weka.classifiers.functions.neural
Class SigmoidUnit

java.lang.Object
  |
  +--weka.classifiers.functions.neural.SigmoidUnit

All Implemented Interfaces:

NeuralMethod, java.io.Serializable

public class SigmoidUnit

extends java.lang.Object

implements NeuralMethod

This can be used by the neuralnode to perform all it's computations (as a sigmoid unit).

Version:

$Revision: 1.4 $

Author:

Malcolm Ware (mfw4@cs.waikato.ac.nz)

See Also:

Serialized Form

Constructor Summary

SigmoidUnit()

Method Summary

double	errorValue(NeuralNode node) This function calculates what the error value should be.
double	outputValue(NeuralNode node) This function calculates what the output value should be.
void	updateWeights(NeuralNode node, double learn, double momentum) This function will calculate what the change in weights should be and also update them.

Methods inherited from class java.lang.Object

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

Constructor Detail

SigmoidUnit

public SigmoidUnit()

Method Detail

outputValue

public double outputValue(NeuralNode node)

This function calculates what the output value should be.

Specified by:

outputValue in interface NeuralMethod

Parameters:

node - The node to calculate the value for.

Returns:

The value.

errorValue

public double errorValue(NeuralNode node)

This function calculates what the error value should be.

Specified by:

errorValue in interface NeuralMethod

Parameters:

node - The node to calculate the error for.

Returns:

The error.

updateWeights

public void updateWeights(NeuralNode node,
                          double learn,
                          double momentum)

This function will calculate what the change in weights should be and also update them.

Specified by:

updateWeights in interface NeuralMethod

Parameters:

node - The node to update the weights for.

learn - The learning rate to use.

momentum - The momentum to use.