yawn.nn.mlp
Class MultiLayerPerceptron

java.lang.Object
  yawn.nn.NeuralNetwork
      yawn.nn.mlp.MultiLayerPerceptron

All Implemented Interfaces:

java.io.Serializable

public class MultiLayerPerceptron
extends NeuralNetwork

A multi-layer perceptron with backpropagation of errors learning. $Id: MultiLayerPerceptron.java,v 1.8 2005/04/07 17:28:07 supermarti Exp $

Version:

$Revision: 1.8 $

Author:

Luis Martí (luis dot marti at uc3m dot es)

See Also:

Serialized Form

Field Summary

protected MultiLayerPerceptronConfig	config
protected long	currentEpoch
protected double	errorSum
protected java.util.List	layers
protected double	learningRate
protected long	maxEpochs
protected double	momentumRate
protected double	predictionError

Constructor Summary

MultiLayerPerceptron()

Method Summary

void	addLayer(int index, MLPLayer element)
void	addLayer(MLPLayer layer)
MultiLayerPerceptron	copy()
int	getInputSize()
MLPLayer	getLayer(int i)
java.util.List	getLayers()
double	getLearningRate()
long	getMaxEpochs()
double	getMomentumRate()
java.lang.String	getNeuralNetworkName() Returns a human readable
int	getOutputSize()
double	getPredictionError()
protected void	init()
int	layersCount()
java.util.Iterator	layersIterator()
java.util.ListIterator	layersListIterator()
void	oneLearningStep(Pattern input, Pattern output) Implements a learning iteration as: propagates the input storing the activation of all nodes; sums in errorSum the mean square error of the prediction; calculation of the deltas; backpropagation of errors, and; weights update.
Pattern	predict(Pattern input) Computes a network prediction
void	removeLayer(int index)
void	setLayers(java.util.List layers)
void	setLearningRate(double learningRate)
void	setMaxEpochs(long maxEpochs)
void	setMomentumRate(double momentumRate)
void	setPredictionError(double predictionError)
void	setup(NeuralNetworkConfig config)
void	train(InputOutputPattern[] iop) Trains the network until the stop criteria is met.
NeuralNetworkConfig	yieldConfiguration()

Methods inherited from class yawn.nn.NeuralNetwork

getAdapting, getStatisticsFacility, isAdapting, setAdapting, setStatisticsFacility

Methods inherited from class java.lang.Object

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

Field Detail

currentEpoch

protected long currentEpoch

errorSum

protected double errorSum

layers

protected java.util.List layers

learningRate

protected double learningRate

momentumRate

protected double momentumRate

maxEpochs

protected long maxEpochs

predictionError

protected double predictionError

config

protected MultiLayerPerceptronConfig config

Constructor Detail

MultiLayerPerceptron

public MultiLayerPerceptron()

Method Detail

addLayer

public void addLayer(int index,
                     MLPLayer element)

Parameters:

index - index of insertion

element - the layer to add

See Also:

List.add(int, Object)

addLayer

public void addLayer(MLPLayer layer)

Parameters:

layer - The layer to add.

getLayer

public MLPLayer getLayer(int i)

getLayers

public java.util.List getLayers()

Returns:

Returns the layers.

getLearningRate

public double getLearningRate()

Returns:

Returns the learningRate.

getMaxEpochs

public long getMaxEpochs()

Returns:

Returns the maxEpochs.

getNeuralNetworkName

public java.lang.String getNeuralNetworkName()

Description copied from class: NeuralNetwork

Returns a human readable

Specified by:

getNeuralNetworkName in class NeuralNetwork

Returns:

The name

See Also:

NeuralNetwork.getNeuralNetworkName()

getPredictionError

public double getPredictionError()

Returns:

Returns the predictionError.

init

protected void init()

See Also:

yawn.nn.NeuralNetwork#init()

getInputSize

public int getInputSize()

Specified by:

getInputSize in class NeuralNetwork

See Also:

NeuralNetwork.getInputSize()

layersCount

public int layersCount()

Returns:

the number of layers in the network

layersIterator

public java.util.Iterator layersIterator()

Returns:

a java.util.Iterator of the layers list

See Also:

List.iterator()

layersListIterator

public java.util.ListIterator layersListIterator()

Returns:

The ListIterator of the layers list.

See Also:

List.listIterator()

oneLearningStep

public void oneLearningStep(Pattern input,
                            Pattern output)

Implements a learning iteration as:

• propagates the input storing the activation of all nodes;
• sums in errorSum the mean square error of the prediction;
• calculation of the deltas;
• backpropagation of errors, and;
• weights update.

Specified by:

oneLearningStep in class NeuralNetwork

Parameters:

input - The input presented to the network.

output - The expected output.

See Also:

NeuralNetwork.oneLearningStep(yawn.util.Pattern, yawn.util.Pattern)

getOutputSize

public int getOutputSize()

Specified by:

getOutputSize in class NeuralNetwork

See Also:

NeuralNetwork.getOutputSize()

predict

public Pattern predict(Pattern input)

Description copied from class: NeuralNetwork

Computes a network prediction

Specified by:

predict in class NeuralNetwork

Parameters:

input - The input to propagate.

Returns:

the network output.

See Also:

NeuralNetwork.predict(yawn.util.Pattern)

removeLayer

public void removeLayer(int index)

Parameters:

index -

setLayers

public void setLayers(java.util.List layers)

Parameters:

layers - The layers to set.

setLearningRate

public void setLearningRate(double learningRate)

Parameters:

learningRate - The learningRate to set.

setMaxEpochs

public void setMaxEpochs(long maxEpochs)

Parameters:

maxEpochs - The maxEpochs to set.

setPredictionError

public void setPredictionError(double predictionError)

Parameters:

predictionError - The predictionError to set.

train

public void train(InputOutputPattern[] iop)

Description copied from class: NeuralNetwork

Trains the network until the stop criteria is met.

Specified by:

train in class NeuralNetwork

Parameters:

iop - The training set to be learned.

See Also:

NeuralNetwork.train(yawn.util.InputOutputPattern[])

setup

public void setup(NeuralNetworkConfig config)
           throws ConfigurationException

Specified by:

setup in class NeuralNetwork

Throws:

ConfigurationException

See Also:

NeuralNetwork.setup(NeuralNetworkConfig)

yieldConfiguration

public NeuralNetworkConfig yieldConfiguration()

Specified by:

yieldConfiguration in class NeuralNetwork

See Also:

NeuralNetwork.yieldConfiguration()

getMomentumRate

public double getMomentumRate()

Returns:

Returns the momentumRate.

setMomentumRate

public void setMomentumRate(double momentumRate)

Parameters:

momentumRate - The momentumRate to set.

copy

public MultiLayerPerceptron copy()