Thesis of Adrienne Tuynman

Optimalities in sequential identification

The objective of this thesis is to develop new adaptive algorithms, and to prove theoretical guarantees on the necessary sample size to perform a high confidence test in a small sample size setting. As the limitation on the sample size is often due to the difficulty of gathering observations, which would prevent the algorithm from adapting at each time step, we want to know how much adaptivity is necessary to still see significant improvements when comparing to non adaptive methods. Finally, with clinical trials in mind, we'll study how to stop the test as soon as possible while guaranteeing that patients receive efficient treatment.

Jury

Composition du jury proposé Mme Emilie KAUFMANN Chargée de recherche Université de Lille Directrice de thèse, M. Bruno GAUJAL Directeur de recherche Université Grenoble Alpes Rapporteur, M. Gergely NEU Full professor Universitat Pompeu Fabra Rapporteur, Mme Claire VERNADE Full professor UTN Examinatrice, M. Alexandre PROUTIèRE Professeur EECS school, KTH Examinateur, M. Rémy DEGENNE Chargé de recherche Université de Lille Co-directeur de thèse, Mme Rianne DE HEIDE Associate Professor University of Twente Examinatrice.

Thesis of the team SCOOL defended on 08/10/2026