weka.classifiers.functions.neural

Interface NeuralMethod

All Superinterfaces:

Serializable

All Known Implementing Classes:

LinearUnit, SigmoidUnit

public interface NeuralMethod
extends Serializable

This is an interface used to create classes that can be used by the neuralnode to perform all it's computations.

Version:

$Revision: 1.6 $

Author:

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

Method Summary

Modifier and Type	Method and Description
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.

Method Detail

outputValue

double outputValue(NeuralNode node)

This function calculates what the output value should be.

Parameters:

node - The node to calculate the value for.

Returns:

The value.

errorValue

double errorValue(NeuralNode node)

This function calculates what the error value should be.

Parameters:

node - The node to calculate the error for.

Returns:

The error.

updateWeights

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

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

Parameters:

node - The node to update the weights for.

learn - The learning rate to use.

momentum - The momentum to use.