Bandits on graphs and structures

Valko, Michal

Type de document :

Habilitation à diriger des recherches

Titre :

Bandits on graphs and structures

Titre en anglais :

Bandits on graphs and structures

Auteur(s) :

Valko, Michal [Auteur]

Sequential Learning [SEQUEL]

Directeur(s) de thèse :

Aurélien Garivier

Date de soutenance :

2016-06-15

Président du jury :

Nicolas Vayatis (Garant & Examinateur)
Aurélien Garivier (Président & Rapporteur)
Gábor Lugosi (Rapporteur)
Vianney Perchet (Rapporteur)
Nicolò Cesa-Bianchi (Examinateur)
Mark Herbster (Examinateur)
Rémi Munos (Examinateur)

Membre(s) du jury :

Organisme de délivrance :

École normale supérieure de Cachan - ENS Cachan

Mot(s)-clé(s) :

apprentissage statistique

Mot(s)-clé(s) en anglais :

machine learning
sequential decision-making
bandits
graphs
structured learning

Discipline(s) HAL :

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

Résumé en anglais : [en]

We investigate the structural properties of certain sequential decision-making problems with limited feedback (bandits) in order to bring the known algorithmic solutions closer to a practical use. In the first part, we put ...
Lire la suite >We investigate the structural properties of certain sequential decision-making problems with limited feedback (bandits) in order to bring the known algorithmic solutions closer to a practical use. In the first part, we put a special emphasis on structures that can be represented as graphs on actions, in the second part we study the large action spaces that can be of exponential size in the number of base actions or even infinite. We show how to take advantage of structures over the actions and (provably) learn faster.Lire moins >

Langue :

Anglais

Collections :