Rethinking the Utilization of Individual Rewards in Multiagent Reinforcement Learning With Sparse Team Rewards | Synapse