BNW using Octave instead of Matlab.

This version of BNW should perform the same as the original version. The only difference is that it uses Octave instead of Matlab when running BayesNet Toolbox during parameter learning.

I am calling this BNW_1.02. It can be accessed at:
compbio.uthsc.edu/BNW_1.02

7 files changed, 190 insertions, 0 deletions

diff --git a/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_IOhmm.m b/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_IOhmm.m
new file mode 100644
index 00000000..43c6c802
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_IOhmm.m
@@ -0,0 +1,49 @@
+% Sigmoid Belief IOHMM
+% Here is the model
+%
+%  X \  X \
+%  | |  | |
+%  Q-|->Q-|-> ...
+%  | /  | /
+%  Y    Y
+%
+clear all;
+clc;
+rand('state',0); randn('state',0);
+X = 1; Q = 2; Y = 3;
+% intra time-slice graph
+intra=zeros(3);
+intra(X,[Q Y])=1;
+intra(Q,Y)=1;
+% inter time-slice graph
+inter=zeros(3);
+inter(Q,Q)=1;
+
+ns = [1 3 1]; 
+dnodes = [2];
+eclass1 = [1 2 3];
+eclass2 = [1 4 3];
+bnet = mk_dbn(intra, inter, ns, dnodes, eclass1, eclass2);
+bnet.CPD{1} = root_CPD(bnet, 1);
+% ==========================================================
+bnet.CPD{2} = softmax_CPD(bnet, 2);
+bnet.CPD{4} = softmax_CPD(bnet, 5, 'discrete', [2]);
+% ==========================================================
+bnet.CPD{3} = gaussian_CPD(bnet, 3);
+
+% make some data
+T=20;
+cases = cell(3, T);
+cases(1,:)=num2cell(round(rand(1,T)*2)+1);
+%cases(2,:)=num2cell(round(rand(1,T))+1);
+cases(3,:)=num2cell(rand(1,T));
+
+engine = bk_inf_engine(bnet, 'exact', [1 2 3]);
+
+% log lik before learning
+[engine, loglik] = enter_evidence(engine, cases);
+
+% do learning
+ev=cell(1,1);
+ev{1}=cases;
+[bnet2, LL2] = learn_params_dbn_em(engine, ev, 3);
\ No newline at end of file
diff --git a/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_hmdt.m b/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_hmdt.m
new file mode 100644
index 00000000..88c4ae00
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_hmdt.m
@@ -0,0 +1,48 @@
+% Sigmoid Belief Hidden Markov Decision Tree    (Jordan/Gharhamani 1996)
+% 
+clear all;
+%clc;
+rand('state',0); randn('state',0);
+X = 1; Q1 = 2; Q2 = 3; Y = 4;
+% intra time-slice graph
+intra=zeros(4);
+intra(X,[Q1 Q2 Y])=1;
+intra(Q1,[Q2 Y])=1;
+intra(Q2, Y)=1;
+% inter time-slice graph
+inter=zeros(4);
+inter(Q1,Q1)=1;
+inter(Q2,Q2)=1;
+
+ns = [1 2 3 1]; 
+dnodes = [2 3]; 
+eclass1 = [1 2 3 4];
+eclass2 = [1 5 6 4];
+bnet = mk_dbn(intra, inter, ns, dnodes, eclass1, eclass2);
+
+bnet.CPD{1} = root_CPD(bnet, 1);
+% =========================================
+bnet.CPD{2} = softmax_CPD(bnet, 2);
+bnet.CPD{3} = softmax_CPD(bnet, 3, 'discrete', [2]);
+bnet.CPD{5} = softmax_CPD(bnet, 6);
+bnet.CPD{6} = softmax_CPD(bnet, 7, 'discrete', [3 6]);
+% =========================================
+bnet.CPD{4} = gaussian_CPD(bnet, 4);
+
+% make some data
+T=20;
+cases = cell(4, T);
+cases(1,:)=num2cell(round(rand(1,T)*2)+1);
+%cases(2,:)=num2cell(round(rand(1,T))+1);
+%cases(3,:)=num2cell(round(rand(1,T)*2)+1);
+cases(4,:)=num2cell(rand(1,T));
+
+engine = bk_inf_engine(bnet, 'exact', [1 2 3 4]);
+
+% log lik before learning
+[engine, loglik] = enter_evidence(engine, cases);
+
+% do learning
+ev=cell(1,1);
+ev{1}=cases;
+[bnet2, LL2] = learn_params_dbn_em(engine, ev, 10);
\ No newline at end of file
diff --git a/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_hme.m b/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_hme.m
new file mode 100644
index 00000000..0b298d48
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/static/Brutti/Belief_hme.m
@@ -0,0 +1,37 @@
+% Sigmoid Belief Hierarchical Mixtures of Experts
+
+clear all
+clc
+X = 1;
+Q1 = 2;
+Q2 = 3;
+Y = 4;
+dag = zeros(4,4);
+dag(X,[Q1 Q2 Y]) = 1;
+dag(Q1, [Q2 Y]) = 1;
+dag(Q2,Y)=1;
+ns = [1 3 4 3];
+dnodes = [2 3 4];
+onodes=[1 2 3 4];
+bnet = mk_bnet(dag,ns, dnodes);
+
+rand('state',0); randn('state',0);
+
+bnet.CPD{1} = root_CPD(bnet, 1);
+bnet.CPD{2} = softmax_CPD(bnet, 2, 'max_iter', 3);
+bnet.CPD{3} = softmax_CPD(bnet, 3, 'discrete', [2], 'max_iter', 3);
+bnet.CPD{4} = softmax_CPD(bnet, 4, 'discrete', [2 3], 'max_iter', 3);
+
+T=5;
+cases = cell(4, T);
+cases(1,:)=num2cell(rand(1,T));
+%cases(2,:)=num2cell(round(rand(1,T)*2)+1);
+%cases(3,:)=num2cell(round(rand(1,T)*3)+1);
+cases(4,:)=num2cell(round(rand(1,T)*2)+1);
+
+engine = jtree_inf_engine(bnet, onodes);
+
+[engine, loglik] = enter_evidence(engine, cases);
+
+disp('learning-------------------------------------------')
+[bnet2, LL2] = learn_params_em(engine, cases, 4);
\ No newline at end of file
diff --git a/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Entries b/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Entries
new file mode 100644
index 00000000..bf214c0a
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Entries
@@ -0,0 +1,5 @@
+/Belief_IOhmm.m/1.1.1.1/Wed May 29 15:59:54 2002//
+/Belief_hmdt.m/1.1.1.1/Wed May 29 15:59:54 2002//
+/Belief_hme.m/1.1.1.1/Wed May 29 15:59:54 2002//
+/Sigmoid_Belief.m/1.1.1.1/Wed May 29 15:59:54 2002//
+D
diff --git a/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Repository b/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Repository
new file mode 100644
index 00000000..52d4e2ed
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Repository
@@ -0,0 +1 @@
+FullBNT/BNT/examples/static/Brutti
diff --git a/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Root b/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Root
new file mode 100644
index 00000000..f3bd14a6
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/static/Brutti/CVS/Root
@@ -0,0 +1 @@
+:ext:nsaunier@bnt.cvs.sourceforge.net:/cvsroot/bnt
diff --git a/sourcecodes/bnt-master/BNT/examples/static/Brutti/Sigmoid_Belief.m b/sourcecodes/bnt-master/BNT/examples/static/Brutti/Sigmoid_Belief.m
new file mode 100644
index 00000000..1a6ecc35
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/static/Brutti/Sigmoid_Belief.m
@@ -0,0 +1,49 @@
+% Sigmoid Belief Net
+
+clear all
+clc
+dum1 = 1;
+dum2 = 2;
+dum3 = 3;
+Q1 = 4;
+Q2 = 5;
+Y = 6;
+dag = zeros(6,6);
+dag(dum1,[Q1 Y]) = 1;
+dag(dum2, Q2)=1;
+dag(dum3, [Q1 Q2])=1;
+dag(Q1,[Q2 Y]) = 1;
+dag(Q2, Y)=1;
+
+ns = [2 2 3 3 4 3];
+dnodes = [1:6];
+bnet = mk_bnet(dag,ns, dnodes);
+
+rand('state',0); randn('state',0);
+n_iter=10;
+clamped=0;
+
+bnet.CPD{1} = tabular_CPD(bnet, 1);
+bnet.CPD{2} = tabular_CPD(bnet, 2);
+bnet.CPD{3} = tabular_CPD(bnet, 3);
+% CPD = dsoftmax_CPD(bnet, self, dummy_pars, w, b, clamped, max_iter, verbose, wthresh,...
+%    llthresh, approx_hess)
+bnet.CPD{4} = softmax_CPD(bnet, 4, 'discrete', [1 3]);
+bnet.CPD{5} = softmax_CPD(bnet, 5, 'discrete', [2 3]);
+bnet.CPD{6} = softmax_CPD(bnet, 6, 'discrete', [1 4]);
+
+T=5;
+cases = cell(6, T);
+cases(1,:)=num2cell(round(rand(1,T)*1)+1);
+%cases(2,:)=num2cell(round(rand(1,T)*1)+1);
+cases(3,:)=num2cell(round(rand(1,T)*2)+1);
+cases(4,:)=num2cell(round(rand(1,T)*2)+1); 
+%cases(5,:)=num2cell(round(rand(1,T)*3)+1);
+cases(6,:)=num2cell(round(rand(1,T)*2)+1);
+
+engine = jtree_inf_engine(bnet);
+
+[engine, loglik] = enter_evidence(engine, cases);
+
+disp('learning-------------------------------------------')
+[bnet2, LL2, eng2] = learn_params_em(engine, cases, n_iter);
\ No newline at end of file