A Human Mixed Strategy Approach to Deep Reinforcement Learning