Non-Stationary Lipschitz Bandits

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Nguyen, Nicolas, Gaucher, Solenne, Vernade, Claire
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866912663404019712
author Nguyen, Nicolas
Gaucher, Solenne
Vernade, Claire
author_facet Nguyen, Nicolas
Gaucher, Solenne
Vernade, Claire
contents We study the problem of non-stationary Lipschitz bandits, where the number of actions is infinite and the reward function, satisfying a Lipschitz assumption, can change arbitrarily over time. We design an algorithm that adaptively tracks the recently introduced notion of significant shifts, defined by large deviations of the cumulative reward function. To detect such reward changes, our algorithm leverages a hierarchical discretization of the action space. Without requiring any prior knowledge of the non-stationarity, our algorithm achieves a minimax-optimal dynamic regret bound of $\mathcal{\widetilde{O}}(\tilde{L}^{1/3}T^{2/3})$, where $\tilde{L}$ is the number of significant shifts and $T$ the horizon. This result provides the first optimal guarantee in this setting.
format Preprint
id arxiv_https___arxiv_org_abs_2505_18871
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Non-Stationary Lipschitz Bandits
Nguyen, Nicolas
Gaucher, Solenne
Vernade, Claire
Machine Learning
We study the problem of non-stationary Lipschitz bandits, where the number of actions is infinite and the reward function, satisfying a Lipschitz assumption, can change arbitrarily over time. We design an algorithm that adaptively tracks the recently introduced notion of significant shifts, defined by large deviations of the cumulative reward function. To detect such reward changes, our algorithm leverages a hierarchical discretization of the action space. Without requiring any prior knowledge of the non-stationarity, our algorithm achieves a minimax-optimal dynamic regret bound of $\mathcal{\widetilde{O}}(\tilde{L}^{1/3}T^{2/3})$, where $\tilde{L}$ is the number of significant shifts and $T$ the horizon. This result provides the first optimal guarantee in this setting.
title Non-Stationary Lipschitz Bandits
topic Machine Learning
url https://arxiv.org/abs/2505.18871