Posterior Weighted Reinforcement Learning with State Uncertainty