Top-m identification for linear bandits

Clémence Réda (1) , Emilie Kaufmann (2, 3, 4) , Andrée Delahaye-Duriez (5, 6, 1)

1 NeuroDiderot (UMR_S_1141 / U1141) - Maladies neurodéveloppementales et neurovasculaires
2 CNRS - Centre National de la Recherche Scientifique
3 Centrale Lille
4 Scool - Scool
5 Université Sorbonne Paris Nord
6 Hôpital Jean Verdier [AP-HP]

Motivated by an application to drug repurposing, we propose the first algorithms to tackle the identification of the m ≥ 1 arms with largest means in a linear bandit model, in the fixed-confidence setting. These algorithms belong to the generic family of Gap-Index Focused Algorithms (GIFA) that we introduce for Top-m identification in linear bandits. We propose a unified analysis of these algorithms, which shows how the use of features might decrease the sample complexity. We further validate these algorithms empirically on simulated data and on a simple drug repurposing task.

Mots clés: Multi-armed bandit; Best arm identification; Adaptive sampling; Drug repurposing; Fixed-confidence setting

Domaines: Intelligence artificielle [cs.AI]; Bio-informatique [q-bio.QM]; Apprentissage [cs.LG]; Statistiques [math.ST]

Clémence Réda, Emilie Kaufmann, Andrée Delahaye-Duriez. Top-m identification for linear bandits. Proceedings of the 24th International Conference on Artificial Intelligence and Statistics (AISTATS), 2021, Virtual, United States. ⟨hal-03172145⟩