An implementation of EUBRL, a Bayesian RL algorithm that leverages epistemic guidance to achieve principled exploration.

python install.py

Examples are given for running EUBRL. Any other algorithms in the same folder can be run in a similar manner.

bash scripts/Chain/EUBRL.sh

bash scripts/Loop/EUBRL.sh

bash scripts/DeepSea-Stochastic/EUBRL.sh

bash scripts/DeepSea-Deterministic/EUBRL.sh

bash scripts/LazyChain-Stochastic/EUBRL.sh

bash scripts/LazyChain-Deterministic/EUBRL.sh

Our code is based on open-source repository BayesRL.

This project is licensed under the GNU General Public License v3.0 (GPLv3).

Should you find this work useful for your research, please consider citing:

@inproceedings{ma2026eubrl,
  title={{EUBRL}: Epistemic Uncertainty Directed Bayesian Reinforcement Learning},
  author={Jianfei Ma and Wee Sun Lee},
  booktitle={The Fourteenth International Conference on Learning Representations},
  year={2026},
  url={https://openreview.net/forum?id=KASqlcI6Nm}
}