Logo du site
  • English
  • Français
  • Se connecter
Logo du site
  • English
  • Français
  • Se connecter
  1. Accueil
  2. Université de Neuchâtel
  3. Publications
  4. Achieving Privacy in the Adversarial Multi-Armed Bandit
 
  • Details
Options
Vignette d'image

Achieving Privacy in the Adversarial Multi-Armed Bandit

Auteur(s)
Aristide C. Y. Tossou
Dimitrakakis, Christos 
Institut d'informatique 
Date de parution
2017
In
Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence
De la page
2653
A la page
2659
Mots-clés
  • Machine Learning (cs.LG)
  • Artificial Intelligence (cs.AI)
  • Cryptography and Security (cs.CR)
  • Machine Learning (cs....

  • Artificial Intelligen...

  • Cryptography and Secu...

Résumé
In this paper, we improve the previously best known regret bound to achieve ϵ-differential privacy in oblivious adversarial bandits from O(T2/3/ϵ) to O(T−−√lnT/ϵ). This is achieved by combining a Laplace Mechanism with EXP3. We show that though EXP3 is already differentially private, it leaks a linear amount of information in T. However, we can improve this privacy by relying on its intrinsic exponential mechanism for selecting actions. This allows us to reach O(lnT−−−√)-DP, with a regret of O(T2/3) that holds against an adaptive adversary, an improvement from the best known of O(T3/4). This is done by using an algorithm that run EXP3 in a mini-batch loop. Finally, we run experiments that clearly demonstrate the validity of our theoretical analysis.
Nom de l'événement
Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence
Lieu
San Francisco, California, USA
Identifiants
https://libra.unine.ch/handle/123456789/30967
_
10.48550/arXiv.1701.04222
Type de publication
conference paper
Dossier(s) à télécharger
 main article: 1701.04222.pdf (608.53 KB)
google-scholar
Présentation du portailGuide d'utilisationStratégie Open AccessDirective Open Access La recherche à l'UniNE Open Access ORCIDNouveautés

Service information scientifique & bibliothèques
Rue Emile-Argand 11
2000 Neuchâtel
contact.libra@unine.ch

Propulsé par DSpace, DSpace-CRIS & 4Science | v2022.02.00