wolffd@0: function CPD = hhmmF_CPD(bnet, self, Qself, Fbelow, varargin)
wolffd@0: % HHMMF_CPD Make the CPD for an F node in a hierarchical HMM
wolffd@0: % CPD = hhmmF_CPD(bnet, self, Qself,  Fbelow, ...)
wolffd@0: %
wolffd@0: %        Qps
wolffd@0: %          \
wolffd@0: %           \
wolffd@0: %           Fself
wolffd@0: %         /   |
wolffd@0: %        /    |
wolffd@0: %       Qself Fbelow
wolffd@0: %
wolffd@0: % We assume nodes are ordered (numbered) as follows: Qps, Q, Fbelow, F
wolffd@0: % All nodes numbers should be from slice 1.
wolffd@0: %
wolffd@0: % If Fbelow if missing, this becomes a regular tabular_CPD.
wolffd@0: % Qps may be omitted.
wolffd@0: %
wolffd@0: % optional args [defaults]
wolffd@0: % 
wolffd@0: % Qps - node numbers.
wolffd@0: % termprob - termprob(k,i,2) = prob finishing given Q(d)=i and Q(1:d-1)=k [ finish in last state wp 0.9]
wolffd@0: %
wolffd@0: % hhmmF_CPD is a subclass of tabular_CPD so we inherit inference methods like CPD_to_pot, etc.
wolffd@0: %
wolffd@0: % We create an isolated tabular_CPD with no F parent to learn termprob
wolffd@0: % so we can avail of e.g., entropic or Dirichlet priors.
wolffd@0: %
wolffd@0: % For details, see "Linear-time inference in hierarchical HMMs", Murphy and Paskin, NIPS'01.
wolffd@0: 
wolffd@0: 
wolffd@0: 
wolffd@0: Qps = [];
wolffd@0: % get parents
wolffd@0: for i=1:2:length(varargin)
wolffd@0:   switch varargin{i},
wolffd@0:    case 'Qps', Qps = varargin{i+1}; 
wolffd@0:   end
wolffd@0: end
wolffd@0: 
wolffd@0: ns = bnet.node_sizes(:);
wolffd@0: Qsz = ns(Qself);
wolffd@0: Qpsz = prod(ns(Qps));
wolffd@0: CPD.Qsz = Qsz;
wolffd@0: CPD.Qpsz = Qpsz;
wolffd@0: 
wolffd@0: ps = parents(bnet.dag, self);
wolffd@0: CPD.Fbelow_ndx = find_equiv_posns(Fbelow, ps);
wolffd@0: CPD.Qps_ndx = find_equiv_posns(Qps, ps);
wolffd@0: CPD.Qself_ndx = find_equiv_posns(Qself, ps);
wolffd@0: 
wolffd@0: % set default arguments
wolffd@0: p = 0.9;
wolffd@0: %termprob(k,i,t) Might terminate if i=Qsz; will not terminate if i<Qsz
wolffd@0: termprob = zeros(Qpsz, Qsz, 2);
wolffd@0: termprob(:, Qsz, 2) = p; 
wolffd@0: termprob(:, Qsz, 1) = 1-p; 
wolffd@0: termprob(:, 1:(Qsz-1), 1) = 1; 
wolffd@0:     
wolffd@0: for i=1:2:length(varargin)
wolffd@0:   switch varargin{i},
wolffd@0:    case 'termprob', termprob = varargin{i+1}; 
wolffd@0:   end
wolffd@0: end
wolffd@0: 
wolffd@0: CPD.sub_CPD_term = mk_isolated_tabular_CPD([Qpsz Qsz 2], {'CPT', termprob});
wolffd@0: S = struct(CPD.sub_CPD_term);
wolffd@0: CPD.termprob = S.CPT;
wolffd@0: 
wolffd@0: CPD = class(CPD, 'hhmmF_CPD', tabular_CPD(bnet, self));
wolffd@0: 
wolffd@0: CPD = update_CPT(CPD);
wolffd@0: