDecreaseFactor()
double	getIncreaseFactor()
double	getInitialDelta()
double	getMaxDelta()
double	getMinDelta()
protected void	onStart() This method is executed when learning starts, before the first epoch.
protected void	resillientWeightUpdate(Weight weight) Weight update by done by ResilientPropagation learning rule Executed at the end of epoch (in batch mode)
void	setBatchMode(boolean batchMode) Sets batch mode on/off (true/false)
void	setDecreaseFactor(double decreaseFactor)
void	setIncreaseFactor(double increaseFactor)
void	setInitialDelta(double initialDelta)
void	setMaxDelta(double maxDelta)
void	setMinDelta(double minDelta)

Methods inherited from class org.neuroph.nnet.learning.BackPropagation

calculateErrorAndUpdateHiddenNeurons, calculateErrorAndUpdateOutputNeurons, calculateHiddenNeuronError, calculateWeightChanges

Methods inherited from class org.neuroph.core.learning.SupervisedLearning

afterEpoch, beforeEpoch, doLearningEpoch, getErrorFunction, getMaxError, getMinErrorChange, getMinErrorChangeIterationsCount, getMinErrorChangeIterationsLimit, getPreviousEpochError, getTotalNetworkError, isBatchMode, learn, learn, learnPattern, setErrorFunction, setMaxError, setMinErrorChange, setMinErrorChangeIterationsLimit

Methods inherited from class org.neuroph.core.learning.IterativeLearning

doOneLearningIteration, getCurrentIteration, getLearningRate, getMaxIterations, hasReachedStopCondition, isIterationsLimited, isPausedLearning, learn, learn, pause, resume, setLearningRate, setMaxIterations

Methods inherited from class org.neuroph.core.learning.LearningRule

addListener, fireLearningEvent, getNeuralNetwork, getTrainingSet, isStopped, onStop, removeListener, setNeuralNetwork, setTrainingSet, stopLearning

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

ResilientPropagation

public ResilientPropagation()

Method Detail

onStart

protected void onStart()

Description copied from class: IterativeLearning

This method is executed when learning starts, before the first epoch. Used for initialisation.

Overrides:

onStart in class SupervisedLearning

calculateWeightChanges

public void calculateWeightChanges(Neuron neuron)

Calculate and sum gradients for each neuron's weight, the actual weight update is done in batch mode.

Overrides:

calculateWeightChanges in class LMS

Parameters:

neuron - neuron to update weights

See Also:

resillientWeightUpdate(org.neuroph.core.Weight)

doBatchWeightsUpdate

protected void doBatchWeightsUpdate()

Description copied from class: SupervisedLearning

This method updates network weights in batch mode - use accumulated weights change stored in Weight.deltaWeight It is executed after each learning epoch, only if learning is done in batch mode.

Overrides:

doBatchWeightsUpdate in class SupervisedLearning

See Also:

SupervisedLearning.doLearningEpoch(org.neuroph.core.data.DataSet)

resillientWeightUpdate

protected void resillientWeightUpdate(Weight weight)

Weight update by done by ResilientPropagation learning rule Executed at the end of epoch (in batch mode)

Parameters:

weight -

getDecreaseFactor

public double getDecreaseFactor()

setDecreaseFactor

public void setDecreaseFactor(double decreaseFactor)

getIncreaseFactor

public double getIncreaseFactor()

setIncreaseFactor

public void setIncreaseFactor(double increaseFactor)

getInitialDelta

public double getInitialDelta()

setInitialDelta

public void setInitialDelta(double initialDelta)

getMaxDelta

public double getMaxDelta()

setMaxDelta

public void setMaxDelta(double maxDelta)

getMinDelta

public double getMinDelta()

setMinDelta

public void setMinDelta(double minDelta)

setBatchMode

public void setBatchMode(boolean batchMode)

Description copied from class: SupervisedLearning

Sets batch mode on/off (true/false)

Overrides:

setBatchMode in class SupervisedLearning

Parameters:

batchMode - batch mode setting