view toolboxes/distance_learning/mlr/distance/setDistanceFullMKL.m @ 0:cc4b1211e677 tip

initial commit to HG from
Changeset:
646 (e263d8a21543) added further path and more save "camirversion.m"

function D = setDistanceFullMKL(X, W, Ifrom, Ito)
%
% D = setDistanceFullMKL(X, W, Ifrom, Ito)
%
%   X       = d-by-n-by-m data matrix
%   W       = d-by-d-by-m PSD matrix
%   Ifrom   = k-by-1 vector of source points
%   Ito     = j-by-1 vector of destination points
%
%   D = n-by-n matrix of squared euclidean distances from Ifrom to Ito
%       D is sparse, and only the rows corresponding to Ifrom and
%       columns corresponding to Ito are populated.

    [d,n,m]       = size(X);

    D = 0;
    for i = 1:m
        [vecs,vals] = eig(0.5 * (W(:,:,i) + W(:,:,i)'));
        L           = real(abs(vals)).^0.5 * vecs';

        Vfrom   = L * X(:,Ifrom,i);

        if nargin == 4
            Vto     = L * X(:,Ito,i);
        else
            Vto     = L * X(:,:,i);
            Ito     = 1:n;
        end

        D = D + distToFrom(n, Vto, Vfrom, Ito, Ifrom);
    end
end