Package elki.clustering.kmeans.initialization.betula

Class CFKPlusPlusLeaves

java.lang.Object

elki.clustering.kmeans.initialization.betula.AbstractCFKMeansInitialization

elki.clustering.kmeans.initialization.betula.CFKPlusPlusLeaves

Direct Known Subclasses:

CFKPlusPlusTrunk

@Alias("leaves")
@Reference(authors="Andreas Lang and Erich Schubert",
           title="BETULA: Fast Clustering of Large Data with Improved BIRCH CF-Trees",
           booktitle="Information Systems",
           url="https://doi.org/10.1016/j.is.2021.101918",
           bibkey="DBLP:journals/is/LangS22")
public class CFKPlusPlusLeaves
extends AbstractCFKMeansInitialization

K-Means++-like initialization for BETULA k-means, treating the leaf clustering features as a flat list, and called "leaves" in the publication. To initialize regular k-means, use KMeansPlusPlus instead.

References:

Andreas Lang and Erich Schubert
BETULA: Fast Clustering of Large Data with Improved BIRCH CF-Trees
Information Systems

Since:

0.8.0

Author:

Andreas Lang

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	CFKPlusPlusLeaves.Par	Parameterization class.

Field Summary

Fields

Modifier and Type	Field	Description
protected CFInitWeight	distance	Distance function
protected boolean	firstUniform	Choose the first center uniformly from the leaves.

Fields inherited from class elki.clustering.kmeans.initialization.betula.AbstractCFKMeansInitialization

rf

Constructor Summary

Constructors

Constructor	Description
CFKPlusPlusLeaves(CFInitWeight dist, boolean firstUniform, RandomFactory rf)	Constructor.

Method Summary

Modifier and Type	Method	Description
double[][]	chooseInitialMeans(CFTree<?> tree, java.util.List<? extends ClusterFeature> cfs, int k)	Build the initial models.
private double	initialWeights(ClusterFeature first, java.util.List<? extends AsClusterFeature> cfs, double[] weights)	Initialize the weight list.
double[][]	run(CFTree<?> tree, java.util.List<? extends ClusterFeature> cfs, int k)	Perform k-means++ initialization.
private ClusterFeature	sampleFirst(ClusterFeature root, java.util.List<? extends AsClusterFeature> cfs, java.util.Random rnd)	Sample the first cluster center.
private double	updateWeights(ClusterFeature latest, java.util.List<? extends AsClusterFeature> cfs, double[] weights)	Update the weight list.

Methods inherited from class java.lang.Object

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

Field Detail

distance

protected CFInitWeight distance

Distance function

firstUniform

protected boolean firstUniform

Choose the first center uniformly from the leaves.

Constructor Detail

CFKPlusPlusLeaves

public CFKPlusPlusLeaves(CFInitWeight dist,
                         boolean firstUniform,
                         RandomFactory rf)

Constructor.

Parameters:

dist - distance function

firstUniform - choose the first center uniformly from leaves

rf - random generator

Method Detail

chooseInitialMeans

public double[][] chooseInitialMeans(CFTree<?> tree,
                                     java.util.List<? extends ClusterFeature> cfs,
                                     int k)

Description copied from class: AbstractCFKMeansInitialization

Build the initial models.

Specified by:

chooseInitialMeans in class AbstractCFKMeansInitialization

Parameters:

tree - CF tree

cfs - Cluster features of the tree (may be ignored for tree-based initializations, should be an array list for efficiency)

k - Number of clusters.

Returns:

initial cluster means

run

public double[][] run(CFTree<?> tree,
                      java.util.List<? extends ClusterFeature> cfs,
                      int k)

Perform k-means++ initialization.

Parameters:

tree - CFTree

cfs - Cluster features

k - K

Returns:

Initial cluster centers

sampleFirst

private ClusterFeature sampleFirst(ClusterFeature root,
                                   java.util.List<? extends AsClusterFeature> cfs,
                                   java.util.Random rnd)

Sample the first cluster center.

Parameters:

root - Root node of the tree

cfs - Cluster features to sample from

rnd - Random generator

Returns:

Selected cluster feature

initialWeights

private double initialWeights(ClusterFeature first,
                              java.util.List<? extends AsClusterFeature> cfs,
                              double[] weights)

Initialize the weight list.

Parameters:

first - Id of first mean.

cfs - Cluster features

weights - Weights output

Returns:

Sum of weights

updateWeights

private double updateWeights(ClusterFeature latest,
                             java.util.List<? extends AsClusterFeature> cfs,
                             double[] weights)

Update the weight list.

Parameters:

latest - Latest center

cfs - Cluster features

weights - Weights

Returns:

Weight sum