EADIST

Create evidence accumulation dissimilarity matrix

    D = EADIST(LABELS)
    D = LABELS*EADIST

Input
LABELS	Multilevel labeling, size [M,K], for M objects.

Output
D	MxM distance matrix

Description

This routine constructs from K sets of clusterlabels (doubles) of M objects stored in the MxK array LABELS an MxM dissimilarity matrix D.

The preferred clustering of D is the single linkage hierarchical  clustering, see DCLUSTH.

Reference(s)

A.L.N. Fred, A.K. Jain. Combining multiple clusterings using evidence accumulation, PAMI, 27.6 (2005): 835-850.

See also

datasets, mappings, dclusth,