Sciweavers

Share
AGI
2011

Reinforcement Learning and the Bayesian Control Rule

8 years 2 months ago
Reinforcement Learning and the Bayesian Control Rule
We present an actor-critic scheme for reinforcement learning in complex domains. The main contribution is to show that planning and I/O dynamics can be separated such that an intractable planning problem reduces to a simple multi-armed bandit problem, where each lever stands for a potentially arbitrarily complex policy. Furthermore, we use the Bayesian control rule to construct an adaptive bandit player that is universal with respect to a given class of optimal bandit players, thus indirectly constructing an adaptive agent that is universal with respect to a given class of policies.
Pedro Alejandro Ortega, Daniel Alexander Braun, Si
Added 24 Aug 2011
Updated 24 Aug 2011
Type Journal
Year 2011
Where AGI
Authors Pedro Alejandro Ortega, Daniel Alexander Braun, Simon J. Godsill
Comments (0)
books