yawn.nn.appart
Class RadialBasisFunctionsNeuralNode

java.lang.Object
  yawn.nn.NeuralNode
      yawn.nn.appart.RadialBasisFunctionsNeuralNode

Direct Known Subclasses:

GasRecognitionNode

public class RadialBasisFunctionsNeuralNode
extends NeuralNode

A classifier (extended RBF) node intended for the recognition (F2) layer of a AppART network. $Id: RadialBasisFunctionsNeuralNode.java,v 1.7 2005/03/24 17:59:53 supermarti Exp $

Version:

$Revision: 1.8 $

Author:

Luis Martí (luis dot marti at uc3m dot es)

Field Summary

protected int	categoryIndex
protected double	eta
protected Pattern	lambda
protected Pattern	mu
protected Pattern	sigma
protected double	threshold holds the vigilance parameter specified in GF2

Fields inherited from class yawn.nn.NeuralNode

input, inputSize

Constructor Summary

RadialBasisFunctionsNeuralNode(int inputSize, double aThreshold, int aCategoryIndex)

Method Summary

protected double	activationFunction(Pattern input)
double	bigG()
protected double	bigG(Pattern input)
double	deviationsProduct()
int	getCategoryIndex()
double	getEta()
int	getInputSize()
Pattern	getLambda()
Pattern	getMu()
Pattern	getSigma()
double	getThreshold()
void	learn(double v)
void	learnNewClass(Pattern gamma, int n) Sets up the node to exactly represents the class specified by the current input.
double	netInput()
protected double	netInput(Pattern input)
void	setCategoryIndex(int categoryIndex)
void	setEta(double eta)
void	setLambda(Pattern lambda)
void	setMu(Pattern mu)
void	setSigma(Pattern sigma)
void	setThreshold(double threshold)

Methods inherited from class yawn.nn.NeuralNode

output, reset, setInput, setInputElement, setInputSize

Methods inherited from class java.lang.Object

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

Field Detail

categoryIndex

protected int categoryIndex

eta

protected double eta

lambda

protected Pattern lambda

mu

protected Pattern mu

sigma

protected Pattern sigma

threshold

protected double threshold

holds the vigilance parameter specified in GF2

Constructor Detail

RadialBasisFunctionsNeuralNode

public RadialBasisFunctionsNeuralNode(int inputSize,
                                      double aThreshold,
                                      int aCategoryIndex)

Parameters:

inputSize -

aThreshold -

aCategoryIndex -

Method Detail

activationFunction

protected double activationFunction(Pattern input)

Specified by:

activationFunction in class NeuralNode

See Also:

NeuralNode.activationFunction(yawn.util.Pattern)

bigG

public double bigG()

bigG

protected double bigG(Pattern input)

deviationsProduct

public double deviationsProduct()

getCategoryIndex

public int getCategoryIndex()

Returns:

the index of this category node

getEta

public double getEta()

Returns:

the vakue of eta

getInputSize

public int getInputSize()

Overrides:

getInputSize in class NeuralNode

See Also:

NeuralNode.getInputSize()

getLambda

public Pattern getLambda()

Returns:

the value

getMu

public Pattern getMu()

Returns:

the value

getSigma

public Pattern getSigma()

Returns:

the value

getThreshold

public double getThreshold()

Returns:

the threshold value

learn

public void learn(double v)

learnNewClass

public void learnNewClass(Pattern gamma,
                          int n)

Sets up the node to exactly represents the class specified by the current input. This is used when committing a node.

Parameters:

gamma, - initial standard deviations

n, - index of the class being learned

netInput

public double netInput()

netInput

protected double netInput(Pattern input)

setCategoryIndex

public void setCategoryIndex(int categoryIndex)

Parameters:

categoryIndex -

setEta

public void setEta(double eta)

Parameters:

eta -

setLambda

public void setLambda(Pattern lambda)

Parameters:

lambda -

setMu

public void setMu(Pattern mu)

Parameters:

mu -

setSigma

public void setSigma(Pattern sigma)

Parameters:

sigma -

setThreshold

public void setThreshold(double threshold)

Parameters:

threshold -