org.openimaj.ml.clustering.spectral

Class SpectralClusteringConf<DATATYPE>

java.lang.Object

org.openimaj.ml.clustering.spectral.SpectralClusteringConf<DATATYPE>

Type Parameters:

DATATYPE -

Direct Known Subclasses:

MultiviewSpectralClusteringConf

public class SpectralClusteringConf<DATATYPE>
extends Object

Author:

Sina Samangooei (ss@ecs.soton.ac.uk)

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static interface	SpectralClusteringConf.ClustererProvider<DATATYPE> A function which can represent itself as a string
protected static class	SpectralClusteringConf.DefaultClustererFunction<DATATYPE>

Field Summary

Fields

Modifier and Type	Field and Description
EigenChooser	eigenChooser The method used to select the number of eigen vectors from the lower valued eigenvalues
boolean	eigenValueScale
GraphLaplacian	laplacian The graph laplacian creator
int	skipEigenVectors

Constructor Summary

Constructors

Constructor and Description
SpectralClusteringConf(SpatialClusterer<? extends SpatialClusters<DATATYPE>,DATATYPE> internal) The underlying EigenChooser is set to an ChangeDetectingEigenChooser which looks for a 100x gap between eigen vectors to select number of clusters.
SpectralClusteringConf(SpatialClusterer<? extends SpatialClusters<DATATYPE>,DATATYPE> internal, GraphLaplacian laplacian) The underlying EigenChooser is set to an ChangeDetectingEigenChooser which looks for a 100x gap between eigen vectors to select number of clusters.
SpectralClusteringConf(SpatialClusterer<? extends SpatialClusters<DATATYPE>,DATATYPE> internal, GraphLaplacian lap, int top)
SpectralClusteringConf(SpatialClusterer<? extends SpatialClusters<DATATYPE>,DATATYPE> internal, int eigK)
SpectralClusteringConf(SpectralClusteringConf.ClustererProvider<DATATYPE> internal) The underlying EigenChooser is set to an ChangeDetectingEigenChooser which looks for a 100x gap between eigen vectors to select number of clusters.

Method Summary

Methods inherited from class java.lang.Object

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

Field Detail

laplacian

public GraphLaplacian laplacian

The graph laplacian creator

eigenChooser

public EigenChooser eigenChooser

The method used to select the number of eigen vectors from the lower valued eigenvalues

skipEigenVectors

public int skipEigenVectors

eigenValueScale

public boolean eigenValueScale

Constructor Detail

SpectralClusteringConf

public SpectralClusteringConf(SpatialClusterer<? extends SpatialClusters<DATATYPE>,DATATYPE> internal,
                      int eigK)

Parameters:

internal - the internal clusterer

eigK - the value handed to HardCodedEigenChooser

SpectralClusteringConf

public SpectralClusteringConf(SpatialClusterer<? extends SpatialClusters<DATATYPE>,DATATYPE> internal)

The underlying EigenChooser is set to an ChangeDetectingEigenChooser which looks for a 100x gap between eigen vectors to select number of clusters. It also insists upon a maximum of 0.1 * number of data items (so 10 items per cluster)

Parameters:

internal - the internal clusterer

SpectralClusteringConf

public SpectralClusteringConf(SpatialClusterer<? extends SpatialClusters<DATATYPE>,DATATYPE> internal,
                      GraphLaplacian lap,
                      int top)

Parameters:

internal - an internal clusterer

lap - the laplacian

top - the top eigen vectors

SpectralClusteringConf

public SpectralClusteringConf(SpatialClusterer<? extends SpatialClusters<DATATYPE>,DATATYPE> internal,
                      GraphLaplacian laplacian)

The underlying EigenChooser is set to an ChangeDetectingEigenChooser which looks for a 100x gap between eigen vectors to select number of clusters. It also insists upon a maximum of 0.1 * number of data items (so 10 items per cluster)

Parameters:

internal - the internal clusterer

laplacian - the graph laplacian

SpectralClusteringConf

public SpectralClusteringConf(SpectralClusteringConf.ClustererProvider<DATATYPE> internal)

The underlying EigenChooser is set to an ChangeDetectingEigenChooser which looks for a 100x gap between eigen vectors to select number of clusters. It also insists upon a maximum of 0.1 * number of data items (so 10 items per cluster)

Parameters:

internal - the internal clusterer