[2206.01085] Incorporating Explicit Uncertainty Estimates into Deep Offline Reinforcement Learning