Daniel@0: function bnet = learn_params_dbn(bnet, data)
Daniel@0: % LEARN_PARAM_DBN Estimate params of a DBN for a fully observed model
Daniel@0: % bnet = learn_params_dbn(bnet, data)
Daniel@0: %
Daniel@0: % data(i,t) is the value of node i in slice t (can be a cell array)
Daniel@0: % We currently assume there is a single time series
Daniel@0: %
Daniel@0: % We set bnet.CPD{i} to its ML/MAP estimate.
Daniel@0: %
Daniel@0: % Currently we assume each node in the first 2 slices has its own CPD (no param tying);
Daniel@0: % all nodes in slices >2 share their params with slice 2 as usual.
Daniel@0: 
Daniel@0: [ss T] = size(data);
Daniel@0: 
Daniel@0: % slice 1
Daniel@0: for j=1:ss
Daniel@0:   if adjustable_CPD(bnet.CPD{j})
Daniel@0:     fam = family(bnet.dag,j);
Daniel@0:     bnet.CPD{j} = learn_params(bnet.CPD{j}, data(fam,1));
Daniel@0:   end
Daniel@0: end
Daniel@0: 
Daniel@0: 
Daniel@0: % slices 2:T
Daniel@0: % data2(:,t) contains [data(:,t-1); data(:,t)].
Daniel@0: % Then we extract out the rows corresponding to the parents in the current and previous slice.
Daniel@0: data2 = [data(:,1:T-1);
Daniel@0: 	 data(:,2:T)];
Daniel@0: for j=1:ss
Daniel@0:   j2 = j+ss;
Daniel@0:   if adjustable_CPD(bnet.CPD{j2})
Daniel@0:     fam = family(bnet.dag,j2);
Daniel@0:     bnet.CPD{j2} = learn_params(bnet.CPD{j2}, data2(fam,:));
Daniel@0:   end
Daniel@0: end
Daniel@0: