The State-Action-Reward-State-Action Algorithm in Spatial Prisoner's Dilemma Game