camir-aes2014: toolboxes/FullBNT-1.0.7/bnt/CPDs/@gaussian_CPD/learn

first hg version after svn

rev	line source
wolffd@0	1 function CPD = learn_params(CPD, fam, data, ns, cnodes)
wolffd@0	2 %function CPD = learn_params(CPD, fam, data, ns, cnodes)
wolffd@0	3 % LEARN_PARAMS Compute the maximum likelihood estimate of the params of a gaussian CPD given complete data
wolffd@0	4 % CPD = learn_params(CPD, fam, data, ns, cnodes)
wolffd@0	5 %
wolffd@0	6 % data(i,m) is the value of node i in case m (can be cell array).
wolffd@0	7 % We assume this node has a maximize_params method.
wolffd@0	8
wolffd@0	9 ncases = size(data, 2);
wolffd@0	10 CPD = reset_ess(CPD);
wolffd@0	11 % make a fully observed joint distribution over the family
wolffd@0	12 fmarginal.domain = fam;
wolffd@0	13 fmarginal.T = 1;
wolffd@0	14 fmarginal.mu = [];
wolffd@0	15 fmarginal.Sigma = [];
wolffd@0	16 if ~iscell(data)
wolffd@0	17 cases = num2cell(data);
wolffd@0	18 else
wolffd@0	19 cases = data;
wolffd@0	20 end
wolffd@0	21 hidden_bitv = zeros(1, max(fam));
wolffd@0	22 for m=1:ncases
wolffd@0	23 % specify (as a bit vector) which elements in the family domain are hidden
wolffd@0	24 hidden_bitv = zeros(1, max(fmarginal.domain));
wolffd@0	25 ev = cases(:,m);
wolffd@0	26 hidden_bitv(find(isempty(ev)))=1;
wolffd@0	27 CPD = update_ess(CPD, fmarginal, ev, ns, cnodes, hidden_bitv);
wolffd@0	28 end
wolffd@0	29 CPD = maximize_params(CPD);
wolffd@0	30
wolffd@0	31

Mercurial > hg > camir-aes2014