Beijing Zhongke Journal Publising Co. Ltd. Many real-world multi-agent scenarios can be naturally modeled as partially observable cooperative multi-agent reinforcement learning (MARL) problems. Agents ...