Deep Coordination Graphs
This addresses coordination challenges in multi-agent systems, offering a method that balances representational flexibility with efficiency, though it appears incremental as it builds on existing coordination graph concepts.
The paper tackles the problem of collaborative multi-agent reinforcement learning by introducing the deep coordination graph (DCG), which factors joint value functions into pairwise payoffs for improved trade-offs between capacity and generalization, and demonstrates its ability to solve predator-prey and StarCraft II tasks.
This paper introduces the deep coordination graph (DCG) for collaborative multi-agent reinforcement learning. DCG strikes a flexible trade-off between representational capacity and generalization by factoring the joint value function of all agents according to a coordination graph into payoffs between pairs of agents. The value can be maximized by local message passing along the graph, which allows training of the value function end-to-end with Q-learning. Payoff functions are approximated with deep neural networks that employ parameter sharing and low-rank approximations to significantly improve sample efficiency. We show that DCG can solve predator-prey tasks that highlight the relative overgeneralization pathology, as well as challenging StarCraft II micromanagement tasks.