Learning Reward Functions for Cooperative Resilience in Multi-Agent Systems
This work addresses the challenge of fostering resilient cooperation in multi-agent systems operating in dynamic and uncertain environments, representing an incremental step toward robust systems.
The paper tackled the problem of cooperative resilience in mixed-motive multi-agent systems by learning reward functions from ranked trajectories, showing that a hybrid reward strategy significantly improves robustness under disruptions without degrading task performance and reduces catastrophic outcomes like resource overuse.
Multi-agent systems often operate in dynamic and uncertain environments, where agents must not only pursue individual goals but also safeguard collective functionality. This challenge is especially acute in mixed-motive multi-agent systems. This work focuses on cooperative resilience, the ability of agents to anticipate, resist, recover, and transform in the face of disruptions, a critical yet underexplored property in Multi-Agent Reinforcement Learning. We study how reward function design influences resilience in mixed-motive settings and introduce a novel framework that learns reward functions from ranked trajectories, guided by a cooperative resilience metric. Agents are trained in a suite of social dilemma environments using three reward strategies: i) traditional individual reward; ii) resilience-inferred reward; and iii) hybrid that balance both. We explore three reward parameterizations-linear models, hand-crafted features, and neural networks, and employ two preference-based learning algorithms to infer rewards from behavioral rankings. Our results demonstrate that hybrid strategy significantly improve robustness under disruptions without degrading task performance and reduce catastrophic outcomes like resource overuse. These findings underscore the importance of reward design in fostering resilient cooperation, and represent a step toward developing robust multi-agent systems capable of sustaining cooperation in uncertain environments.