databionics.esom.neighborhood
Class Neighborhood

java.lang.Object
  databionics.esom.neighborhood.Neighborhood

Direct Known Subclasses:

BubbleNeighborhood, CachedNeighborhood

public abstract class Neighborhood
extends java.lang.Object

Abstract base class for neighborhood kernels on a SOM grid. Subclasses need to override the calculate(int distance) method.

On a grid only integer distances from the center need to be considered. The kernels are assumed to be radial symmetric, so only a one dimensional array for the weights is needed.

Field Summary

protected static org.apache.log4j.Logger	log interface to log4j system
protected int	radius maximum distance from bestmatch on grid to be considered
protected double	scaling scaling factor for all weights

Constructor Summary

Neighborhood()
Create a new neighborhood object

Neighborhood(int radius, double scaling)
Create a new neighborhood object

Method Summary

protected abstract double	calculate(int distance) Calculate the weight value for this distance.
double	get(int distance) Get weight for distance on grid from kernel center
abstract void	init() Initialize the neighborhood.
void	setRadius(int radius) Set the maximum distance from bestmatch on grid to be considered
void	setScaling(double scaling) Set the scaling factor for all weights

Methods inherited from class java.lang.Object

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

Field Detail

log

protected static final org.apache.log4j.Logger log

interface to log4j system

radius

protected int radius

maximum distance from bestmatch on grid to be considered

scaling

protected double scaling

scaling factor for all weights

Constructor Detail

Neighborhood

public Neighborhood()

Create a new neighborhood object

Neighborhood

public Neighborhood(int radius,
                    double scaling)

Create a new neighborhood object

Parameters:

radius - radius of kernel

scaling - scaling factor for all weights

Method Detail

calculate

protected abstract double calculate(int distance)

Calculate the weight value for this distance. Override this method in subclasses.

Parameters:

distance - distance from center ([0..radius])

Returns:

weight for this distance

init

public abstract void init()

Initialize the neighborhood. Call this after setting radius and scaling

get

public double get(int distance)

Get weight for distance on grid from kernel center

Parameters:

distance - distance from center

Returns:

weight for this distance

setRadius

public void setRadius(int radius)

Set the maximum distance from bestmatch on grid to be considered

Parameters:

radius - radius

setScaling

public void setScaling(double scaling)

Set the scaling factor for all weights

Parameters:

scaling - scaling