Mercurial > hg > camir-aes2014
annotate toolboxes/FullBNT-1.0.7/bnt/learning/learn_params.m @ 0:e9a9cd732c1e tip
first hg version after svn
| author | wolffd |
|---|---|
| date | Tue, 10 Feb 2015 15:05:51 +0000 |
| parents | |
| children |
| rev | line source |
|---|---|
| wolffd@0 | 1 function bnet = learn_params(bnet, data) |
| wolffd@0 | 2 % LEARN_PARAMS Find the maximum likelihood params for a fully observed model |
| wolffd@0 | 3 % bnet = learn_params(bnet, data) |
| wolffd@0 | 4 % |
| wolffd@0 | 5 % data(i,m) is the value of node i in case m (can be a cell array) |
| wolffd@0 | 6 % |
| wolffd@0 | 7 % We set bnet.CPD{i} to its ML/MAP estimate. |
| wolffd@0 | 8 % |
| wolffd@0 | 9 % Currently we assume no param tying |
| wolffd@0 | 10 |
| wolffd@0 | 11 % AND THAT EACH DATA POINT IS A SCALAR - no longer assumed |
| wolffd@0 | 12 |
| wolffd@0 | 13 %if iscell(data) |
| wolffd@0 | 14 % data=cell2num(data); |
| wolffd@0 | 15 %end |
| wolffd@0 | 16 [n ncases] = size(data); |
| wolffd@0 | 17 for j=1:n |
| wolffd@0 | 18 e = bnet.equiv_class(j); |
| wolffd@0 | 19 assert(e==j); |
| wolffd@0 | 20 if adjustable_CPD(bnet.CPD{e}) |
| wolffd@0 | 21 fam = family(bnet.dag,j); |
| wolffd@0 | 22 %bnet.CPD{j} = learn_params(bnet.CPD{j}, data(fam,:)); |
| wolffd@0 | 23 bnet.CPD{j} = learn_params(bnet.CPD{j}, fam, data, bnet.node_sizes, bnet.cnodes); |
| wolffd@0 | 24 end |
| wolffd@0 | 25 end |
| wolffd@0 | 26 |