%KLLDC Linear classifier built on the KL expansion of the common covariance matrix
%
% W = KLLDC(A,N)
% W = KLLDC(A,ALF)
%
% INPUT
% A Dataset
% N Number of significant eigenvectors
% ALF 0 < ALF <= 1, percentage of the total variance explained (default: 0.9)
%
% OUTPUT
% W Linear classifier
%
% DESCRIPTION
% Finds the linear discriminant function W for the dataset A. This is done
% by computing the LDC on the data projected on the first eigenvectors of
% the averaged covariance matrix of the classes. Either first N eigenvectors
% are used or the number of eigenvectors is determined such that ALF, the
% percentage of the total variance is explained. (Karhunen Loeve expansion)
%
% If N (ALF) is NaN it is optimised by REGOPTC.
%
% SEE ALSO (<a href="http://37steps.com/prtools">PRTools Guide</a>)
% MAPPINGS, DATASETS, PCLDC, KLM, FISHERM, REGOPTC
% Copyright: R.P.W. Duin, duin@ph.tn.tudelft.nl
% Faculty of Applied Physics, Delft University of Technology
% P.O. Box 5046, 2600 GA Delft, The Netherlands
% $Id: klldc.m,v 1.4 2007/06/13 21:59:42 duin Exp $
function W = klldc(a,n)
if nargin < 2
n = [];
prwarning(4,'number of significant eigenvectors not supplied, 0.9 variance explained');
end
if nargin == 0 | isempty(a)
W = prmapping('klldc',{n});
elseif isnan(n) % optimize regularisation parameter
defs = {1};
parmin_max = [1,size(a,2)];
W = regoptc(a,mfilename,{n},defs,[1],parmin_max,testc([],'soft'),0);
else
islabtype(a,'crisp','soft');
isvaldfile(a,2,2); % at least 2 object per class, 2 classes
a = testdatasize(a,'features');
a = setprior(a,getprior(a));
v = klm(a,n);
W = v*ldc(a*v);
W = setcost(W,a);
end
W = setname(W,'KL Bayes-Normal-1');
return