Frequency-Based Hyperparameter Selection in Games

Aniket Sanyal, Baraah A. M. Sidahmed, Rebekka Burkholz, Tatjana Chavdarova
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:3808-3816, 2026.

Abstract

Learning in smooth games fundamentally differs from standard minimization due to rotational dynamics, which invalidate classical hyperparameter tuning strategies. Despite their practical importance, effective methods for tuning in games remain underexplored. A notable example is LookAhead (LA), which achieves strong empirical performance but introduces additional parameters that critically influence performance. We propose a principled approach to hyperparameter selection in games by leveraging frequency estimation of oscillatory dynamics. Specifically, we analyze oscillations both in continuous-time trajectories and through the spectrum of the discrete dynamics in the associated frequency-based space. Building on this analysis, we introduce \emph{Modal LookAhead (MoLA)}, an extension of LA that selects the hyperparameters adaptively to a given problem. We provide convergence guarantees and demonstrate in experiments that MoLA accelerates training in both purely rotational games and mixed regimes, all with minimal computational overhead.

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-sanyal26a, title = { Frequency-Based Hyperparameter Selection in Games }, author = {Sanyal, Aniket and Sidahmed, Baraah A. M. and Burkholz, Rebekka and Chavdarova, Tatjana}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {3808--3816}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/sanyal26a/sanyal26a.pdf}, url = {https://proceedings.mlr.press/v300/sanyal26a.html}, abstract = { Learning in smooth games fundamentally differs from standard minimization due to rotational dynamics, which invalidate classical hyperparameter tuning strategies. Despite their practical importance, effective methods for tuning in games remain underexplored. A notable example is LookAhead (LA), which achieves strong empirical performance but introduces additional parameters that critically influence performance. We propose a principled approach to hyperparameter selection in games by leveraging frequency estimation of oscillatory dynamics. Specifically, we analyze oscillations both in continuous-time trajectories and through the spectrum of the discrete dynamics in the associated frequency-based space. Building on this analysis, we introduce \emph{Modal LookAhead (MoLA)}, an extension of LA that selects the hyperparameters adaptively to a given problem. We provide convergence guarantees and demonstrate in experiments that MoLA accelerates training in both purely rotational games and mixed regimes, all with minimal computational overhead. } }
Endnote
%0 Conference Paper %T Frequency-Based Hyperparameter Selection in Games %A Aniket Sanyal %A Baraah A. M. Sidahmed %A Rebekka Burkholz %A Tatjana Chavdarova %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-sanyal26a %I PMLR %P 3808--3816 %U https://proceedings.mlr.press/v300/sanyal26a.html %V 300 %X Learning in smooth games fundamentally differs from standard minimization due to rotational dynamics, which invalidate classical hyperparameter tuning strategies. Despite their practical importance, effective methods for tuning in games remain underexplored. A notable example is LookAhead (LA), which achieves strong empirical performance but introduces additional parameters that critically influence performance. We propose a principled approach to hyperparameter selection in games by leveraging frequency estimation of oscillatory dynamics. Specifically, we analyze oscillations both in continuous-time trajectories and through the spectrum of the discrete dynamics in the associated frequency-based space. Building on this analysis, we introduce \emph{Modal LookAhead (MoLA)}, an extension of LA that selects the hyperparameters adaptively to a given problem. We provide convergence guarantees and demonstrate in experiments that MoLA accelerates training in both purely rotational games and mixed regimes, all with minimal computational overhead.
APA
Sanyal, A., Sidahmed, B.A.M., Burkholz, R. & Chavdarova, T.. (2026). Frequency-Based Hyperparameter Selection in Games . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:3808-3816 Available from https://proceedings.mlr.press/v300/sanyal26a.html.

Related Material