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

Class KMeansElkan<V extends NumberVector>

java.lang.Object

de.lmu.ifi.dbs.elki.algorithm.AbstractAlgorithm<R>

de.lmu.ifi.dbs.elki.algorithm.AbstractNumberVectorDistanceBasedAlgorithm<V,Clustering<M>>

de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans.AbstractKMeans<V,KMeansModel>

de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans.KMeansElkan<V>

Type Parameters:

V - vector datatype

All Implemented Interfaces:

Algorithm, ClusteringAlgorithm<Clustering<KMeansModel>>, KMeans<V,KMeansModel>, DistanceBasedAlgorithm<V>

@Reference(authors="C. Elkan",
           title="Using the triangle inequality to accelerate k-means",
           booktitle="Proc. 20th International Conference on Machine Learning, ICML 2003",
           url="http://www.aaai.org/Library/ICML/2003/icml03-022.php")
public class KMeansElkan<V extends NumberVector>
extends AbstractKMeans<V,KMeansModel>

Elkan's fast k-means by exploiting the triangle inequality. This variant needs O(n*k) additional memory to store bounds. See KMeansHamerly for a close variant that only uses O(n*2) additional memory for bounds.

Reference:
C. Elkan
Using the triangle inequality to accelerate k-means
Proc. 20th International Conference on Machine Learning, ICML 2003

Author:

Erich Schubert

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	KMeansElkan.Parameterizer<V extends NumberVector> Parameterization class.

Field Summary

Fields

Modifier and Type	Field and Description
private static String	KEY Key for statistics logging.
private static Logging	LOG The logger for this class.

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

initializer, k, maxiter

Fields inherited from class de.lmu.ifi.dbs.elki.algorithm.AbstractNumberVectorDistanceBasedAlgorithm

distanceFunction

Fields inherited from interface de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans.KMeans

INIT_ID, K_ID, MAXITER_ID, SEED_ID

Fields inherited from interface de.lmu.ifi.dbs.elki.algorithm.DistanceBasedAlgorithm

DISTANCE_FUNCTION_ID

Constructor Summary

Constructors

Constructor and Description
KMeansElkan(NumberVectorDistanceFunction<? super V> distanceFunction, int k, int maxiter, KMeansInitialization<? super V> initializer) Constructor.

Method Summary

Methods

Modifier and Type	Method and Description
private int	assignToNearestCluster(Relation<V> relation, List<Vector> means, List<Vector> sums, List<ModifiableDBIDs> clusters, WritableIntegerDataStore assignment, double[] sep, double[][] cdist, WritableDoubleDataStore upper, WritableDataStore<double[]> lower) Reassign objects, but only if their bounds indicate it is necessary to do so.
protected Logging	getLogger() Get the (STATIC) logger for this class.
private int	initialAssignToNearestCluster(Relation<V> relation, List<Vector> means, List<Vector> sums, List<ModifiableDBIDs> clusters, WritableIntegerDataStore assignment, WritableDoubleDataStore upper, WritableDataStore<double[]> lower) Reassign objects, but only if their bounds indicate it is necessary to do so.
private double	maxMoved(List<Vector> means, List<Vector> newmeans, double[] dists) Maximum distance moved.
private void	recomputeSeperation(List<Vector> means, double[] sep, double[][] cdist) Recompute the separation of cluster means.
Clustering<KMeansModel>	run(Database database, Relation<V> relation) Run the clustering algorithm.
private void	updateBounds(Relation<V> relation, WritableIntegerDataStore assignment, WritableDoubleDataStore upper, WritableDataStore<double[]> lower, double[] move) Update the bounds for k-means.

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

assignToNearestCluster, getInputTypeRestriction, incrementalUpdateMean, logVarstat, macQueenIterate, means, medians, setDistanceFunction, setK

Methods inherited from class de.lmu.ifi.dbs.elki.algorithm.AbstractNumberVectorDistanceBasedAlgorithm

getDistanceFunction

Methods inherited from class de.lmu.ifi.dbs.elki.algorithm.AbstractAlgorithm

makeParameterDistanceFunction, run

Methods inherited from class java.lang.Object

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

Methods inherited from interface de.lmu.ifi.dbs.elki.algorithm.clustering.ClusteringAlgorithm

run

Methods inherited from interface de.lmu.ifi.dbs.elki.algorithm.DistanceBasedAlgorithm

getDistanceFunction

Field Detail

LOG

private static final Logging LOG

The logger for this class.

KEY

private static final String KEY

Key for statistics logging.

Constructor Detail

KMeansElkan

public KMeansElkan(NumberVectorDistanceFunction<? super V> distanceFunction,
           int k,
           int maxiter,
           KMeansInitialization<? super V> initializer)

Constructor.

Parameters:

distanceFunction - distance function

k - k parameter

maxiter - Maxiter parameter

initializer - Initialization method

Method Detail

run

public Clustering<KMeansModel> run(Database database,
                          Relation<V> relation)

Description copied from interface: KMeans

Run the clustering algorithm.

Parameters:

database - Database to run on.

relation - Relation to process.

Returns:

Clustering result

recomputeSeperation

private void recomputeSeperation(List<Vector> means,
                       double[] sep,
                       double[][] cdist)

Recompute the separation of cluster means.

Parameters:

means - Means

sep - Output array of separation

cdist - Center-to-Center distances

initialAssignToNearestCluster

private int initialAssignToNearestCluster(Relation<V> relation,
                                List<Vector> means,
                                List<Vector> sums,
                                List<ModifiableDBIDs> clusters,
                                WritableIntegerDataStore assignment,
                                WritableDoubleDataStore upper,
                                WritableDataStore<double[]> lower)

Reassign objects, but only if their bounds indicate it is necessary to do so.

Parameters:

relation - Data

means - Current means

sums - New means

clusters - Current clusters

assignment - Cluster assignment

upper - Upper bounds

lower - Lower bounds

Returns:

Number of changes (i.e. relation size)

assignToNearestCluster

private int assignToNearestCluster(Relation<V> relation,
                         List<Vector> means,
                         List<Vector> sums,
                         List<ModifiableDBIDs> clusters,
                         WritableIntegerDataStore assignment,
                         double[] sep,
                         double[][] cdist,
                         WritableDoubleDataStore upper,
                         WritableDataStore<double[]> lower)

Reassign objects, but only if their bounds indicate it is necessary to do so.

Parameters:

relation - Data

means - Current means

sums - New means

clusters - Current clusters

assignment - Cluster assignment

sep - Separation of means

cdist - Center-to-center distances

upper - Upper bounds

lower - Lower bounds

Returns:

true when the object was reassigned

maxMoved

private double maxMoved(List<Vector> means,
              List<Vector> newmeans,
              double[] dists)

Maximum distance moved.

Parameters:

means - Old means

newmeans - New means

dists - Distances moved

Returns:

Maximum distance moved

updateBounds

private void updateBounds(Relation<V> relation,
                WritableIntegerDataStore assignment,
                WritableDoubleDataStore upper,
                WritableDataStore<double[]> lower,
                double[] move)

Update the bounds for k-means.

Parameters:

relation - Relation

assignment - Cluster assignment

upper - Upper bounds

lower - Lower bounds

move - Movement of centers

getLogger

protected Logging getLogger()

Description copied from class: AbstractAlgorithm

Get the (STATIC) logger for this class.

Specified by:

getLogger in class AbstractAlgorithm<Clustering<KMeansModel>>

Returns:

the static logger