PRODIGY
- The PRODIGY system uses EBL to
improve search. It does AI planning.
- It uses a means-ends planner that decomposes problems into
subgoals, solves them, then combines their solutions into a
solution for the full problem.
- At each step it faces the question "which operator should
I consider for solving this subgoal?"
- EBL in Prodigy chooses appropriate target concepts, such as
"the set of states in which subgoal A should be solved before
subgoal B". A sample rule learned is
SolveBefore(On(y,z), On(x,y)) ← NeedToSolve(On(y,z)) ∧ NeedToSolve(On(z,y))
- Prodigy learns this rule by first doing it wrong,
encountering the conflict, and explaining to itself why it had
to unstack the block it had just stacked.
José M. Vidal
.
26 of 29