keel.Algorithms.Neural_Networks.IRPropPlus_Clas

Interface IOptimizableFunc

All Known Implementing Classes:

MSEOptimizablePUNeuralNetClassifier, MSEOptimizablePUNeuralNetRegressor, MSEOptimizableSigmNeuralNetClassifier, MSEOptimizableSigmNeuralNetRegressor

public interface IOptimizableFunc

Since:

JDK1.5

Version:

0.1

Author:

by Pedro Antonio Gutierrez Penia (University of Cordoba) 27/10/2007

Method Summary

Modifier and Type	Method and Description
double[]	getCoefficients() Returns the initial value of a[], that is, the coefficients of the model
double	getLastError() Last error of the model
double[]	gradient(double[][] x, double[][] y) Returns the gradient vector of the derivative of an error function (E) with respect to each coefficient of the model, using an input observation matrix (x[]) and an expected output matrix (y[]).
void	setCoefficients(double[] a) Establish the final value of a[], that is, the coefficients of model

Method Detail

getCoefficients

double[] getCoefficients()

Returns the initial value of a[], that is, the coefficients of the model

Returns:

double array of initial coefficients values

setCoefficients

void setCoefficients(double[] a)

Establish the final value of a[], that is, the coefficients of model

Parameters:

a - double array of final coefficients values

gradient

double[] gradient(double[][] x,
                  double[][] y)

Returns the gradient vector of the derivative of an error function (E) with respect to each coefficient of the model, using an input observation matrix (x[]) and an expected output matrix (y[]). Also returns the error associated.

Parameters:

x - Array with all inputs of all observations

y - Array with all expected outputs of all observations

Returns:

double Resulting gradient vector of dE/da for all coefficients

getLastError

double getLastError()

Last error of the model

Returns:

double Error of the function of the model with respect to data y[]