Improved Regret Bounds for Linear Adversarial MDPs via Linear Optimization | Read Paper on Bytez