%PERC Percentile combining classifier
%
% W = PERC(V,P)
% W = V*PERC([],P)
%
% INPUT
% V Set of classifiers
% P Percentile, 0 <= P <= 100
%
% OUTPUT
% W Percentile combining classifier on V
%
% DESCRIPTION
% If V = [V1,V2,V3, ... ] is a set of classifiers trained on the
% same classes and W is the percentile combiner: it selects the class
% defined by the percentile of the outputs of the input classifiers. This
% might also be used as A*[V1,V2,V3]*PERC([],P) in which A is a dataset to
% be classified.
%
% PERC([],0) is equal to MINC
% PERC([],50) is equal to MEDIANC
% PERC([],100) is equal to MAXC
%
% If it is desired to operate on posterior probabilities then the
% input classifiers should be extended like V = V*CLASSC;
%
% The base classifiers may be combined in a stacked way (operating
% in the same feature space by V = [V1,V2,V3, ... ] or in a parallel
% way (operating in different feature spaces) by V = [V1;V2;V3; ... ]
%
% SEE ALSO (<a href="http://37steps.com/prtools">PRTools Guide</a>)
% MAPPINGS, DATASETS, VOTEC, MAXC, MINC, MEANC, MEDIANC, PRODC,
% AVERAGEC, STACKED, PARALLEL
%
% EXAMPLES
% See PREX_COMBINING
% Copyright: R.P.W. Duin, r.p.w.duin@37steps.com
% Faculty EWI, Delft University of Technology
% P.O. Box 5031, 2600 GA Delft, The Netherlands
function w = perc(p1,par)
if nargin < 2 | par < 0 | par > 100
error('Percentile between 0 and 100 should be defined for percentile combiner')
end
type = 'perc'; % define the operation processed by FIXEDCC.
% define the name of the combiner.
% this is the general procedure for all possible calls of fixed combiners
% handled by FIXEDCC
name = 'Percentile combiner';
if nargin == 0
w = prmapping('fixedcc','combiner',{[],type,name,par});
else
w = fixedcc(p1,[],type,name,par);
end
if isa(w,'prmapping')
w = setname(w,name);
end
return