Reinforcement learning algorithms often suffer from poor sample efficiency, making them challenging to apply in multi-task or continual learning settings. Efficiency can be improved by transferring knowledge from a previously trained teacher policy to guide exploration in new but related tasks. However, if the new task sufficiently differs from the teacher's training task, the transferred guidance may be sub-optimal and bias exploration toward low-reward behaviors. We propose an energy-based transfer learning method that uses out-of-distribution detection to selectively issue guidance, enabling the teacher to intervene only in states within its training distribution. We theoretically show that energy scores reflect the teacher's state-visitation density and empirically demonstrate improved sample efficiency and performance across both single-task and multi-task settings.
@article{arxiv.2506.16590,
title = {Energy-Based Transfer for Reinforcement Learning},
author = {Zeyun Deng and Jasorsi Ghosh and Fiona Xie and Yuzhe Lu and Katia Sycara and Joseph Campbell},
journal= {arXiv preprint arXiv:2506.16590},
year = {2025}
}