org.joone.engine.extenders

Class GradientExtender

java.lang.Object

org.joone.engine.extenders.LearnerExtender

org.joone.engine.extenders.GradientExtender

public abstract class GradientExtender
extends LearnerExtender

This abstract class describes the methods needed for a gradient extender, that is, a class that computes / changes the gradient value according to some algorithm.

Author:

Boris Jansen

Constructor Summary

Constructors

Constructor and Description
GradientExtender() Creates a new instance of DeltaExtender

Method Summary

Methods

Modifier and Type	Method and Description
abstract double	getGradientBias(double[] currentGradientOuts, int j, double aPreviousGradient) Computes the gradient value for a bias.
abstract double	getGradientWeight(double[] currentInps, int j, double[] currentPattern, int k, double aPreviousGradient) Computes the gradient value for a weight.

Methods inherited from class org.joone.engine.extenders.LearnerExtender

getLearner, isEnabled, postBiasUpdate, postWeightUpdate, preBiasUpdate, preWeightUpdate, setEnabled, setLearner

Methods inherited from class java.lang.Object

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

Constructor Detail

GradientExtender

public GradientExtender()

Creates a new instance of DeltaExtender

Method Detail

getGradientBias

public abstract double getGradientBias(double[] currentGradientOuts,
                     int j,
                     double aPreviousGradient)

Computes the gradient value for a bias.

Parameters:

currentGradientOuts - the back propagated gradients.

j - the index of the bias.

aPreviousGradient - a gradient value calculated by a previous gradient extender.

getGradientWeight

public abstract double getGradientWeight(double[] currentInps,
                       int j,
                       double[] currentPattern,
                       int k,
                       double aPreviousGradient)

Computes the gradient value for a weight.

Parameters:

currentInps - the forwarded input.

j - the input index of the weight.

currentPattern - the back propagated gradients.

k - the output index of the weight.

aPreviousGradient - a gradients value calculated by a previous gradients extender.