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

8 files changed, 537 insertions, 0 deletions

diff --git a/sourcecodes/bnt-master/BNT/examples/limids/CVS/Entries b/sourcecodes/bnt-master/BNT/examples/limids/CVS/Entries
new file mode 100644
index 00000000..f7f12046
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/limids/CVS/Entries
@@ -0,0 +1,6 @@
+/amnio.m/1.1.1.1/Mon Sep 13 03:21:04 2004//
+/asia_dt1.m/1.1.1.1/Mon Jun  7 15:53:54 2004//
+/id1.m/1.1.1.1/Wed May 29 15:59:54 2002//
+/oil1.m/1.1.1.1/Mon Sep 13 02:27:08 2004//
+/pigs1.m/1.1.1.1/Wed May 29 15:59:54 2002//
+D
diff --git a/sourcecodes/bnt-master/BNT/examples/limids/CVS/Repository b/sourcecodes/bnt-master/BNT/examples/limids/CVS/Repository
new file mode 100644
index 00000000..bd5dd5af
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/limids/CVS/Repository
@@ -0,0 +1 @@
+FullBNT/BNT/examples/limids
diff --git a/sourcecodes/bnt-master/BNT/examples/limids/CVS/Root b/sourcecodes/bnt-master/BNT/examples/limids/CVS/Root
new file mode 100644
index 00000000..f3bd14a6
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/limids/CVS/Root
@@ -0,0 +1 @@
+:ext:nsaunier@bnt.cvs.sourceforge.net:/cvsroot/bnt
diff --git a/sourcecodes/bnt-master/BNT/examples/limids/amnio.m b/sourcecodes/bnt-master/BNT/examples/limids/amnio.m
new file mode 100644
index 00000000..fd621b6c
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/limids/amnio.m
@@ -0,0 +1,135 @@
+
+clear all
+B0 = 1; Rtriple = 2; Damnio = 3;
+B1 = 4; Ramnio = 5; Dabort = 6; 
+B2 = 7; U = 8;
+
+N = 8;
+dag = zeros(N,N);
+dag(B0, [Rtriple B1 Ramnio]) = 1;
+dag(Rtriple, [Damnio Dabort]) = 1;
+dag(Damnio, [B1 Ramnio]) = 1;
+dag(B1, B2) = 1;
+dag(Ramnio, [Dabort U]) = 1;
+dag(Dabort, B2) = 1;
+dag(B2, U) = 1;
+
+
+
+ns = zeros(1,N);
+ns(B0) = 2;
+ns(B1) = 3;
+ns(B2) = 4;
+ns(Rtriple) = 2;
+ns(Ramnio) = 3;
+ns(Damnio) = 2;
+ns(Dabort) = 2;
+ns(U) = 1;
+
+limid = mk_limid(dag, ns, 'chance', [B0 B1 B2], ...
+		 'decision', [Damnio Dabort], 'utility', [U]);
+
+% states of nature
+healthy = 1; downs = 2; miscarry = 3; aborted = 4;
+% test results
+pos = 1; neg = 2; unk = 3;
+% actions
+yes = 1; no = 2;
+
+% Prior probability baby has downs syndrome
+tbl = zeros(2,1);
+p = 1/1000; % from www.downs-syndrome.org.uk figure
+p = 24/10000; % www-personal.umich.edu/~bobwolfe/560/review/Downs.pdf (for women agen 35-40)
+tbl(healthy) = 1-p;
+tbl(downs) = p;
+limid.CPD{B0} = tabular_CPD(limid, B0, tbl);
+
+% Reliability of triple screen test
+% Unreliable sensor
+% B0 -> Rtriple
+tbl = zeros(2,2); % Rtriple = pos, neg
+p = 0.5;  % high false positive rate (guess)
+tbl(healthy, :) = [p 1-p];
+p = 0.6; % low detection rate (march of dimes figure)
+tbl(downs, :) = [p 1-p]; 
+limid.CPD{Rtriple} = tabular_CPD(limid, Rtriple, tbl);
+
+limid.CPD{Damnio} = tabular_decision_node(limid, Damnio);
+
+% Effect of amnio on baby  B0,Damnio -> B1
+ % 1/200 risk of miscarry 
+p = 1/200; % (march of dimes figure)
+tbl = zeros(2, 2, 3); % B1 = healthy, downs, miscarry
+tbl(healthy, no, :) =  [1     0     0];
+tbl(downs, no, :) =    [0     1     0];
+tbl(healthy, yes, :) = [1-p     0   p];
+tbl(downs, yes, :) =   [0     1-p   p];
+limid.CPD{B1} = tabular_CPD(limid, B1, tbl);
+
+% Reliability of amnio  B0, Damnio -> Ramnio
+% Perfect sensor
+tbl = zeros(2,2,3); % Ramnio = pos, neg, unk
+tbl(:, no, :) =        repmat([0 0 1], 2 ,1);
+tbl(healthy, yes, :) = [0 1 0]; 
+tbl(downs, yes, :) =   [1 0 0]; 
+limid.CPD{Ramnio} = tabular_CPD(limid, Ramnio, tbl);
+
+limid.CPD{Dabort} = tabular_decision_node(limid, Dabort);
+
+% Effect of abortion on baby  B1, Dabort -> B2
+tbl = zeros(3, 2, 4); % B2 = healthy, downs, miscarry, aborted
+tbl(:, yes, :) =       repmat([0 0 0 1], 3, 1);
+tbl(healthy, no, :) =  [1 0 0 0];
+tbl(downs, no, :) =    [0 1 0 0];
+tbl(miscarry, no, :) = [0 0 1 0];
+limid.CPD{B2} = tabular_CPD(limid, B2, tbl);
+
+% Utility U(Ramnio, B2)
+tbl = zeros(3, 4);
+tbl(:, healthy) = 5000;
+tbl(:, downs) = -50000;
+tbl(:, miscarry) = -1000;
+tbl(:, aborted) = -1000;
+
+if 0
+%tbl(unk, miscarry) = 0; % this case is impossible
+tbl(pos, miscarry) = -1;
+tbl(neg, miscarry) = -1000;
+if 1
+  tbl(unk, aborted) = -100;
+  tbl(pos, aborted) = -1;
+  tbl(neg, aborted) = -500;
+else % pro-life utility fn
+  tbl(unk, aborted) = -500000;
+  tbl(pos, aborted) = -500000;
+  tbl(neg, aborted) = -500000;
+end 
+end
+
+limid.CPD{U} = tabular_utility_node(limid, U,  tbl);
+
+
+
+engine = jtree_limid_inf_engine(limid);
+[strategy, MEU] = solve_limid(engine);
+
+% Rtriple U(Damnio=1=yes)  U(Damnio=2=no)
+% 1=pos    0               1
+% 2=neg    0               1
+dispcpt(strategy{Damnio})
+if isequal(strategy{Damnio}(1,:), strategy{Damnio}(2,:))
+  % Rtriple result irrelevant
+  doAmnio = argmax(strategy{Damnio}(1,:))
+else
+  doAmnio = 1;
+end
+
+% Rtriple Ramnio U(Dabort=yes=1) U(Dabort=no=2)
+% 1=pos   1=pos  1               0
+% 2=neg   1=pos  1               0
+% 1=pos   2=neg  0               1
+% 2=neg   2=neg  0               1
+% 1=pos   3=unk  0               1
+% 2=neg   3=unk  0               1
+dispcpt(strategy{Dabort})
+
diff --git a/sourcecodes/bnt-master/BNT/examples/limids/asia_dt1.m b/sourcecodes/bnt-master/BNT/examples/limids/asia_dt1.m
new file mode 100644
index 00000000..de305818
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/limids/asia_dt1.m
@@ -0,0 +1,100 @@
+% decision theoretic version of asia network
+% Cowell et al, p177
+% We explicitely add the no-forgetting arcs.
+
+Smoking = 1;
+VisitToAsia = 2;
+Bronchitis = 3;
+LungCancer = 4;
+TB = 5;
+Do_Xray = 6;
+TBorCancer = 7;
+Util_Xray = 8;
+Dys = 9;
+posXray = 10;
+Do_Hosp = 11;
+Util_Hosp = 12;
+
+n = 12;
+dag = zeros(n);
+dag(Smoking, [Bronchitis LungCancer]) = 1;
+dag(VisitToAsia, [TB Do_Xray Do_Hosp]) = 1;
+dag(Bronchitis, Dys) = 1;
+dag(LungCancer, [Util_Hosp TBorCancer]) = 1;
+dag(TB, [Util_Hosp TBorCancer Util_Xray]) = 1;
+dag(Do_Xray, [posXray Util_Xray Do_Hosp]) = 1;
+dag(TBorCancer, [Dys posXray]) = 1;
+dag(Dys, Do_Hosp) = 1;
+dag(posXray, Do_Hosp) = 1;
+dag(Do_Hosp, Util_Hosp) = 1;
+
+dnodes = [Do_Xray Do_Hosp];
+unodes = [Util_Xray Util_Hosp];
+cnodes = mysetdiff(1:n, [dnodes unodes]); % chance nodes
+ns = 2*ones(1,n);
+ns(unodes) = 1;
+limid = mk_limid(dag, ns, 'chance', cnodes, 'decision', dnodes, 'utility', unodes);
+
+% 1 = yes, 2 = no
+limid.CPD{VisitToAsia} = tabular_CPD(limid, VisitToAsia, [0.01 0.99]);
+limid.CPD{Bronchitis} = tabular_CPD(limid, Bronchitis, [0.6 0.3  0.4 0.7]);
+limid.CPD{Dys} = tabular_CPD(limid, Dys, [0.9 0.7 0.8 0.1  0.1 0.3 0.2 0.9]);
+limid.CPD{TBorCancer} = tabular_CPD(limid, TBorCancer, [1 1 1 0  0 0 0 1]);
+
+limid.CPD{LungCancer} = tabular_CPD(limid, LungCancer, [0.1 0.01  0.9 0.99]);
+limid.CPD{Smoking} = tabular_CPD(limid, Smoking, [0.5 0.5]);
+limid.CPD{TB} = tabular_CPD(limid, TB, [0.05 0.01  0.95 0.99]);
+limid.CPD{posXray} = tabular_CPD(limid, posXray, [0.98 0.5 0.05 0.5  0.02 0.5 0.95 0.5]);
+
+limid.CPD{Util_Hosp} = tabular_utility_node(limid, Util_Hosp, [180 120 160 15  2 4 0 40]);
+limid.CPD{Util_Xray} = tabular_utility_node(limid, Util_Xray, [0 1 10 10]);
+
+for i=dnodes(:)'
+  limid.CPD{i} = tabular_decision_node(limid, i);
+end
+
+engines = {};
+engines{end+1} = global_joint_inf_engine(limid);
+engines{end+1} = jtree_limid_inf_engine(limid);
+%engines{end+1} = belprop_inf_engine(limid);
+
+exact = [1 2];
+%approx = 3;
+approx = [];
+
+
+NE = length(engines);
+MEU = zeros(1, NE);
+niter = zeros(1, NE);
+strategy = cell(1, NE);
+
+tol = 1e-2;
+for e=1:length(engines)
+  [strategy{e}, MEU(e), niter(e)] = solve_limid(engines{e});
+end
+
+for e=exact(:)'
+  assert(approxeq(MEU(e), 47.49, tol))
+  assert(isequal(strategy{e}{Do_Xray}(:)', [1 0 0 1]))
+  
+  % Check the hosptialize strategy is correct (p180)
+  % We assume the patient has not been to Asia and therefore did not have an Xray.
+  % In this case it is optimal not to hospitalize regardless of whether the patient has
+  % dyspnoea or not (and of course regardless of the value of pos_xray).
+  asia = 2;
+  do_xray = 2;
+  for dys = 1:2
+    for pos_xray = 1:2
+      assert(argmax(squeeze(strategy{e}{Do_Hosp}(asia, do_xray, dys, pos_xray, :))) == 2)
+    end
+  end
+end
+
+
+for e=approx(:)'
+  approxeq(strategy{exact(1)}{Do_Xray}, strategy{e}{Do_Xray})
+  approxeq(strategy{exact(1)}{Do_Hosp}, strategy{e}{Do_Hosp})
+end
+
+ 
+
diff --git a/sourcecodes/bnt-master/BNT/examples/limids/id1.m b/sourcecodes/bnt-master/BNT/examples/limids/id1.m
new file mode 100644
index 00000000..ddacd67a
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/limids/id1.m
@@ -0,0 +1,50 @@
+% influence diagram with no loops
+%
+% rv  dec
+%  \  /
+%  utility
+
+N = 3;
+dag = zeros(N);
+X = 1; D = 2; U = 3;
+dag([X D], U)=1;
+
+ns = zeros(1,N);
+ns(X) = 2; ns(D) = 2; ns(U) = 1;
+
+limid = mk_limid(dag, ns, 'chance', X, 'decision', D, 'utility', U);
+
+% use random params
+limid.CPD{X} = tabular_CPD(limid, X);
+limid.CPD{D} = tabular_decision_node(limid, D);
+limid.CPD{U} = tabular_utility_node(limid, U);
+
+%fname = '/home/cs/murphyk/matlab/Misc/loopybel.txt';
+global BNT_HOME
+fname = sprintf('%s/loopybel.txt', BNT_HOME);
+
+engines = {};
+engines{end+1} = global_joint_inf_engine(limid);
+engines{end+1} = jtree_limid_inf_engine(limid);
+%engines{end+1} = belprop_inf_engine(limid, 'max_iter', 2*N, 'filename', fname);
+engines{end+1} = belprop_inf_engine(limid, 'max_iter', 2*N);
+
+exact = [1 2];
+approx = 3;
+
+E = length(engines);
+strategy = cell(1, E);
+MEU = zeros(1, E);
+for e=1:E
+  [strategy{e}, MEU(e)] = solve_limid(engines{e});
+  MEU
+end
+MEU
+
+for e=exact(:)'
+  assert(approxeq(strategy{exact(1)}{D}, strategy{e}{D}))
+end
+
+for e=approx(:)'
+  approxeq(strategy{exact(1)}{D}, strategy{e}{D})
+end
diff --git a/sourcecodes/bnt-master/BNT/examples/limids/oil1.m b/sourcecodes/bnt-master/BNT/examples/limids/oil1.m
new file mode 100644
index 00000000..192acc9d
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/limids/oil1.m
@@ -0,0 +1,91 @@
+% oil wildcatter influence diagram in Cowell et al p172
+
+% T = test for oil?
+% UT = utility (negative cost) of testing
+% O = amount of oil = Dry, Wet or Soaking
+% R = results of test = NoStrucure, OpenStructure, ClosedStructure or NoResult
+% D = drill?
+% UD = utility of drilling
+
+% Decision sequence = T R D O
+
+T = 1; UT = 2; O = 3; R = 4; D = 5; UD = 6;
+N = 6;
+dag = zeros(N);
+dag(T, [UT R D]) = 1;
+dag(O, [R UD]) = 1;
+dag(R, D) = 1;
+dag(D, UD) = 1;
+
+ns = zeros(1,N);
+ns(O) = 3; ns(R) = 4; ns(T) = 2; ns(D) = 2; ns(UT) = 1; ns(UD) = 1;
+
+limid = mk_limid(dag, ns, 'chance', [O R], 'decision', [T D], 'utility', [UT UD]);
+
+limid.CPD{O} = tabular_CPD(limid, O, [0.5 0.3 0.2]);
+tbl = [0.6 0 0.3 0 0.1 0  0.3 0 0.4 0 0.4 0  0.1 0 0.3 0 0.5 0  0 1 0 1 0 1];
+limid.CPD{R} = tabular_CPD(limid, R, tbl);
+
+limid.CPD{UT} = tabular_utility_node(limid, UT,  [-10 0]);
+limid.CPD{UD} = tabular_utility_node(limid, UD, [-70 50 200  0 0 0]);
+
+if 1
+  % start with uniform policies
+  limid.CPD{T} = tabular_decision_node(limid, T);
+  limid.CPD{D} = tabular_decision_node(limid, D);
+else
+  % hard code optimal policies
+  limid.CPD{T} = tabular_decision_node(limid, T, [1.0 0.0]);        
+  a = 0.5; b = 1-a; % arbitrary value
+  tbl = myreshape([0 a 1 a 1 a a a  1 b 0 b 0 b b b], ns([T R D]));
+  limid.CPD{D} = tabular_decision_node(limid, D,  tbl);
+end
+
+%fname = '/home/cs/murphyk/matlab/Misc/loopybel.txt';
+
+engines = {};
+engines{end+1} = global_joint_inf_engine(limid);
+engines{end+1} = jtree_limid_inf_engine(limid);
+%engines{end+1} = belprop_inf_engine(limid, 'max_iter', 3*N, 'filename', fname);
+
+exact = [1 2];
+%approx = 3;
+approx = [];
+
+E = length(engines);
+strategy = cell(1, E);
+MEU = zeros(1, E);
+for e=1:E
+  [strategy{e}, MEU(e)] = solve_limid(engines{e});
+  MEU
+end
+MEU
+
+for e=exact(:)'
+  assert(approxeq(MEU(e), 22.5))
+  % U(T=yes)  U(T=no)
+  % 1         0
+  assert(argmax(strategy{e}{T}) == 1); % test = yes
+  t = 1; % test = yes
+  % strategy{D} T       R      U(D=yes=1)  U(D=no=2)
+  %             1=yes   1=noS  0           1         Don't drill
+  %             2=no    1=noS  1           0
+  %             1=yes   2=opS  1           0
+  %             2=no    2=opS  1           0
+  %             1=yes   3=clS  1           0
+  %             2=no    3=clS  1           0
+  %             1=yes   4=unk  1           0
+  %             2=no    4=unk  1           0
+  
+  for r=[2 3] % OpS, ClS
+    assert(argmax(squeeze(strategy{e}{D}(t,r,:))) == 1); % drill = yes
+  end
+  r = 1; % noS
+  assert(argmax(squeeze(strategy{e}{D}(t,r,:))) == 2); % drill = no
+end
+
+
+for e=approx(:)'
+  approxeq(strategy{exact(1)}{T}, strategy{e}{T})
+  approxeq(strategy{exact(1)}{D}, strategy{e}{D})
+end
diff --git a/sourcecodes/bnt-master/BNT/examples/limids/pigs1.m b/sourcecodes/bnt-master/BNT/examples/limids/pigs1.m
new file mode 100644
index 00000000..6a56ad04
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/examples/limids/pigs1.m
@@ -0,0 +1,153 @@
+% pigs model from Lauritzen and Nilsson, 2001
+
+seed = 0;
+rand('state', seed);
+randn('state', seed);
+
+% we number nodes down and to the right
+h = [1 5 9 13];
+t = [2 6 10];
+d = [3 7 11];
+u = [4 8 12 14];
+
+N = 14;
+dag = zeros(N);
+
+% causal arcs
+for i=1:3
+  dag(h(i), [t(i) h(i+1)]) = 1;
+  dag(d(i), [u(i) h(i+1)]) = 1;
+end
+dag(h(4), u(4)) = 1;
+
+% information arcs
+fig = 2;
+switch fig
+ case 0,
+  % no info arcs
+ case 1,
+   % no-forgetting policy (figure 1)
+   for i=1:3
+     dag(t(i), d(i:3)) = 1;
+   end
+ case 2,
+  % reactive policy (figure 2)
+  for i=1:3
+    dag(t(i), d(i)) = 1;
+  end
+ case 7,
+  % omniscient policy (figure 7: di has access to hidden state h(i-1))
+  dag(t(1), d(1)) = 1;
+  for i=2:3
+    %dag([h(i-1) t(i-1) d(i-1)], d(i)) = 1;
+    dag([h(i-1) d(i-1)], d(i)) = 1; % t(i-1) is redundant given h(i-1)
+  end
+end
+
+
+ns = 2*ones(1,N);
+ns(u) = 1;
+
+% parameter tying
+params = ones(1,N);
+uparam = 1;
+final_uparam = 2;
+tparam = 3;
+h1_param = 4;
+hparam = 5;
+dparams = 6:8;
+
+params(u(1:3)) = uparam;
+params(u(4)) = final_uparam;
+params(t) = tparam;
+params(h(1)) = h1_param;
+params(h(2:end)) = hparam;
+params(d) = dparams;
+
+limid = mk_limid(dag, ns, 'chance', [h t], 'decision', d, 'utility', u, 'equiv_class', params);
+
+% h = 1 means healthy, h = 2 means diseased
+% d = 1 means don't treat, d = 2 means treat
+% t = 1 means test shows healthy, t = 2 means test shows diseased
+
+if 0
+  % use random params
+  limid.CPD{final_uparam} = tabular_utility_node(limid, u(4));
+  limid.CPD{uparam} = tabular_utility_node(limid, u(1));
+  limid.CPD{tparam} = tabular_CPD(limid, t(1));
+  limid.CPD{h1_param} = tabular_CPD(limid, h(1));
+  limid.CPD{hparam} = tabular_CPD(limid, h(2));
+else
+  limid.CPD{final_uparam} = tabular_utility_node(limid, u(4), [1000 300]);
+  limid.CPD{uparam} = tabular_utility_node(limid, u(1), [0 -100]); % costs have negative utility!
+  
+  % h  P(t=1) P(t=2)
+  % 1  0.9   0.1
+  % 2  0.2   0.8
+  limid.CPD{tparam} = tabular_CPD(limid, t(1), [0.9 0.2 0.1 0.8]);
+  
+  % P(h1)
+  limid.CPD{h1_param} = tabular_CPD(limid, h(1), [0.9 0.1]);
+  
+  % hi di P(hj=1) P(hj=2),  j = i+1, i=1:3
+  % 1  1  0.8     0.2
+  % 2  1  0.1     0.9
+  % 1  2  0.9     0.1
+  % 2  2  0.5     0.5
+  limid.CPD{hparam} = tabular_CPD(limid, h(2), [0.8 0.1 0.9 0.5 0.2 0.9 0.1 0.5]);
+end
+
+% Decision nodes get assigned uniform policies by default
+for i=1:3
+  limid.CPD{dparams(i)} = tabular_decision_node(limid, d(i));
+end
+
+
+fname = '/home/cs/murphyk/matlab/Misc/loopybel.txt';
+
+engines = {};
+engines{end+1} = global_joint_inf_engine(limid);
+engines{end+1} = jtree_limid_inf_engine(limid);
+%engines{end+1} = belprop_inf_engine(limid, 'max_iter', 1*N, 'filename', fname, 'tol', 1e-3);
+
+exact = [1 2];
+%approx = 3;
+approx = [];
+
+max_iter = 1;
+order = d(end:-1:1);
+%order = d(1:end);
+
+NE = length(engines);
+MEU = zeros(1, NE);
+niter = zeros(1, NE);
+strategy = cell(1, NE);
+for e=1:NE
+  [strategy{e}, MEU(e), niter(e)] = solve_limid(engines{e}, 'max_iter', max_iter, 'order',  order);
+end
+MEU
+
+% check results match those in the paper (p. 22)
+direct_policy = eye(2); % treat iff test is positive
+never_policy = [1 0; 1 0]; % never treat
+tol = 1e-0; % results in paper are reported to 0dp
+for e=exact(:)'
+  switch fig
+   case 2, % reactive policy
+    assert(approxeq(MEU(e), 727, tol));
+    assert(approxeq(strategy{e}{d(1)}(:), never_policy(:)))
+    assert(approxeq(strategy{e}{d(2)}(:), direct_policy(:)))
+    assert(approxeq(strategy{e}{d(3)}(:), direct_policy(:)))
+   case 1, assert(approxeq(MEU(e), 729, tol));
+   case 7, assert(approxeq(MEU(e), 732, tol));
+  end
+end
+
+
+for e=approx(:)'
+  for i=1:3
+    approxeq(strategy{exact(1)}{d(i)}, strategy{e}{d(i)})
+    dispcpt(strategy{e}{d(i)})
+  end
+end
+