mlpack::kmeans::DualTreeKMeans< MetricType, MatType, TreeType > Class Template Reference

An algorithm for an exact Lloyd iteration which simply uses dual-tree nearest-neighbor search to find the nearest centroid for each point in the dataset. More...

#include <dual_tree_kmeans.hpp>

Public Types
typedef TreeType< MetricType, DualTreeKMeansStatistic, MatType >	Tree
	Convenience typedef.
template<typename TreeMetricType , typename IgnoredStatType , typename TreeMatType >
using	NNSTreeType = TreeType< TreeMetricType, DualTreeKMeansStatistic, TreeMatType >

Public Member Functions
	DualTreeKMeans (const MatType &dataset, MetricType &metric)
	Construct the DualTreeKMeans object, which will construct a tree on the points.
	~DualTreeKMeans ()
	Delete the tree constructed by the DualTreeKMeans object.
double	Iterate (const arma::mat &centroids, arma::mat &newCentroids, arma::Col< size_t > &counts)
	Run a single iteration of the dual-tree nearest neighbor algorithm for k-means, updating the given centroids into the newCentroids matrix. More...
size_t	DistanceCalculations () const
	Return the number of distance calculations.
size_t &	DistanceCalculations ()
	Modify the number of distance calculations.

Detailed Description

template<typename MetricType, typename MatType, template< typename TreeMetricType, typename TreeStatType, typename TreeMatType > class TreeType = tree::KDTree>
class mlpack::kmeans::DualTreeKMeans< MetricType, MatType, TreeType >

An algorithm for an exact Lloyd iteration which simply uses dual-tree nearest-neighbor search to find the nearest centroid for each point in the dataset.

The conditions under which this will perform best are probably limited to the case where k is close to the number of points in the dataset, and the number of iterations of the k-means algorithm will be few.

Member Function Documentation

Iterate()

template<typename MetricType , typename MatType , template< typename TreeMetricType, typename TreeStatType, typename TreeMatType > class TreeType>

double mlpack::kmeans::DualTreeKMeans< MetricType, MatType, TreeType >::Iterate	(	const arma::mat &	centroids,
		arma::mat &	newCentroids,
		arma::Col< size_t > &	counts
	)

Run a single iteration of the dual-tree nearest neighbor algorithm for k-means, updating the given centroids into the newCentroids matrix.

Parameters

centroids	Current cluster centroids.
newCentroids	New cluster centroids.
counts	Current counts, to be overwritten with new counts.

---

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

• src/mlpack/methods/kmeans/dual_tree_kmeans.hpp
• src/mlpack/methods/kmeans/dual_tree_kmeans_impl.hpp