Exploration and Anti-Exploration with Distributional Random Network Distillation

Kai Yang, Jian Tao, Jiafei Lyu, Xiu Li
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:56397-56421, 2024.

Abstract

Exploration remains a critical issue in deep reinforcement learning for an agent to attain high returns in unknown environments. Although the prevailing exploration Random Network Distillation (RND) algorithm has been demonstrated to be effective in numerous environments, it often needs more discriminative power in bonus allocation. This paper highlights the “bonus inconsistency” issue within RND, pinpointing its primary limitation. To address this issue, we introduce the Distributional RND (DRND), a derivative of the RND. DRND enhances the exploration process by distilling a distribution of random networks and implicitly incorporating pseudo counts to improve the precision of bonus allocation. This refinement encourages agents to engage in more extensive exploration. Our method effectively mitigates the inconsistency issue without introducing significant computational overhead. Both theoretical analysis and experimental results demonstrate the superiority of our approach over the original RND algorithm. Our method excels in challenging online exploration scenarios and effectively serves as an anti-exploration mechanism in D4RL offline tasks. Our code is publicly available at https://github.com/yk7333/DRND.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-yang24w, title = {Exploration and Anti-Exploration with Distributional Random Network Distillation}, author = {Yang, Kai and Tao, Jian and Lyu, Jiafei and Li, Xiu}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {56397--56421}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/yang24w/yang24w.pdf}, url = {https://proceedings.mlr.press/v235/yang24w.html}, abstract = {Exploration remains a critical issue in deep reinforcement learning for an agent to attain high returns in unknown environments. Although the prevailing exploration Random Network Distillation (RND) algorithm has been demonstrated to be effective in numerous environments, it often needs more discriminative power in bonus allocation. This paper highlights the “bonus inconsistency” issue within RND, pinpointing its primary limitation. To address this issue, we introduce the Distributional RND (DRND), a derivative of the RND. DRND enhances the exploration process by distilling a distribution of random networks and implicitly incorporating pseudo counts to improve the precision of bonus allocation. This refinement encourages agents to engage in more extensive exploration. Our method effectively mitigates the inconsistency issue without introducing significant computational overhead. Both theoretical analysis and experimental results demonstrate the superiority of our approach over the original RND algorithm. Our method excels in challenging online exploration scenarios and effectively serves as an anti-exploration mechanism in D4RL offline tasks. Our code is publicly available at https://github.com/yk7333/DRND.} }
Endnote
%0 Conference Paper %T Exploration and Anti-Exploration with Distributional Random Network Distillation %A Kai Yang %A Jian Tao %A Jiafei Lyu %A Xiu Li %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-yang24w %I PMLR %P 56397--56421 %U https://proceedings.mlr.press/v235/yang24w.html %V 235 %X Exploration remains a critical issue in deep reinforcement learning for an agent to attain high returns in unknown environments. Although the prevailing exploration Random Network Distillation (RND) algorithm has been demonstrated to be effective in numerous environments, it often needs more discriminative power in bonus allocation. This paper highlights the “bonus inconsistency” issue within RND, pinpointing its primary limitation. To address this issue, we introduce the Distributional RND (DRND), a derivative of the RND. DRND enhances the exploration process by distilling a distribution of random networks and implicitly incorporating pseudo counts to improve the precision of bonus allocation. This refinement encourages agents to engage in more extensive exploration. Our method effectively mitigates the inconsistency issue without introducing significant computational overhead. Both theoretical analysis and experimental results demonstrate the superiority of our approach over the original RND algorithm. Our method excels in challenging online exploration scenarios and effectively serves as an anti-exploration mechanism in D4RL offline tasks. Our code is publicly available at https://github.com/yk7333/DRND.
APA
Yang, K., Tao, J., Lyu, J. & Li, X.. (2024). Exploration and Anti-Exploration with Distributional Random Network Distillation. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:56397-56421 Available from https://proceedings.mlr.press/v235/yang24w.html.

Related Material