Small Total-Cost Constraints in Contextual Bandits with Knapsacks, with Application to Fairness - Inria - Institut national de recherche en sciences et technologies du numérique
Conference Papers Year : 2023

Small Total-Cost Constraints in Contextual Bandits with Knapsacks, with Application to Fairness

Abstract

We consider contextual bandit problems with knapsacks [CBwK], a problem where at each round, a scalar reward is obtained and vector-valued costs are suffered. The learner aims to maximize the cumulative rewards while ensuring that the cumulative costs are lower than some predetermined cost constraints. We assume that contexts come from a continuous set, that costs can be signed, and that the expected reward and cost functions, while unknown, may be uniformly estimated---a typical assumption in the literature. In this setting, total cost constraints had so far to be at least of order $T^{3/4}$, where $T$ is the number of rounds, and were even typically assumed to depend linearly on $T$. We are however motivated to use CBwK to impose a fairness constraint of equalized average costs between groups: the budget associated with the corresponding cost constraints should be as close as possible to the natural deviations, of order $\sqrt{T}$. To that end, we introduce a dual strategy based on projected-gradient-descent updates, that is able to deal with total-cost constraints of the order of $\sqrt{T}$ up to poly-logarithmic terms. This strategy is more direct and simpler than existing strategies in the literature. It relies on a careful, adaptive, tuning of the step size.
Fichier principal
Vignette du fichier
Fair-CBwK--by-CGLS-v2-HAL.pdf (1.66 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04105622 , version 1 (24-05-2023)
hal-04105622 , version 2 (24-10-2023)

Identifiers

Cite

Evgenii Chzhen, Christophe Giraud, Zhen Li, Gilles Stoltz. Small Total-Cost Constraints in Contextual Bandits with Knapsacks, with Application to Fairness. Advances in Neural Information Processing Systems, Dec 2023, New Orleans, United States. ⟨hal-04105622v2⟩
211 View
85 Download

Altmetric

Share

More