Titre :

Algorithms for Differentially Private Multi-Armed Bandits

Auteur(s) :

Tossou, Aristide [Auteur]
Dimitrakakis, Christos [Auteur]
Sequential Learning [SEQUEL]
Université de Lille, Sciences Humaines et Sociales

Titre de la manifestation scientifique :

AAAI 2016

Ville :

Phoenix, Arizona

Pays :

Etats-Unis d'Amérique

Date de début de la manifestation scientifique :

2016-02-11

Mot(s)-clé(s) en anglais :

differential privacy
regret
reinforcement learning
stochastic multi-armed bandits

Discipline(s) HAL :

Statistiques [stat]/Machine Learning [stat.ML]
Informatique [cs]/Cryptographie et sécurité [cs.CR]

Résumé en anglais : [en]

We present differentially private algorithms for the stochastic Multi-Armed Bandit (MAB) problem. This is a problem for applications such as adaptive clinical trials, experiment design, and user-targeted advertising ...
Lire la suite >We present differentially private algorithms for the stochastic Multi-Armed Bandit (MAB) problem. This is a problem for applications such as adaptive clinical trials, experiment design, and user-targeted advertising where private information is connected to individual rewards. Our major contribution is to show that there exist $(\epsilon, \delta)$ differentially private variants of Upper Confidence Bound algorithms which have optimal regret, $O(\epsilon^{-1} + \log T)$. This is a significant improvement over previous results, which only achieve poly-log regret $O(\epsilon^{-2} \log^{2} T)$, because of our use of a novel interval-based mechanism. We also substantially improve the bounds of previous family of algorithms which use a continual release mechanism. Experiments clearly validate our theoretical bounds.Lire moins >

Langue :

Anglais

Comité de lecture :

Oui

Audience :

Internationale

Vulgarisation :

Non

Collections :

• Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189

Source :

Harvested from HAL