Memory-Based Reinforcement Learning: Efficient Computation with Prioritized Sweeping