weka.classifiers.neural.common
Class CommonNeuralAlgorithmAncestor

java.lang.Object
  weka.classifiers.neural.common.CommonNeuralAlgorithmAncestor

All Implemented Interfaces:

java.io.Serializable, NeuralModel

Direct Known Subclasses:

BackPropagationAlgorithm, SLPAlgorithmAncestor

public abstract class CommonNeuralAlgorithmAncestor
extends java.lang.Object
implements NeuralModel

Date: 31/05/2004 File: CommonNeuralAlgorithmAncestor.java

Author:

Jason Brownlee

See Also:

Serialized Form

Constructor Summary

CommonNeuralAlgorithmAncestor(TransferFunction aTransferFunction, RandomWrapper aRand)

Method Summary

double[]

getDistributionForInstance(weka.core.Instance instance) Responsible for returning a class distribution for the provided instance.

Methods inherited from class java.lang.Object

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

Methods inherited from interface weka.classifiers.neural.common.NeuralModel

finishedEpoch, getAllWeights, getLearningRate, getModelInformation, getNetworkOutputs, getNumOutputNeurons, startingEpoch, updateModel

Constructor Detail

CommonNeuralAlgorithmAncestor

public CommonNeuralAlgorithmAncestor(TransferFunction aTransferFunction,
                                     RandomWrapper aRand)

Method Detail

getDistributionForInstance

public double[] getDistributionForInstance(weka.core.Instance instance)

Responsible for returning a class distribution for the provided instance. Each element is given as a fractional part of the output vectors magnitude.

Specified by:

getDistributionForInstance in interface NeuralModel

Parameters:

instance - - the instance to retrieve the class distribution for

Returns:

double - class distribution vecotr for classification problems otherwise the raw network output for regression problems