Reward Function In Reinforcement Learning