de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans

Class KMeansHamerly.Instance

java.lang.Object

de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans.AbstractKMeans.Instance

de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans.KMeansHamerly.Instance

Direct Known Subclasses:

KMeansAnnulus.Instance, KMeansExponion.Instance

Enclosing class:

KMeansHamerly<V extends NumberVector>

protected static class KMeansHamerly.Instance
extends AbstractKMeans.Instance

Inner instance, storing state for a single data set.

Author:

Erich Schubert

Field Summary

Fields

Modifier and Type	Field and Description
(package private) WritableDoubleDataStore	lower Lower bounding distance
(package private) double[][]	newmeans Temporary storage for the new means.
(package private) double[]	sep Separation of means / distance moved.
(package private) double[][]	sums Sum aggregate for the new mean.
(package private) WritableDoubleDataStore	upper Upper bounding distance

Fields inherited from class de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans.AbstractKMeans.Instance

assignment, clusters, isSquared, k, key, means, relation, varsum

Constructor Summary

Constructors

Constructor and Description
Instance(Relation<? extends NumberVector> relation, NumberVectorDistanceFunction<?> df, double[][] means) Constructor.

Method Summary

Modifier and Type	Method and Description
protected int	assignToNearestCluster() Reassign objects, but avoid unnecessary computations based on their bounds.
protected Logging	getLogger() Get the class logger.
protected int	initialAssignToNearestCluster() Perform initial cluster assignment.
protected int	iterate(int iteration) Main loop function.
protected void	recomputeSeperation(double[][] means, double[] sep) Recompute the separation of cluster means.
protected void	updateBounds(double[] move, double delta) Update the bounds for k-means.

Methods inherited from class de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans.AbstractKMeans.Instance

buildResult, buildResult, copyMeans, distance, isSquared, meansFromSums, movedDistance, recomputeSeperation, run

Methods inherited from class java.lang.Object

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

Field Detail

sums

double[][] sums

Sum aggregate for the new mean.

newmeans

double[][] newmeans

Temporary storage for the new means.

sep

double[] sep

Separation of means / distance moved.

upper

WritableDoubleDataStore upper

Upper bounding distance

lower

WritableDoubleDataStore lower

Lower bounding distance

Constructor Detail

Instance

public Instance(Relation<? extends NumberVector> relation,
                NumberVectorDistanceFunction<?> df,
                double[][] means)

Constructor.

Parameters:

relation - Relation

means - Initial means

Method Detail

iterate

protected int iterate(int iteration)

Description copied from class: AbstractKMeans.Instance

Main loop function.

Specified by:

iterate in class AbstractKMeans.Instance

Parameters:

iteration - Iteration number (beginning at 1)

Returns:

Number of reassigned points

initialAssignToNearestCluster

protected int initialAssignToNearestCluster()

Perform initial cluster assignment.

Returns:

Number of changes (i.e. relation size)

assignToNearestCluster

protected int assignToNearestCluster()

Reassign objects, but avoid unnecessary computations based on their bounds.

Overrides:

assignToNearestCluster in class AbstractKMeans.Instance

Returns:

number of objects reassigned

recomputeSeperation

protected void recomputeSeperation(double[][] means,
                                   double[] sep)

Recompute the separation of cluster means.

Used by Hamerly.

Parameters:

means - Means

sep - Output array of separation (half-sqrt scaled)

updateBounds

protected void updateBounds(double[] move,
                            double delta)

Update the bounds for k-means.

Parameters:

move - Movement of centers

delta - Maximum center movement.

getLogger

protected Logging getLogger()

Description copied from class: AbstractKMeans.Instance

Get the class logger.

Specified by:

getLogger in class AbstractKMeans.Instance

Returns:

Logger