jsat.clustering.kmeans

Class ElkanKMeans

java.lang.Object

jsat.clustering.ClustererBase

jsat.clustering.KClustererBase

jsat.clustering.kmeans.KMeans

jsat.clustering.kmeans.ElkanKMeans

All Implemented Interfaces:

java.io.Serializable, Clusterer, KClusterer, Parameterized

public class ElkanKMeans
extends KMeans

An efficient implementation of the K-Means algorithm. This implementation uses the triangle inequality to accelerate computation while maintaining the exact same solution. This requires that the DistanceMetric used support DistanceMetric.isSubadditive().
Implementation based on the paper: Using the Triangle Inequality to Accelerate k-Means, by Charles Elkan

See Also:

Serialized Form

Field Summary

Fields inherited from class jsat.clustering.kmeans.KMeans

DEFAULT_SEED_SELECTION

Constructor Summary

Constructors

Constructor and Description
ElkanKMeans() Creates a new KMeans instance.
ElkanKMeans(DistanceMetric dm) Creates a new KMeans instance
ElkanKMeans(DistanceMetric dm, java.util.Random rand) Creates a new KMeans instance
ElkanKMeans(DistanceMetric dm, java.util.Random rand, SeedSelectionMethods.SeedSelection seedSelection) Creates a new KMeans instance.
ElkanKMeans(ElkanKMeans toCopy)

Method Summary

Modifier and Type	Method and Description
ElkanKMeans	clone()
boolean	isUseDenseSparse() Returns if Dense Sparse acceleration will be used if available
void	setUseDenseSparse(boolean useDenseSparse) Sets whether or not to use DenseSparseMetric when computing.

Methods inherited from class jsat.clustering.kmeans.KMeans

cluster, cluster, cluster, cluster, getDistanceMetric, getIterationLimit, getMeans, getSeedSelection, setIterationLimit, setSeedSelection, setStoreMeans, supportsWeightedData

Methods inherited from class jsat.clustering.ClustererBase

createClusterListFromAssignmentArray, getDatapointsFromCluster

Methods inherited from class java.lang.Object

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

Methods inherited from interface jsat.parameters.Parameterized

getParameter, getParameters

Methods inherited from interface jsat.clustering.KClusterer

cluster, cluster, cluster, cluster, cluster, cluster

Methods inherited from interface jsat.clustering.Clusterer

cluster, cluster

Constructor Detail

ElkanKMeans

public ElkanKMeans(DistanceMetric dm,
                   java.util.Random rand,
                   SeedSelectionMethods.SeedSelection seedSelection)

Creates a new KMeans instance.

Parameters:

dm - the distance metric to use, must support DistanceMetric.isSubadditive().

rand - the random number generator to use during seed selection

seedSelection - the method of seed selection to use

ElkanKMeans

public ElkanKMeans(DistanceMetric dm,
                   java.util.Random rand)

Creates a new KMeans instance

Parameters:

dm - the distance metric to use, must support DistanceMetric.isSubadditive().

rand - the random number generator to use during seed selection

ElkanKMeans

public ElkanKMeans(DistanceMetric dm)

Creates a new KMeans instance

Parameters:

dm - the distance metric to use, must support DistanceMetric.isSubadditive().

ElkanKMeans

public ElkanKMeans()

Creates a new KMeans instance. The EuclideanDistance will be used by default.

ElkanKMeans

public ElkanKMeans(ElkanKMeans toCopy)

Method Detail

setUseDenseSparse

public void setUseDenseSparse(boolean useDenseSparse)

Sets whether or not to use DenseSparseMetric when computing. This may or may not provide a speed increase.

Parameters:

useDenseSparse - whether or not to compute the distance from dense mean vectors to sparse ones using acceleration

isUseDenseSparse

public boolean isUseDenseSparse()

Returns if Dense Sparse acceleration will be used if available

Returns:

if Dense Sparse acceleration will be used if available

clone

public ElkanKMeans clone()

Specified by:

clone in interface Clusterer

Specified by:

clone in interface KClusterer

Specified by:

clone in class KMeans