Reinforcement Learning Basis Independent