perClass Documentation
version 5.1 (31-May-2017)
 SDKCENTRES k-centres classifier or clustering

    P=SDKCENTRES(DATA,options)
    P=SDKCENTRES(DATA,K)

 INPUT
   DATA        training dataset
   K           number of centres per class (may be vector)

 OPTIONS
  'k'          number of centres per class (required)
  'all'        execute k-centres on entire data set (not per class)
  'iter'       number of iterations (opt, def:20)
  'maxsamples' max.number of samples used (opt.def:4000)
  'cluster'    return one output per cluster (default: return one
               output per class=classifier)
  'rounds'     repetition rounds to choose best output (minimizing
               the sum of distances between centers)
  'nodisplay'  do not show any output

 OUTPUT
   P           pipeline object

 DESCRIPTION
 SDKCENTRES describes data by k centroids (samples). The centroids are
 selected such that the maximum distance in each corresponding cluster
 is minimized.
 By default SDKCENTRES trains a classifier which handles each class in
 DATA separately and returns one output per class (distance to closest
 centroid).  Data clustering may be performed using 'cluster' option.
 SDKCENTRES then returns one output per cluster.
 The number of centers may be specified using 'k' parameter (vector of
 'k', one per class is supported')

 READ MORE
http://perclass.com/doc/guide/clustering.html#sdkcenters

 SEE ALSO
 SDMIXTURE, SDKMEANS

sdkcentres is referenced in examples: