org.encog.neural.networks.training.lma
Class JacobianChainRule

java.lang.Object
  org.encog.neural.networks.training.lma.JacobianChainRule

All Implemented Interfaces:

ComputeJacobian

public class JacobianChainRule
extends Object
implements ComputeJacobian

Calculate the Jacobian using the chain rule. ---------------------------------------------------- This implementation of the Levenberg Marquardt algorithm is based heavily on code published in an article by Cesar Roberto de Souza. The original article can be found here: http://crsouza.blogspot.com/2009/11/neural-network-learning-by-levenberg_18.html Portions of this class are under the following copyright/license. Copyright 2009 by Cesar Roberto de Souza, Released under the LGPL.

Constructor Summary

JacobianChainRule(BasicNetwork network, MLDataSet indexableTraining)
Construct the chain rule calculation.

Method Summary

double	calculate(double[] weights) Calculate the Jacobian matrix.
double	getError()
double[][]	getJacobian()
double[]	getRowErrors()

Methods inherited from class java.lang.Object

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

Constructor Detail

JacobianChainRule

public JacobianChainRule(BasicNetwork network,
                         MLDataSet indexableTraining)

Construct the chain rule calculation.

Parameters:

network - The network to use.

indexableTraining - The training set to use.

Method Detail

calculate

public double calculate(double[] weights)

Calculate the Jacobian matrix.

Specified by:

calculate in interface ComputeJacobian

Parameters:

weights - The weights for the neural network.

Returns:

The sum squared of the weights.

getError

public double getError()

Returns:

The sum squared errors.

getJacobian

public double[][] getJacobian()

Specified by:

getJacobian in interface ComputeJacobian

Returns:

The Jacobian matrix.

getRowErrors

public double[] getRowErrors()

Specified by:

getRowErrors in interface ComputeJacobian

Returns:

The errors for each row of the Jacobian.