Summary: Learning Partially Observable Action Models: Efficient Algorithms
Dafna Shahaf Allen Chang Eyal Amir
Computer Science Department
University of Illinois, Urbana-Champaign
Urbana, IL 61801, USA
We present tractable, exact algorithms for learning actions'
effects and preconditions in partially observable domains.
Our algorithms maintain a propositional logical representa-
tion of the set of possible action models after each obser-
vation and action execution. The algorithms perform exact
learning of preconditions and effects in any deterministic ac-
tion domain. This includes STRIPS actions and actions with
conditional effects. In contrast, previous algorithms rely on
approximations to achieve tractability, and do not supply ap-
proximation guarantees. Our algorithms take time and space
that are polynomial in the number of domain features, and
can maintain a representation that stays compact indefinitely.
Our experimental results show that we can learn efficiently