Added online filtering inference.

matlab/meet/learn/ahmm/inferenceahmm.m

@@ -1,6 +1,14 @@
 function R = inferenceahmm(ahmm, data, predictNode, method)
 
-engine = smoother_engine(jtree_2TBN_inf_engine(ahmm));
+switch method
+  case 'fixed-interval-smoothing'
+  case 'viterbi'
+    engine = smoother_engine(jtree_2TBN_inf_engine(ahmm));
+  case 'filtering'
+    engine = filter_engine(jtree_2TBN_inf_engine(ahmm));
+  otherwise
+    error(['Inference method not implemented: ' method]);
+end
 
 nseq = length(data);
 R = cell(1, nseq);
@@ -10,6 +18,19 @@
     case 'fixed-interval-smoothing'
       engine = enter_evidence(engine, evidence);
       R{i} = mapest(engine, predictNode, length(evidence));
+    case 'filtering'
+      T = size(evidence, 2);
+      nhnode = length(predictNode);
+      mapEst = cell(nhnode, T);
+      for t = 1 : T
+        for n = 1 : nhnode
+          engine = enter_evidence(engine, evidence, t);
+          m = marginal_nodes(engine, predictNode, t);
+          [~, ndx] = max(m.T);
+          mapEst{n, t} = ndx;
+        end
+      end
+      R{i} = mapEst;
     case 'viterbi'
       % Find the most probable explanation (Viterbi).
       mpe = find_mpe(engine, evidence);

matlab/meet/learn/ahmm/learnahmm.m

@@ -18,7 +18,7 @@
 trainData = createInputData(Y.train, X.train, param);
 finalAhmm = trainahmm(param, trainData);
 
-predictNode = [param.G1 param.F1];
+predictNode = [param.G1];
 param.onodes = [param.X1];
 finalAhmm = sethiddenbit(finalAhmm, param.onodes);
 checkahmm(finalAhmm);