vt::SimpleKmeans< Feature, Distance, FeatureAllocator > Class Template Reference

Class for performing K-means clustering, optimized for a particular feature type and metric. More...

#include <simple_kmeans.h>

List of all members.

Public Types
typedef boost::function< void(const std::vector< Feature * > &, size_t, std::vector < Feature, FeatureAllocator > &, Distance)	Initializer )
typedef Distance::result_type	squared_distance_type
Public Member Functions
squared_distance_type	cluster (const std::vector< Feature, FeatureAllocator > &features, size_t k, std::vector< Feature, FeatureAllocator > &centers, std::vector< unsigned int > &membership) const
	Partition a set of features into k clusters.
squared_distance_type	clusterPointers (const std::vector< Feature * > &features, size_t k, std::vector< Feature, FeatureAllocator > &centers, std::vector< unsigned int > &membership) const
	Partition a set of features into k clusters.
size_t	getMaxIterations () const
size_t	getRestarts () const
void	setInitMethod (const Initializer &init)
	Set function object used to choose initial cluster centers.
void	setMaxIterations (size_t iters)
void	setRestarts (size_t restarts)
	SimpleKmeans (const Feature &zero=Feature(), Distance d=Distance())
	Constructor.
Private Member Functions
squared_distance_type	clusterOnce (const std::vector< Feature * > &features, size_t k, std::vector< Feature, FeatureAllocator > &centers, std::vector< unsigned int > &membership) const
Private Attributes
Initializer	choose_centers_
Distance	distance_
size_t	max_iterations_
size_t	restarts_
Feature	zero_

---

Detailed Description

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>
class vt::SimpleKmeans< Feature, Distance, FeatureAllocator >

Class for performing K-means clustering, optimized for a particular feature type and metric.

Todo:

InitKmeanspp

The standard Lloyd's algorithm is used. By default, cluster centers are initialized randomly.

Definition at line 26 of file simple_kmeans.h.

---

Member Typedef Documentation

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

typedef boost::function<void(const std::vector<Feature*>&, size_t, std::vector<Feature, FeatureAllocator>&, Distance) vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::Initializer)

Definition at line 30 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

typedef Distance::result_type vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::squared_distance_type

Definition at line 29 of file simple_kmeans.h.

---

Constructor & Destructor Documentation

template<class Feature , class Distance , class FeatureAllocator >

vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::SimpleKmeans	(	const Feature &	zero = Feature(),
		Distance	d = Distance()
	)

Constructor.

Parameters:

zero	Object representing zero in the feature space
d	Functor for calculating squared distance

Todo:

FeatureAllocator parameter

Definition at line 93 of file simple_kmeans.h.

---

Member Function Documentation

template<class Feature , class Distance , class FeatureAllocator >

SimpleKmeans< Feature, Distance, FeatureAllocator >::squared_distance_type vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::cluster	(	const std::vector< Feature, FeatureAllocator > &	features,
		size_t	k,
		std::vector< Feature, FeatureAllocator > &	centers,
		std::vector< unsigned int > &	membership
	)		const

Partition a set of features into k clusters.

Parameters:

	features	The features to be clustered.
	k	The number of clusters.
[out]	centers	A set of k cluster centers.
[out]	membership	Cluster assignment for each feature

Definition at line 104 of file simple_kmeans.h.

template<class Feature , class Distance , class FeatureAllocator >

SimpleKmeans< Feature, Distance, FeatureAllocator >::squared_distance_type vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::clusterOnce	(	const std::vector< Feature * > &	features,
		size_t	k,
		std::vector< Feature, FeatureAllocator > &	centers,
		std::vector< unsigned int > &	membership
	)		const [private]

Todo:

Kahan summation?

Definition at line 142 of file simple_kmeans.h.

template<class Feature , class Distance , class FeatureAllocator >

SimpleKmeans< Feature, Distance, FeatureAllocator >::squared_distance_type vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::clusterPointers	(	const std::vector< Feature * > &	features,
		size_t	k,
		std::vector< Feature, FeatureAllocator > &	centers,
		std::vector< unsigned int > &	membership
	)		const

Partition a set of features into k clusters.

This version is more convenient for hierarchical clustering, as you do not have to copy feature objects.

Parameters:

	features	The features to be clustered.
	k	The number of clusters.
[out]	centers	A set of k cluster centers.
[out]	membership	Cluster assignment for each feature

Definition at line 117 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

size_t vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::getMaxIterations

(

)

const [inline]

Definition at line 45 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

size_t vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::getRestarts

(

)

const [inline]

Definition at line 48 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

void vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::setInitMethod

(

const Initializer &

init

)

[inline]

Set function object used to choose initial cluster centers.

Definition at line 43 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

void vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::setMaxIterations

(

size_t

iters

)

[inline]

Definition at line 46 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

void vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::setRestarts

(

size_t

restarts

)

[inline]

Definition at line 49 of file simple_kmeans.h.

---

Member Data Documentation

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

Initializer vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::choose_centers_ [private]

Definition at line 86 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

Distance vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::distance_ [private]

Definition at line 85 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

size_t vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::max_iterations_ [private]

Definition at line 87 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

size_t vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::restarts_ [private]

Definition at line 88 of file simple_kmeans.h.

template<class Feature, class Distance = distance::L2<Feature>, class FeatureAllocator = typename DefaultAllocator<Feature>::type>

Feature vt::SimpleKmeans< Feature, Distance, FeatureAllocator >::zero_ [private]

Definition at line 84 of file simple_kmeans.h.

---

The documentation for this class was generated from the following file:

• simple_kmeans.h