Towards Optimal Robustness in Learning-Augmented Paging

Peng Chen, Hailiang Zhao, Xueyan Tang, Yixuan Wang, Shuiguang Deng
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:18492-18516, 2026.

Abstract

Learning-augmented paging has been extensively studied in recent years. A key advantage over naive ML-based approaches is bounded robustness, which guarantees worst-case performance even when predictions are inaccurate, making these algorithms valuable for real-world systems. Prior work achieves robustness bounds of $2H_k + O(1)$ in the randomized setting, leaving a gap to the optimal competitive ratio $H_k$. In this paper, we study how to close this gap. We begin by reviewing online optimality and proving a new property of the latest $H_k$-competitive algorithm, which facilitates our analysis in the learning-augmented setting. Then, we review existing learning-augmented paging algorithms and introduce a unifying primitive, the relative prediction budget, which captures the essence of establishing robustness and reveals that prior algorithms either overuse or underutilize predictions. Guided by the above analysis, we develop a new framework that achieves the best-possible robustness up to an additive constant for learning-augmented paging: $H_k + O(1)$. Experiments further demonstrate strong practical performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chen26gt, title = {Towards Optimal Robustness in Learning-Augmented Paging}, author = {Chen, Peng and Zhao, Hailiang and Tang, Xueyan and Wang, Yixuan and Deng, Shuiguang}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {18492--18516}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/chen26gt/chen26gt.pdf}, url = {https://proceedings.mlr.press/v306/chen26gt.html}, abstract = {Learning-augmented paging has been extensively studied in recent years. A key advantage over naive ML-based approaches is bounded robustness, which guarantees worst-case performance even when predictions are inaccurate, making these algorithms valuable for real-world systems. Prior work achieves robustness bounds of $2H_k + O(1)$ in the randomized setting, leaving a gap to the optimal competitive ratio $H_k$. In this paper, we study how to close this gap. We begin by reviewing online optimality and proving a new property of the latest $H_k$-competitive algorithm, which facilitates our analysis in the learning-augmented setting. Then, we review existing learning-augmented paging algorithms and introduce a unifying primitive, the relative prediction budget, which captures the essence of establishing robustness and reveals that prior algorithms either overuse or underutilize predictions. Guided by the above analysis, we develop a new framework that achieves the best-possible robustness up to an additive constant for learning-augmented paging: $H_k + O(1)$. Experiments further demonstrate strong practical performance.} }
Endnote
%0 Conference Paper %T Towards Optimal Robustness in Learning-Augmented Paging %A Peng Chen %A Hailiang Zhao %A Xueyan Tang %A Yixuan Wang %A Shuiguang Deng %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-chen26gt %I PMLR %P 18492--18516 %U https://proceedings.mlr.press/v306/chen26gt.html %V 306 %X Learning-augmented paging has been extensively studied in recent years. A key advantage over naive ML-based approaches is bounded robustness, which guarantees worst-case performance even when predictions are inaccurate, making these algorithms valuable for real-world systems. Prior work achieves robustness bounds of $2H_k + O(1)$ in the randomized setting, leaving a gap to the optimal competitive ratio $H_k$. In this paper, we study how to close this gap. We begin by reviewing online optimality and proving a new property of the latest $H_k$-competitive algorithm, which facilitates our analysis in the learning-augmented setting. Then, we review existing learning-augmented paging algorithms and introduce a unifying primitive, the relative prediction budget, which captures the essence of establishing robustness and reveals that prior algorithms either overuse or underutilize predictions. Guided by the above analysis, we develop a new framework that achieves the best-possible robustness up to an additive constant for learning-augmented paging: $H_k + O(1)$. Experiments further demonstrate strong practical performance.
APA
Chen, P., Zhao, H., Tang, X., Wang, Y. & Deng, S.. (2026). Towards Optimal Robustness in Learning-Augmented Paging. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:18492-18516 Available from https://proceedings.mlr.press/v306/chen26gt.html.

Related Material