org.joone.engine.listeners

Class ErrorBasedConvergenceObserver

java.lang.Object

org.joone.util.MonitorPlugin

org.joone.engine.listeners.ConvergenceObserver

org.joone.engine.listeners.ErrorBasedConvergenceObserver

All Implemented Interfaces:

java.io.Serializable, java.util.EventListener, NeuralNetListener

public class ErrorBasedConvergenceObserver
extends ConvergenceObserver

This observer observes if the network has convergenced based on the sequence of training errors.

Author:

Boris Jansen

See Also:

Serialized Form

Field Summary

Fields inherited from class org.joone.engine.listeners.ConvergenceObserver

disableCurrentConvergence

Constructor Summary

Constructors

Constructor and Description
ErrorBasedConvergenceObserver() Creates a new instance of ErrorBasedConvergenceObserver

Method Summary

Methods

Modifier and Type	Method and Description
int	getCycles() Gets the number of cycles over which convergence is checked.
double	getPercentage() Gets the percentage.
protected void	manageCycle(Monitor mon)
protected void	manageError(Monitor mon)
protected void	manageStart(Monitor mon)
protected void	manageStop(Monitor mon)
protected void	manageStopError(Monitor mon, java.lang.String msgErr)
void	setCycles(int aCylces) Sets the number of cycles.
void	setPercentage(double aPercentage) Sets the percentage.

Methods inherited from class org.joone.engine.listeners.ConvergenceObserver

addConvergenceListener, disableCurrentConvergence, fireNetConverged, removeConvergenceListener

Methods inherited from class org.joone.util.MonitorPlugin

cicleTerminated, errorChanged, getName, getNeuralNet, getRate, netStarted, netStopped, netStoppedError, setName, setNeuralNet, setRate, toBeManaged

Methods inherited from class java.lang.Object

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

Constructor Detail

ErrorBasedConvergenceObserver

public ErrorBasedConvergenceObserver()

Creates a new instance of ErrorBasedConvergenceObserver

Method Detail

setPercentage

public void setPercentage(double aPercentage)

Sets the percentage. Whenever a neural network's training error doesn't decrease more than this percentage for a couple of steps in a sequence of training errors, the network is considered as converged.

Parameters:

aPercentage - the percentage to set.

getPercentage

public double getPercentage()

Gets the percentage.

Returns:

the percentage.

setCycles

public void setCycles(int aCylces)

Sets the number of cycles. Whenever the training error of a network doesn't decrease more than a percentage for this number of cycles, the network is considered as converged.

Parameters:

aCylces -

getCycles

public int getCycles()

Gets the number of cycles over which convergence is checked.

Returns:

the number of cycles.

manageStop

protected void manageStop(Monitor mon)

Specified by:

manageStop in class MonitorPlugin

manageCycle

protected void manageCycle(Monitor mon)

Specified by:

manageCycle in class MonitorPlugin

manageStart

protected void manageStart(Monitor mon)

Specified by:

manageStart in class MonitorPlugin

manageError

protected void manageError(Monitor mon)

Specified by:

manageError in class MonitorPlugin

manageStopError

protected void manageStopError(Monitor mon,
                   java.lang.String msgErr)

Specified by:

manageStopError in class MonitorPlugin