Clustering functions. More...

#include "cluster.h"
#include "utilities.h"

Functions
vector< long >	k_means (long n, float *data, long k)
	Generate clusters using a K-means algorithm. More...

vector< long >	affin_prop_clustering (Matrix s, long maxit, long convit, double lambda, long &ncluster)
	Generate clusters from a similarity matrix using affinity propagation. More...

Variables
int	verbose

Detailed Description

Clustering functions.

Author: Bernard Heymann

Date: Created: 20070417; Modified: 20210508

Function Documentation

◆ affin_prop_clustering()

vector< long > affin_prop_clustering	(	Matrix	s,
		long	maxit,
		long	convit,
		double	lambda,
		long &	ncluster
	)

Generate clusters from a similarity matrix using affinity propagation.

Parameters

s	n x n similarity matrix.
maxit	maximum iterations.
convit	convergence iterations.
lambda	damping factor.
&ncluster	number of clusters.

Returns

vector<long> vector of cluster memberships.

Frey, B. J. and D. Dueck (2007). "Clustering by passing messages between 
data points." Science 315(5814): 972-6.
The clustering algorithm proceeds iteratively for a given maximum iterations.
If the cluster solution does not change for a given number of iterations
(convit), convergence is assumed and the function finishes.
The damping factor should be ~0.5, or at least between 0.1 and 0.9.
For larger damping factors, more convergence iterations are required.
The diagonal of the matrix should be set to preferences for exemplars
(typically the average or median of the similarities).

◆ k_means()

vector< long > k_means	(	long	n,
		float *	data,
		long	k
	)

Generate clusters using a K-means algorithm.

Parameters

n	number of data elements.
*data	floating point array.
k	number of classes.

Returns: vector<long> vector of cluster memberships.

Variable Documentation

◆ verbose

int verbose

extern