Weighted Linear Bandits for Non-Stationary Environments - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

Weighted Linear Bandits for Non-Stationary Environments

Claire Vernade

Résumé

We consider a stochastic linear bandit model in which the available actions correspond to arbitrary context vectors whose associated rewards follow a non-stationary linear regression model. In this setting, the unknown regression parameter is allowed to vary in time. To address this problem, we propose D-LinUCB, a novel optimistic algorithm based on discounted linear regression, where exponential weights are used to smoothly forget the past. This involves studying the deviations of the sequential weighted least-squares estimator under generic assumptions. As a by-product, we obtain novel deviation results that can be used beyond non-stationary environments. We provide theoretical guarantees on the behavior of D-LinUCB in both slowly-varying and abruptly-changing environments. We obtain an upper bound on the dynamic regret that is of order d^{2/3} B_T^{1/3}T^{2/3}, where B_T is a measure of non-stationarity (d and T being, respectively, dimension and horizon). This rate is known to be optimal. We also illustrate the empirical performance of D-LinUCB and compare it with recently proposed alternatives in simulated environments.
Fichier principal
Vignette du fichier
main.pdf (475.33 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02291460 , version 1 (18-09-2019)
hal-02291460 , version 2 (19-03-2020)

Identifiants

Citer

Yoan Russac, Claire Vernade, Olivier Cappé. Weighted Linear Bandits for Non-Stationary Environments. NeurIPS 2019 - 33rd Conference on Neural Information Processing Systems, Dec 2019, Vancouver, Canada. ⟨hal-02291460v1⟩
204 Consultations
625 Téléchargements

Altmetric

Partager

Gmail Mastodon Facebook X LinkedIn More