Trading off rewards and errors in multi-armed bandits

Erraqabi, Akram; Lazaric, Alessandro; Valko, Michal; Brunskill, Emma; Liu, Yun-En

Type de document :

Communication dans un congrès avec actes

Titre :

Trading off rewards and errors in multi-armed bandits

Auteur(s) :

Erraqabi, Akram [Auteur]
Université de Montréal [UdeM]
Sequential Learning [SEQUEL]
Lazaric, Alessandro [Auteur]

Sequential Learning [SEQUEL]
Valko, Michal [Auteur]

Sequential Learning [SEQUEL]
Brunskill, Emma [Auteur]
Computer Science Department - Carnegie Mellon University
Liu, Yun-En [Auteur]
Computer Science Department - Carnegie Mellon University

Titre de la manifestation scientifique :

International Conference on Artificial Intelligence and Statistics

Ville :

Fort Lauderdale

Pays :

Etats-Unis d'Amérique

Date de début de la manifestation scientifique :

2017

Discipline(s) HAL :

Statistiques [stat]/Machine Learning [stat.ML]

Résumé en anglais : [en]

In multi-armed bandits, the most common objective is the maximization of the cumulative reward. Alternative settings include active exploration, where a learner tries to gain accurate estimates of the rewards of all arms. ...
Lire la suite >In multi-armed bandits, the most common objective is the maximization of the cumulative reward. Alternative settings include active exploration, where a learner tries to gain accurate estimates of the rewards of all arms. While these objectives are contrasting, in many scenarios it is desirable to trade off rewards and errors. For instance, in educational games the designer wants to gather generalizable knowledge about the behavior of the students and teaching strategies (small estimation errors) but, at the same time, the system needs to avoid giving a bad experience to the players, who may leave the system permanently (large reward). In this paper, we formalize this tradeoff and introduce the ForcingBalance algorithm whose performance is provably close to the best possible tradeoff strategy. Finally, we demonstrate on real-world educational data that ForcingBalance returns useful information about the arms without compromising the overall reward.Lire moins >

Langue :

Anglais

Comité de lecture :

Oui

Audience :

Internationale

Vulgarisation :

Non

Projet ANR :

Extraction et transfert de connaissances dans l'apprentissage par renforcement
Inférence bayésienne à ressources limitées - données massives et modèles coûteux

Collections :