org.encog.mathutil.matrices.hessian

Interface ComputeHessian

All Known Implementing Classes:

BasicHessian, HessianCR, HessianFD

public interface ComputeHessian

Compute (estimate) the Hessian matrix. The Hessian matrix is a matrix of the second derivatives of the neural network. This is a square matrix with rows and columns equal to the number of weights in the neural network. A Hessian matrix is useful for several neural network functions. It is also used by the Levenberg Marquardt training method. http://en.wikipedia.org/wiki/Hessian_matrix

Method Summary

Methods

Modifier and Type	Method and Description
void	clear() Clear the Hessian and gradients.
void	compute() Compute the Hessian.
double[]	getGradients() The gradeints.
double[][]	getHessian()
Matrix	getHessianMatrix()
double	getSSE()
void	init(BasicNetwork theNetwork, MLDataSet theTraining) Init the class.

Method Detail

init

void init(BasicNetwork theNetwork,
        MLDataSet theTraining)

Init the class.

Parameters:

theNetwork - The neural network to train.

theTraining - The training set to train with.

compute

void compute()

Compute the Hessian.

getGradients

double[] getGradients()

The gradeints.

Returns:

The gradients for the Hessian.

getSSE

double getSSE()

Returns:

The sum of squares error over all of the training elements.

clear

void clear()

Clear the Hessian and gradients.

getHessianMatrix

Matrix getHessianMatrix()

Returns:

The Hessian matrix.

getHessian

double[][] getHessian()

Returns:

Get the Hessian as a 2d array.