Action-Dependent Optimality-Preserving Reward Shaping | Read Paper on Bytez