Distributed Value Decomposition Networks with Networked Agents
This work addresses the problem of distributed training for cooperative multi-agent reinforcement learning agents in domains where centralized training is not possible, which is significant for applications in decentralized multi-agent systems.
The authors tackled the problem of distributed training under partial observability in cooperative multi-agent reinforcement learning, achieving performance comparable to centralized training methods in ten MARL tasks. Their approach, DVDN, approximates the performance of value decomposition networks despite information loss during communication.
We investigate the problem of distributed training under partial observability, whereby cooperative multi-agent reinforcement learning agents (MARL) maximize the expected cumulative joint reward. We propose distributed value decomposition networks (DVDN) that generate a joint Q-function that factorizes into agent-wise Q-functions. Whereas the original value decomposition networks rely on centralized training, our approach is suitable for domains where centralized training is not possible and agents must learn by interacting with the physical environment in a decentralized manner while communicating with their peers. DVDN overcomes the need for centralized training by locally estimating the shared objective. We contribute with two innovative algorithms, DVDN and DVDN (GT), for the heterogeneous and homogeneous agents settings respectively. Empirically, both algorithms approximate the performance of value decomposition networks, in spite of the information loss during communication, as demonstrated in ten MARL tasks in three standard environments.