Improve Gibbs learning in CLP(BN).

This commit is contained in:
Vitor Santos Costa 2008-11-14 14:56:18 +00:00
parent 1be78e23ec
commit 1b98de440d
4 changed files with 55 additions and 12 deletions

View File

@ -39,7 +39,7 @@ clpbn_not_var_member([V1|Vs], V) :- V1 \== V,
sort_vars_by_key(AVars, SortedAVars, UnifiableVars) :-
get_keys(AVars, KeysVars),
keysort(KeysVars, KVars),
msort(KeysVars, KVars),
merge_same_key(KVars, SortedAVars, [], UnifiableVars).
get_keys([], []).
@ -51,7 +51,19 @@ get_keys([_|AVars], KeysVars) :- % may be non-CLPBN vars.
merge_same_key([], [], _, []).
merge_same_key([K1-V1,K2-V2|Vs], SortedAVars, Ks, UnifiableVars) :-
K1 == K2, !, V1 = V2,
K1 == K2, !,
(clpbn:get_atts(V1, [evidence(E)])
->
clpbn:put_atts(V2, [evidence(E)])
;
clpbn:get_atts(V2, [evidence(E)])
->
clpbn:put_atts(V1, [evidence(E)])
;
true
),
% V1 = V2,
attributes:fast_unify_attributed(V1,V2),
merge_same_key([K1-V1|Vs], SortedAVars, Ks, UnifiableVars).
merge_same_key([K1-V1,K2-V2|Vs], [V1|SortedAVars], Ks, [K1|UnifiableVars]) :-
(in_keys(K1, Ks) ; \+ \+ K1 == K2), !,

View File

@ -14,7 +14,7 @@
*********************************/
:- module(vel, [vel/3,
:- module(clpbn_vel, [vel/3,
check_if_vel_done/1,
init_vel_solver/4,
run_vel_solver/3]).
@ -92,8 +92,12 @@ init_vel_solver_for_questions([Vs|MVs], G, RG, [NVs|MNVs0], [NVs|LVis]) :-
%clpbn_gviz:clpbn2gviz(user_error, test, NVs, Vs),
init_vel_solver_for_questions(MVs, G, RG, MNVs0, LVis).
run_vel_solver([], [], []).
run_vel_solver([LVs|MoreLVs], [Ps|MorePs], [NVs0|MoreLVis]) :-
% use a findall to recover space without needing for GC
run_vel_solver(LVs, LPs, LNVs) :-
findall(Ps, solve_vel(LVs, LNVs, Ps), LPs).
solve_vel([LVs|_], [NVs0|_], Ps) :-
length(NVs0, L), (L > 64 -> clpbn_gviz:clpbn2gviz(user_error,sort,NVs0,LVs) ; true ),
find_all_clpbn_vars(NVs0, NVs0, LV0, LVi, Tables0),
sort(LV0, LV),
% construct the graph
@ -101,9 +105,19 @@ run_vel_solver([LVs|MoreLVs], [Ps|MorePs], [NVs0|MoreLVis]) :-
process(LVi, LVs, tab(Dist,_,_)),
% move from potentials back to probabilities
normalise_CPT(Dist,MPs),
list_from_CPT(MPs, Ps),
length(Ps,_Len),
run_vel_solver(MoreLVs, MorePs, MoreLVis).
list_from_CPT(MPs, Ps).
solve_vel([_|MoreLVs], [_|MoreLVis], Ps) :-
solve_vel(MoreLVs, MoreLVis, Ps).
keys([],[]).
keys([V|NVs0],[K:E|Ks]) :-
clpbn:get_atts(V,[key(K),evidence(E)]), !,
keys(NVs0,Ks).
keys([V|NVs0],[K|Ks]) :-
clpbn:get_atts(V,[key(K)]),
keys(NVs0,Ks).
%
% just get a list of variables plus associated tables

View File

@ -197,7 +197,7 @@ cpt_score(Lik) :-
clpbn_flag(em_solver, EMSolver),
set_clpbn_flag(solver, EMSolver),
reset_all_dists,
em(Exs, 0.1, 10, _Tables, Lik),
em(Exs, 0.01, 10, _Tables, Lik),
set_clpbn_flag(solver, Solver).
complete_clpbn_cost(_AlephClause).

View File

@ -50,7 +50,7 @@
:- meta_predicate em(:,+,+,-,-), init_em(:,-).
em(Items, MaxError, MaxIts, Tables, Likelihood) :-
init_em(Items, State),
catch(init_em(Items, State),Error,handle_em(Error)),
em_loop(0, 0.0, State, MaxError, MaxIts, Likelihood, Tables),
assert(em_found(Tables, Likelihood)),
fail.
@ -58,6 +58,13 @@ em(Items, MaxError, MaxIts, Tables, Likelihood) :-
em(_, _, _, Tables, Likelihood) :-
retract(em_found(Tables, Likelihood)).
handle_em(error(repeated_parents)) :-
assert(em_found(_, -inf)),
fail.
% This gets you an initial configuration. If there is a lot of evidence
% tables may be filled in close to optimal, otherwise they may be
% close to uniform.
@ -72,9 +79,9 @@ init_em(Items, state( AllDists, AllDistInstances, MargVars, SolverVars)) :-
% randomise_all_dists,
uniformise_all_dists,
attributes:all_attvars(AllVars0),
sort_vars_by_key(AllVars0,AllVars1,[]),
sort_vars_by_key(AllVars0,AllVars,[]),
% remove variables that do not have to do with this query.
check_for_hidden_vars(AllVars1, AllVars1, AllVars),
% check_for_hidden_vars(AllVars1, AllVars1, AllVars),
different_dists(AllVars, AllDists, AllDistInstances, MargVars),
clpbn_flag(em_solver, Solver),
clpbn_init_solver(Solver, MargVars, AllVars, _, SolverVars).
@ -116,6 +123,16 @@ different_dists(AllVars, AllDists, AllInfo, MargVars) :-
all_dists([], []).
all_dists([V|AllVars], [i(Id, [V|Parents], Cases, Hiddens)|Dists]) :-
clpbn:get_atts(V, [dist(Id,Parents)]),
sort([V|Parents], Sorted),
length(Sorted, LengSorted),
length(Parents, LengParents),
(
LengParents+1 =:= LengSorted
->
true
;
throw(error(repeated_parents))
),
generate_hidden_cases([V|Parents], CompactCases, Hiddens),
uncompact_cases(CompactCases, Cases),
all_dists(AllVars, Dists).