wolffd@0: function CPD = maximize_params(CPD, temp)
wolffd@0: % MAXIMIZE_PARAMS Find ML params of an MLP using Scaled Conjugated Gradient (SCG)
wolffd@0: % CPD = maximize_params(CPD, temperature)
wolffd@0: % temperature parameter is ignored
wolffd@0: 
wolffd@0: if ~adjustable_CPD(CPD), return; end
wolffd@0: options = foptions;
wolffd@0: 
wolffd@0: % options(1) >= 0 means print an annoying message when the max. num. iter. is reached
wolffd@0: if CPD.verbose
wolffd@0:   options(1) = 1;
wolffd@0: else
wolffd@0:   options(1) = -1;
wolffd@0: end
wolffd@0: %options(1) = CPD.verbose;
wolffd@0: 
wolffd@0: options(2) = CPD.wthresh;
wolffd@0: options(3) = CPD.llthresh;
wolffd@0: options(14) = CPD.max_iter;
wolffd@0: 
wolffd@0: dpsz=length(CPD.mlp);
wolffd@0: 
wolffd@0: for i=1:dpsz
wolffd@0:     mask=[];
wolffd@0:     mask=find(CPD.eso_weights(:,:,i)>0);    % for adapting the parameters we use only positive weighted example
wolffd@0:     if  ~isempty(mask),
wolffd@0:         CPD.mlp{i} = netopt_weighted(CPD.mlp{i}, options, CPD.parent_vals(mask',:), CPD.self_vals(mask',:,i), CPD.eso_weights(mask',:,i), 'scg');
wolffd@0:         
wolffd@0:         CPD.W1(:,:,i)=CPD.mlp{i}.w1;        % update the parameters matrix
wolffd@0:         CPD.b1(i,:)=CPD.mlp{i}.b1;          %
wolffd@0:         CPD.W2(:,:,i)=CPD.mlp{i}.w2;        % update the parameters matrix
wolffd@0:         CPD.b2(i,:)=CPD.mlp{i}.b2;          %
wolffd@0:     end
wolffd@0: end