Improved sleeping bandits with stochastic action sets and adversarial rewards - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2020

Improved sleeping bandits with stochastic action sets and adversarial rewards

Michal Valko

Résumé

In this paper, we consider the problem of sleeping bandits with stochastic action sets and adversarial rewards. In this setting, in contrast to most work in bandits, the actions may not be available at all times. For instance, some products might be out of stock in item recommendation. The best existing efficient (i.e., polynomial-time) algorithms for this problem only guarantee an $O(T 2/3)$ upper-bound on the regret. Yet, inefficient algorithms based on EXP4 can achieve $O(√ T)$. In this paper , we provide a new computationally efficient algorithm inspired by EXP3 satisfying a regret of order $O(√ T)$ when the availabilities of each action $i ∈ A$ are independent. We then study the most general version of the problem where at each round available sets are generated from some unknown arbitrary distribution (i.e., without the independence assumption) and propose an efficient algorithm with $O(√ 2 K T)$ regret guarantee. Our theoretical results are corroborated with experimental evaluations.
Fichier principal
Vignette du fichier
saha2020improved.pdf (530.58 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02950106 , version 1 (27-09-2020)

Identifiants

  • HAL Id : hal-02950106 , version 1

Citer

Aadirupa Saha, Pierre Gaillard, Michal Valko. Improved sleeping bandits with stochastic action sets and adversarial rewards. ICML 2020 - International Conference on Machine Learning, Jul 2020, Vienna, Austria. ⟨hal-02950106⟩
1007 Consultations
206 Téléchargements

Partager

Gmail Facebook X LinkedIn More