Role-Level Inductive Bias for Cross-Task Generalization in Multi-Agent Reinforcement Learning
Chang Yao, Youfang Lin, Shoucheng Song, Hao Wu, Shengkun Yang, Yuqing Ma, Kai Lv
Abstract
Achieving cross-task generalization remains a critical challenge in Multi-Agent Reinforcement Learning (MARL), fundamentally relying on effective inductive biases. However, existing entity-level biases often overlook collaborative patterns, whereas task-level biases lack sufficient coverage for novel scenarios. To address this, we introduce a role-level inductive bias as an intermediate abstraction that integrates entity-level flexibility with task-level inter-agent collaboration. To instantiate this, we propose Gaussian-mixture-model-based Transferable Role discovery (GTR). Specifically, GTR constructs a structured role space to ensure diverse role assignment, further achieves role decoupling via regularization, and ultimately utilizes these roles for efficient generalization. Empirical results demonstrate that GTR achieves superior zero-shot and few-shot transfer performance on unseen tasks compared to state-of-the-art methods.
BibTeX
@inproceedings{
yao2026rolelevel,
title={Role-Level Inductive Bias for Cross-Task Generalization in Multi-Agent Reinforcement Learning},
author={Chang Yao and Youfang Lin and Shoucheng Song and Hao Wu and Shengkun Yang and Yuqing Ma and Kai Lv},
booktitle={Forty-third International Conference on Machine Learning},
year={2026},
url={https://openreview.net/forum?id=oz8kFbXdpj}
}