この資料をSMS送信: Analysing factorizations of action-value networks for cooperative multi-agent reinforcement learning