Package org.openimaj.ml.clustering.kmeans

K-Means in OpenIMAJ is designed to be both extremely fast and flexible.

See: Description

Class Summary

Class	Description
ByteKMeans	Fast, parallel implementation of the K-Means algorithm with support for bigger-than-memory data.
ByteKMeansInit	Initialisation for K-Means clustering.
ByteKMeansInit.RANDOM	Simple kmeans initialized on randomly selected samples.
DoubleKMeans	Fast, parallel implementation of the K-Means algorithm with support for bigger-than-memory data.
DoubleKMeansInit	Initialisation for K-Means clustering.
DoubleKMeansInit.RANDOM	Simple kmeans initialized on randomly selected samples.
FloatKMeans	Fast, parallel implementation of the K-Means algorithm with support for bigger-than-memory data.
FloatKMeansInit	Initialisation for K-Means clustering.
FloatKMeansInit.RANDOM	Simple kmeans initialized on randomly selected samples.
HierarchicalByteKMeans	Hierarchical Byte K-Means clustering (HierarchicalByteKMeans) is a simple hierarchical version of ByteKMeans.
HierarchicalByteKMeansResult	The result of a HierarchicalByteKMeans clustering operation.
HierarchicalByteKMeansResult.Node	HierarchicalByteKMeans tree node The number of children is not bigger than the HierarchicalByteKMeans K parameter
HierarchicalDoubleKMeans	Hierarchical Double K-Means clustering (HierarchicalDoubleKMeans) is a simple hierarchical version of DoubleKMeans.
HierarchicalDoubleKMeansResult	The result of a HierarchicalDoubleKMeans clustering operation.
HierarchicalDoubleKMeansResult.Node	HierarchicalDoubleKMeans tree node The number of children is not bigger than the HierarchicalDoubleKMeans K parameter
HierarchicalFloatKMeans	Hierarchical Float K-Means clustering (HierarchicalFloatKMeans) is a simple hierarchical version of FloatKMeans.
HierarchicalFloatKMeansResult	The result of a HierarchicalFloatKMeans clustering operation.
HierarchicalFloatKMeansResult.Node	HierarchicalFloatKMeans tree node The number of children is not bigger than the HierarchicalFloatKMeans K parameter
HierarchicalIntKMeans	Hierarchical Integer K-Means clustering (HierarchicalIntKMeans) is a simple hierarchical version of IntKMeans.
HierarchicalIntKMeansResult	The result of a HierarchicalIntKMeans clustering operation.
HierarchicalIntKMeansResult.Node	HierarchicalIntKMeans tree node The number of children is not bigger than the HierarchicalIntKMeans K parameter
HierarchicalLongKMeans	Hierarchical Long K-Means clustering (HierarchicalLongKMeans) is a simple hierarchical version of LongKMeans.
HierarchicalLongKMeansResult	The result of a HierarchicalLongKMeans clustering operation.
HierarchicalLongKMeansResult.Node	HierarchicalLongKMeans tree node The number of children is not bigger than the HierarchicalLongKMeans K parameter
HierarchicalShortKMeans	Hierarchical Short K-Means clustering (HierarchicalShortKMeans) is a simple hierarchical version of ShortKMeans.
HierarchicalShortKMeansResult	The result of a HierarchicalShortKMeans clustering operation.
HierarchicalShortKMeansResult.Node	HierarchicalShortKMeans tree node The number of children is not bigger than the HierarchicalShortKMeans K parameter
IntKMeans	Fast, parallel implementation of the K-Means algorithm with support for bigger-than-memory data.
IntKMeansInit	Initialisation for K-Means clustering.
IntKMeansInit.RANDOM	Simple kmeans initialized on randomly selected samples.
KMeansConfiguration<NN extends NearestNeighbours<DATA,?,?>,DATA>	Configuration for the K-Means algorithm.
LongKMeans	Fast, parallel implementation of the K-Means algorithm with support for bigger-than-memory data.
LongKMeansInit	Initialisation for K-Means clustering.
LongKMeansInit.RANDOM	Simple kmeans initialized on randomly selected samples.
ShortKMeans	Fast, parallel implementation of the K-Means algorithm with support for bigger-than-memory data.
ShortKMeansInit	Initialisation for K-Means clustering.
ShortKMeansInit.RANDOM	Simple kmeans initialized on randomly selected samples.

Package org.openimaj.ml.clustering.kmeans Description

K-Means in OpenIMAJ is designed to be both extremely fast and flexible. To this end, we've separated out the cluster-assignment part of the algorithm to be performed by a separate object. This means that the core implementation (initial assignment, followed by iterative cluster assignment and centroid averaging) can use any kind of cluster assignment approach. There are a number of assignment techniques that can be used, including exact assignment and approximate assignment based on KD-Tree ensembles.