Scaling Up Multiagent Reinforcement Learning for Robotic Systems: Learn an Adaptive Sparse Communication Graph
This addresses the scalability problem for researchers and practitioners applying MARL to large-scale robotic systems, though it is incremental by building on existing attention and graph neural network methods.
The paper tackles the scalability issue in multiagent reinforcement learning (MARL) by exploiting sparse interactions between agents, proposing an adaptive sparse attention mechanism that learns a sparse communication graph to reduce sample complexity and improve efficiency. Results show the algorithm outperforms previous works by a significant margin on large-scale multiagent systems.
The complexity of multiagent reinforcement learning (MARL) in multiagent systems increases exponentially with respect to the agent number. This scalability issue prevents MARL from being applied in large-scale multiagent systems. However, one critical feature in MARL that is often neglected is that the interactions between agents are quite sparse. Without exploiting this sparsity structure, existing works aggregate information from all of the agents and thus have a high sample complexity. To address this issue, we propose an adaptive sparse attention mechanism by generalizing a sparsity-inducing activation function. Then a sparse communication graph in MARL is learned by graph neural networks based on this new attention mechanism. Through this sparsity structure, the agents can communicate in an effective as well as efficient way via only selectively attending to agents that matter the most and thus the scale of the MARL problem is reduced with little optimality compromised. Comparative results show that our algorithm can learn an interpretable sparse structure and outperforms previous works by a significant margin on applications involving a large-scale multiagent system.