changed some variable names

chapter13/HMM/demo.m

@@ -3,10 +3,13 @@
 d = 3;
 k = 2;
 n = 10000;
-
 [x, model] = hmmRnd(d, k, n);
+%%
 [z, v] = hmmViterbi(x,model);
-[alpha,energy] = hmmFilter(x,model);
-[gamma, alpha, beta, c] = hmmSmoother(x,model);
-[model, energy] = hmmEm(x,k);
-plot(energy)
+%%
+% [alpha,llh] = hmmFilter(x,model);
+%%
+% [gamma, alpha, beta, c] = hmmSmoother(x,model);
+%%
+% [model, llh] = hmmEm(x,k);
+% plot(llh)

chapter13/HMM/hmmEm.m

@@ -1,4 +1,4 @@
-function [model, energy] = hmmEm(x, init)
+function [model, llh] = hmmEm(x, init)
 % EM algorithm to fit the parameters of HMM model (a.k.a Baum-Welch algorithm)
 %   x: 1 x n sequence of observations
 %   init: model or k
@@ -21,18 +21,18 @@
 
 tol = 1e-4;
 maxIter = 100;
-energy = -inf(1,maxIter);
+llh = -inf(1,maxIter);
 for iter = 2:maxIter
 %     E-step
     [gamma,alpha,beta,c] = hmmSmoother_(M,A,s);
-    energy(iter) = sum(log(c(c>0)));
-    if energy(iter)-energy(iter-1) < tol*abs(energy(iter-1)); break; end   % check likelihood for convergence
+    llh(iter) = sum(log(c(c>0)));
+    if llh(iter)-llh(iter-1) < tol*abs(llh(iter-1)); break; end   % check likelihood for convergence
 %     M-step 
     A = normalize(A.*(alpha(:,1:n-1)*bsxfun(@times,beta(:,2:n).*M(:,2:n),1./c(2:end))'),2);      % 13.19
     s = gamma(:,1);                                                                             % 13.18
     M = bsxfun(@times,gamma*X',1./sum(gamma,2))*X;                                          
 end
-energy = energy(2:iter);
+llh = llh(2:iter);
 model.A = A;
 model.E = E;
 model.s = s;

chapter13/LDS/demo.m

@@ -1,11 +1,12 @@
 % demo
+clear; close all;
 d = 3;
-k = 3;
-n = 10000;
+k = 2;
+n = 100;
 
 [X,model] = ldsRnd(d,k,n);
-% [mu, V, llh] = kalmanFilter(X, model);
-% [nu, U, Ezz, Ezy, llh] = kalmanSmoother(X, model);
+[mu, V, llh] = kalmanFilter(X, model);
+% % [nu, U, Ezz, Ezy, llh] = kalmanSmoother(X, model);
 [model, llh] = ldsEm(X, model);
-plot(llh);
+% plot(llh);