de.htwdd.rosenkoenig.neuro.net
Class NeighbourhoodFunction

java.lang.Object
  de.htwdd.rosenkoenig.neuro.net.NeighbourhoodFunction

Direct Known Subclasses:

Cone, Cosine, Gauss, WinnerTakesAll

public abstract class NeighbourhoodFunction
extends java.lang.Object

Base class to provide a KohonenNet with a neighbourhood function. After the winning neuron has been calculated, the Kohonen map uses a neighbourhood function to determine surrounding neurons that also have to be activated.

Field Summary

protected double

radius The neighbourhood radius.

Constructor Summary

NeighbourhoodFunction()
Default constructor.

Method Summary

abstract double	computeActivation(double distance) All neighbourhood functions have to implement this method.
double	getRadius() Getter of the property radius
void	setRadius(double radius) Setter of the property radius

Methods inherited from class java.lang.Object

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

Field Detail

radius

protected double radius

The neighbourhood radius.

Constructor Detail

NeighbourhoodFunction

public NeighbourhoodFunction()

Default constructor.

Method Detail

getRadius

public double getRadius()

Getter of the property radius

Returns:

Returns the radius.

setRadius

public void setRadius(double radius)

Setter of the property radius

Parameters:

radius - The radius to set.

computeActivation

public abstract double computeActivation(double distance)

All neighbourhood functions have to implement this method. Computes the activation of a neuran with a certain distance to the winning neuron.

Parameters:

distance - euclidic distance to the winning neuron

Returns:

activation for the given distance