Optimistic Q-learning for average reward and episodic reinforcement learning

Re-assign community