|
wolffd@0
|
1 % to test whether scg inference engine can handl dynameic BN
|
|
wolffd@0
|
2 % Make a linear dynamical system
|
|
wolffd@0
|
3 % X1 -> X2
|
|
wolffd@0
|
4 % | |
|
|
wolffd@0
|
5 % v v
|
|
wolffd@0
|
6 % Y1 Y2
|
|
wolffd@0
|
7
|
|
wolffd@0
|
8 intra = zeros(2);
|
|
wolffd@0
|
9 intra(1,2) = 1;
|
|
wolffd@0
|
10 inter = zeros(2);
|
|
wolffd@0
|
11 inter(1,1) = 1;
|
|
wolffd@0
|
12 n = 2;
|
|
wolffd@0
|
13
|
|
wolffd@0
|
14 X = 2; % size of hidden state
|
|
wolffd@0
|
15 Y = 2; % size of observable state
|
|
wolffd@0
|
16
|
|
wolffd@0
|
17 ns = [X Y];
|
|
wolffd@0
|
18 dnodes = [];
|
|
wolffd@0
|
19 onodes = [2];
|
|
wolffd@0
|
20 eclass1 = [1 2];
|
|
wolffd@0
|
21 eclass2 = [3 2];
|
|
wolffd@0
|
22 bnet = mk_dbn(intra, inter, ns, dnodes, eclass1, eclass2);
|
|
wolffd@0
|
23
|
|
wolffd@0
|
24 x0 = rand(X,1);
|
|
wolffd@0
|
25 V0 = eye(X);
|
|
wolffd@0
|
26 C0 = rand(Y,X);
|
|
wolffd@0
|
27 R0 = eye(Y);
|
|
wolffd@0
|
28 A0 = rand(X,X);
|
|
wolffd@0
|
29 Q0 = eye(X);
|
|
wolffd@0
|
30
|
|
wolffd@0
|
31 bnet.CPD{1} = gaussian_CPD(bnet, 1, 'mean', x0, 'cov', V0);
|
|
wolffd@0
|
32 %bnet.CPD{2} = gaussian_CPD(bnet, 2, 'mean', zeros(Y,1), 'cov', R0, 'weights', C0, 'full', 'untied', 'clamped_mean');
|
|
wolffd@0
|
33 %bnet.CPD{3} = gaussian_CPD(bnet, 3, 'mean', zeros(X,1), 'cov', Q0, 'weights', A0, 'full', 'untied', 'clamped_mean');
|
|
wolffd@0
|
34 bnet.CPD{2} = gaussian_CPD(bnet, 2, 'mean', zeros(Y,1), 'cov', R0, 'weights', C0);
|
|
wolffd@0
|
35 bnet.CPD{3} = gaussian_CPD(bnet, 3, 'mean', zeros(X,1), 'cov', Q0, 'weights', A0);
|
|
wolffd@0
|
36
|
|
wolffd@0
|
37
|
|
wolffd@0
|
38 T = 5; % fixed length sequences
|
|
wolffd@0
|
39
|
|
wolffd@0
|
40 clear engine;
|
|
wolffd@0
|
41 %engine{1} = kalman_inf_engine(bnet, onodes);
|
|
wolffd@0
|
42 engine{1} = scg_unrolled_dbn_inf_engine(bnet, T, onodes);
|
|
wolffd@0
|
43 engine{2} = jtree_unrolled_dbn_inf_engine(bnet, T);
|
|
wolffd@0
|
44
|
|
wolffd@0
|
45 N = length(engine);
|
|
wolffd@0
|
46
|
|
wolffd@0
|
47 % inference
|
|
wolffd@0
|
48
|
|
wolffd@0
|
49 ev = sample_dbn(bnet, T);
|
|
wolffd@0
|
50 evidence = cell(n,T);
|
|
wolffd@0
|
51 evidence(onodes,:) = ev(onodes, :);
|
|
wolffd@0
|
52
|
|
wolffd@0
|
53 t = 2;
|
|
wolffd@0
|
54 query = [1 3];
|
|
wolffd@0
|
55 m = cell(1, N);
|
|
wolffd@0
|
56 ll = zeros(1, N);
|
|
wolffd@0
|
57
|
|
wolffd@0
|
58 engine{1} = enter_evidence(engine{1}, evidence);
|
|
wolffd@0
|
59 [engine{2}, ll(2)] = enter_evidence(engine{2}, evidence);
|
|
wolffd@0
|
60 m{1} = marginal_nodes(engine{1}, query);
|
|
wolffd@0
|
61 m{2} = marginal_nodes(engine{2}, query, t);
|
|
wolffd@0
|
62
|
|
wolffd@0
|
63
|
|
wolffd@0
|
64 % compare all engines to engine{1}
|
|
wolffd@0
|
65 for i=2:N
|
|
wolffd@0
|
66 assert(approxeq(m{1}.mu, m{i}.mu));
|
|
wolffd@0
|
67 assert(approxeq(m{1}.Sigma, m{i}.Sigma));
|
|
wolffd@0
|
68 % assert(approxeq(ll(1), ll(i)));
|
|
wolffd@0
|
69 end
|
|
wolffd@0
|
70
|
