Abstract: Existing multi-agent reinforcement learning (MARL) in adaptive traffic signal control (ATSC) typically models cooperative control of multiple intersections as a cooperative Markov game, ...