Achieving Tractable Minimax Optimal Regret in Average Reward MDPs | Read Paper on Bytez