Computing Optimal Regularizers for Online Linear Optimization

Khashayar Gatmiry, Jon Schneider, Stefanie Jegelka
Proceedings of Thirty Eighth Conference on Learning Theory, PMLR 291:2362-2402, 2025.

Abstract

Follow-the-Regularized-Leader (FTRL) algorithms are a popular class of learning algorithms for online linear optimization (OLO) that guarantee sub-linear regret. However, the choice of regularizer can significantly impact dimension-dependent factors in the regret bound. We present an algorithm that takes as input convex and symmetric action sets and loss sets for a specific OLO instance, and outputs a regularizer such that running FTRL with this regularizer guarantees regret within a universal constant factor of the best possible regret bound. In particular, for any choice of (convex, symmetric) action set and loss set we prove that there exists an instantiation of FTRL that achieves regret within a constant factor of the best possible learning algorithm, strengthening the universality result of Srebro et al., 2011. Our algorithm requires preprocessing time and space exponential in the dimension $d$ of the OLO instance, but can be run efficiently online assuming a membership and linear optimization oracle for the action and loss sets, respectively (and is fully polynomial time for the case of constant dimension $d$). We complement this with a lower bound showing that even deciding whether a given regularizer is $\alpha$-strongly-convex with respect to a given norm is NP-hard.

Cite this Paper

BibTeX
@InProceedings{pmlr-v291-gatmiry25a, title = {Computing Optimal Regularizers for Online Linear Optimization}, author = {Gatmiry, Khashayar and Schneider, Jon and Jegelka, Stefanie}, booktitle = {Proceedings of Thirty Eighth Conference on Learning Theory}, pages = {2362--2402}, year = {2025}, editor = {Haghtalab, Nika and Moitra, Ankur}, volume = {291}, series = {Proceedings of Machine Learning Research}, month = {30 Jun--04 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v291/main/assets/gatmiry25a/gatmiry25a.pdf}, url = {https://proceedings.mlr.press/v291/gatmiry25a.html}, abstract = {Follow-the-Regularized-Leader (FTRL) algorithms are a popular class of learning algorithms for online linear optimization (OLO) that guarantee sub-linear regret. However, the choice of regularizer can significantly impact dimension-dependent factors in the regret bound. We present an algorithm that takes as input convex and symmetric action sets and loss sets for a specific OLO instance, and outputs a regularizer such that running FTRL with this regularizer guarantees regret within a universal constant factor of the best possible regret bound. In particular, for any choice of (convex, symmetric) action set and loss set we prove that there exists an instantiation of FTRL that achieves regret within a constant factor of the best possible learning algorithm, strengthening the universality result of Srebro et al., 2011. Our algorithm requires preprocessing time and space exponential in the dimension $d$ of the OLO instance, but can be run efficiently online assuming a membership and linear optimization oracle for the action and loss sets, respectively (and is fully polynomial time for the case of constant dimension $d$). We complement this with a lower bound showing that even deciding whether a given regularizer is $\alpha$-strongly-convex with respect to a given norm is NP-hard.} }
Endnote
%0 Conference Paper %T Computing Optimal Regularizers for Online Linear Optimization %A Khashayar Gatmiry %A Jon Schneider %A Stefanie Jegelka %B Proceedings of Thirty Eighth Conference on Learning Theory %C Proceedings of Machine Learning Research %D 2025 %E Nika Haghtalab %E Ankur Moitra %F pmlr-v291-gatmiry25a %I PMLR %P 2362--2402 %U https://proceedings.mlr.press/v291/gatmiry25a.html %V 291 %X Follow-the-Regularized-Leader (FTRL) algorithms are a popular class of learning algorithms for online linear optimization (OLO) that guarantee sub-linear regret. However, the choice of regularizer can significantly impact dimension-dependent factors in the regret bound. We present an algorithm that takes as input convex and symmetric action sets and loss sets for a specific OLO instance, and outputs a regularizer such that running FTRL with this regularizer guarantees regret within a universal constant factor of the best possible regret bound. In particular, for any choice of (convex, symmetric) action set and loss set we prove that there exists an instantiation of FTRL that achieves regret within a constant factor of the best possible learning algorithm, strengthening the universality result of Srebro et al., 2011. Our algorithm requires preprocessing time and space exponential in the dimension $d$ of the OLO instance, but can be run efficiently online assuming a membership and linear optimization oracle for the action and loss sets, respectively (and is fully polynomial time for the case of constant dimension $d$). We complement this with a lower bound showing that even deciding whether a given regularizer is $\alpha$-strongly-convex with respect to a given norm is NP-hard.
APA
Gatmiry, K., Schneider, J. & Jegelka, S.. (2025). Computing Optimal Regularizers for Online Linear Optimization. Proceedings of Thirty Eighth Conference on Learning Theory, in Proceedings of Machine Learning Research 291:2362-2402 Available from https://proceedings.mlr.press/v291/gatmiry25a.html.

Related Material