Skip to Main content Skip to Navigation
Preprints, Working Papers, ...

Diversity-Preserving K-Armed Bandits, Revisited

Abstract : We consider the bandit-based framework for diversity-preserving recommendations introduced by Celis et al. (2019), who approached it mainly by a reduction to the setting of linear bandits. We design a UCB algorithm using the specific structure of the setting and show that it enjoys a bounded distribution-dependent regret in the natural cases when the optimal mixed actions put some probability mass on all actions (i.e., when diversity is desirable). Simulations illustrate this fact. We also provide regret lower bounds and briefly discuss distribution-free regret bounds.
Document type :
Preprints, Working Papers, ...
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-02957485
Contributor : Gilles Stoltz <>
Submitted on : Monday, October 5, 2020 - 11:16:40 AM
Last modification on : Wednesday, November 18, 2020 - 3:18:02 PM

Files

FairBandits-ALT20.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02957485, version 1
  • ARXIV : 2010.01874

Citation

Hédi Hadiji, Sébastien Gerchinovitz, Jean-Michel Loubes, Gilles Stoltz. Diversity-Preserving K-Armed Bandits, Revisited. 2020. ⟨hal-02957485⟩

Share

Metrics

Record views

141

Files downloads

49