image thumbnail

Hierarchical Kalman Filter for clinical time series prediction

by

 

It is an implementation of hierarchical (a.k.a. multi-scale) Kalman filter using belief propagation.

inv_s(A, iA)
% this is safe inv, which can handle inverse of inf matrix
function iA = inv_s(A, iA)
    if(~isempty(iA))
       return;
    else
        if(~isempty(A))
            B = eyeInf(size(A, 1));
            if(sum(B ~= A) == 0)
                iA = zeros(size(A));
            else
                iA = inv(A);
            end
        else
            error('both A and iA are empty');
        end
    end
end

Contact us