Mercurial > hg > camir-aes2014
annotate toolboxes/FullBNT-1.0.7/bnt/examples/static/sample1.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 % Check sampling on a mixture of experts model |
wolffd@0 | 2 % |
wolffd@0 | 3 % X \ |
wolffd@0 | 4 % | | |
wolffd@0 | 5 % Q | |
wolffd@0 | 6 % | / |
wolffd@0 | 7 % Y |
wolffd@0 | 8 % |
wolffd@0 | 9 % where all arcs point down. |
wolffd@0 | 10 % We condition everything on X, so X is a root node. Q is a softmax, and Y is a linear Gaussian. |
wolffd@0 | 11 % Q is hidden, X and Y are observed. |
wolffd@0 | 12 |
wolffd@0 | 13 X = 1; |
wolffd@0 | 14 Q = 2; |
wolffd@0 | 15 Y = 3; |
wolffd@0 | 16 dag = zeros(3,3); |
wolffd@0 | 17 dag(X,[Q Y]) = 1; |
wolffd@0 | 18 dag(Q,Y) = 1; |
wolffd@0 | 19 ns = [1 2 2]; |
wolffd@0 | 20 dnodes = [2]; |
wolffd@0 | 21 bnet = mk_bnet(dag, ns, dnodes); |
wolffd@0 | 22 |
wolffd@0 | 23 x = 0.5; |
wolffd@0 | 24 bnet.CPD{1} = root_CPD(bnet, 1, x); |
wolffd@0 | 25 bnet.CPD{2} = softmax_CPD(bnet, 2); |
wolffd@0 | 26 bnet.CPD{3} = gaussian_CPD(bnet, 3); |
wolffd@0 | 27 |
wolffd@0 | 28 data_case = sample_bnet(bnet, 'evidence', {0.8, [], []}) |
wolffd@0 | 29 ll = log_lik_complete(bnet, data_case) |
wolffd@0 | 30 |
wolffd@0 | 31 data_case = sample_bnet(bnet, 'evidence', {-11, [], []}) |
wolffd@0 | 32 ll = log_lik_complete(bnet, data_case) |
wolffd@0 | 33 |
wolffd@0 | 34 |