Mercurial > hg > camir-aes2014
diff toolboxes/FullBNT-1.0.7/bnt/inference/dynamic/@kalman_inf_engine/enter_evidence.m @ 0:e9a9cd732c1e tip
first hg version after svn
author | wolffd |
---|---|
date | Tue, 10 Feb 2015 15:05:51 +0000 |
parents | |
children |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/toolboxes/FullBNT-1.0.7/bnt/inference/dynamic/@kalman_inf_engine/enter_evidence.m Tue Feb 10 15:05:51 2015 +0000 @@ -0,0 +1,83 @@ +function [engine, loglik] = enter_evidence(engine, evidence, varargin) +% ENTER_EVIDENCE Add the specified evidence to the network (kalman) +% [engine, loglik] = enter_evidence(engine, evidence, ...) +% +% evidence{i,t} = [] if if X(i,t) is hidden, and otherwise contains its observed value (scalar or column vector) +% +% The following optional arguments can be specified in the form of name/value pairs: +% [default value in brackets] +% +% maximize - if 1, does max-product (same as sum-product for Gaussians!), else sum-product [0] +% filter - if 1, do filtering, else smoothing [0] +% +% e.g., engine = enter_evidence(engine, ev, 'maximize', 1) + +maximize = 0; +filter = 0; + +% parse optional params +args = varargin; +nargs = length(args); +if nargs > 0 + for i=1:2:nargs + switch args{i}, + case 'maximize', maximize = args{i+1}; + case 'filter', filter = args{i+1}; + otherwise, + error(['invalid argument name ' args{i}]); + end + end +end + +assert(~maximize); + +bnet = bnet_from_engine(engine); +n = length(bnet.intra); +onodes = bnet.observed; +hnodes = mysetdiff(1:n, onodes); +T = size(evidence, 2); +ns = bnet.node_sizes; +O = sum(ns(onodes)); +data = reshape(cat(1, evidence{onodes,:}), [O T]); + +A = engine.trans_mat; +C = engine.obs_mat; +Q = engine.trans_cov; +R = engine.obs_cov; +init_x = engine.init_state; +init_V = engine.init_cov; + +if filter + [x, V, VV, loglik] = kalman_filter(data, A, C, Q, R, init_x, init_V); +else + [x, V, VV, loglik] = kalman_smoother(data, A, C, Q, R, init_x, init_V); +end + + +% Wrap the posterior inside a potential, so it can be marginalized easily +engine.one_slice_marginal = cell(1,T); +engine.two_slice_marginal = cell(1,T); +ns(onodes) = 0; +ns(onodes+n) = 0; +ss = length(bnet.intra); +for t=1:T + dom = (1:n); + engine.one_slice_marginal{t} = mpot(dom+(t-1)*ss, ns(dom), 1, x(:,t), V(:,:,t)); +end +% for t=1:T-1 +% dom = (1:(2*n)); +% mu = [x(:,t); x(:,t)]; +% Sigma = [V(:,:,t) VV(:,:,t+1)'; +% VV(:,:,t+1) V(:,:,t+1)]; +% engine.two_slice_marginal{t} = mpot(dom+(t-1)*ss, ns(dom), 1, mu, Sigma); +% end +for t=2:T + %dom = (1:(2*n)); + current_slice = hnodes; + next_slice = hnodes + ss; + dom = [current_slice next_slice]; + mu = [x(:,t-1); x(:,t)]; + Sigma = [V(:,:,t-1) VV(:,:,t)'; + VV(:,:,t) V(:,:,t)]; + engine.two_slice_marginal{t-1} = mpot(dom+(t-2)*ss, ns(dom), 1, mu, Sigma); +end