org.encog.mathutil.matrices.hessian

Class ChainRuleWorker

java.lang.Object

org.encog.mathutil.matrices.hessian.ChainRuleWorker

All Implemented Interfaces:

EngineTask

public class ChainRuleWorker
extends Object
implements EngineTask

A threaded worker that is used to calculate the first derivatives of the output of the neural network. These values are ultimatly used to calculate the Hessian.

Constructor Summary

Constructors

Constructor and Description
ChainRuleWorker(FlatNetwork theNetwork, MLDataSet theTraining, int theLow, int theHigh) Construct the chain rule worker.

Method Summary

Methods

Modifier and Type	Method and Description
double[]	getDerivative()
double	getError()
double[]	getGradients()
double[][]	getHessian()
FlatNetwork	getNetwork()
int	getOutputNeuron()
void	run() The task to perform.
void	setOutputNeuron(int outputNeuron)

Methods inherited from class java.lang.Object

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

Constructor Detail

ChainRuleWorker

public ChainRuleWorker(FlatNetwork theNetwork,
               MLDataSet theTraining,
               int theLow,
               int theHigh)

Construct the chain rule worker.

Parameters:

theNetwork - The network to calculate a Hessian for.

theTraining - The training data.

theLow - The low range.

theHigh - The high range.

Method Detail

run

public void run()

The task to perform.

Specified by:

run in interface EngineTask

getOutputNeuron

public int getOutputNeuron()

Returns:

the outputNeuron

setOutputNeuron

public void setOutputNeuron(int outputNeuron)

Parameters:

outputNeuron - the outputNeuron to set

getDerivative

public double[] getDerivative()

Returns:

The first derivatives, used to calculate the Hessian.

getGradients

public double[] getGradients()

Returns:

the gradients

getError

public double getError()

Returns:

The SSE error.

getNetwork

public FlatNetwork getNetwork()

Returns:

The flat network.

getHessian

public double[][] getHessian()

Returns:

the hessian