Efficient Skill Discovery via Regret-Aware Optimization

He Zhang, Ming Zhou, Shaopeng Zhai, Ying Sun, Hui Xiong
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:77216-77232, 2025.

Abstract

Unsupervised skill discovery aims to learn diverse and distinguishable behaviors in open-ended reinforcement learning. For the existing methods, they focus on improving the diversity via pure exploration, mutual information optimization and learning temporal representation. Despite they perform well on exploration, they remain limited in terms of efficiency, especially for the high-dimensional situations. In this work, we frame the skill discovery as a min-max game of skill generation and policy learning, proposing a regret-aware method on top of temporal representation learning that expands the discovered skill space along the direction of upgradable policy strength. The key insight behind the proposed method is that the skill discovery is adversarial to the policy learning, i.e., skills with weak strength should be further explored while less exploration for the skills with converged strength. As an implementation, we score the degree of strength convergence with regret, and guide the skill discovery with a learnable skill generator. To avoid degeneration, the skill generation comes from an upgradable population of skill generators. We conduct experiments on environments with varying complexities and dimension sizes. Empirical results show that our method outperforms baselines on both efficiency and diversity. Moreover, our method achieves 15% zero-shot improvement on high-dimensional environments, compared to existing methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-zhang25dp, title = {Efficient Skill Discovery via Regret-Aware Optimization}, author = {Zhang, He and Zhou, Ming and Zhai, Shaopeng and Sun, Ying and Xiong, Hui}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {77216--77232}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/zhang25dp/zhang25dp.pdf}, url = {https://proceedings.mlr.press/v267/zhang25dp.html}, abstract = {Unsupervised skill discovery aims to learn diverse and distinguishable behaviors in open-ended reinforcement learning. For the existing methods, they focus on improving the diversity via pure exploration, mutual information optimization and learning temporal representation. Despite they perform well on exploration, they remain limited in terms of efficiency, especially for the high-dimensional situations. In this work, we frame the skill discovery as a min-max game of skill generation and policy learning, proposing a regret-aware method on top of temporal representation learning that expands the discovered skill space along the direction of upgradable policy strength. The key insight behind the proposed method is that the skill discovery is adversarial to the policy learning, i.e., skills with weak strength should be further explored while less exploration for the skills with converged strength. As an implementation, we score the degree of strength convergence with regret, and guide the skill discovery with a learnable skill generator. To avoid degeneration, the skill generation comes from an upgradable population of skill generators. We conduct experiments on environments with varying complexities and dimension sizes. Empirical results show that our method outperforms baselines on both efficiency and diversity. Moreover, our method achieves 15% zero-shot improvement on high-dimensional environments, compared to existing methods.} }
Endnote
%0 Conference Paper %T Efficient Skill Discovery via Regret-Aware Optimization %A He Zhang %A Ming Zhou %A Shaopeng Zhai %A Ying Sun %A Hui Xiong %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-zhang25dp %I PMLR %P 77216--77232 %U https://proceedings.mlr.press/v267/zhang25dp.html %V 267 %X Unsupervised skill discovery aims to learn diverse and distinguishable behaviors in open-ended reinforcement learning. For the existing methods, they focus on improving the diversity via pure exploration, mutual information optimization and learning temporal representation. Despite they perform well on exploration, they remain limited in terms of efficiency, especially for the high-dimensional situations. In this work, we frame the skill discovery as a min-max game of skill generation and policy learning, proposing a regret-aware method on top of temporal representation learning that expands the discovered skill space along the direction of upgradable policy strength. The key insight behind the proposed method is that the skill discovery is adversarial to the policy learning, i.e., skills with weak strength should be further explored while less exploration for the skills with converged strength. As an implementation, we score the degree of strength convergence with regret, and guide the skill discovery with a learnable skill generator. To avoid degeneration, the skill generation comes from an upgradable population of skill generators. We conduct experiments on environments with varying complexities and dimension sizes. Empirical results show that our method outperforms baselines on both efficiency and diversity. Moreover, our method achieves 15% zero-shot improvement on high-dimensional environments, compared to existing methods.
APA
Zhang, H., Zhou, M., Zhai, S., Sun, Y. & Xiong, H.. (2025). Efficient Skill Discovery via Regret-Aware Optimization. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:77216-77232 Available from https://proceedings.mlr.press/v267/zhang25dp.html.

Related Material