Model-agnostic out-of-distribution detection using combined statistical tests - Laboratoire Jean-Alexandre Dieudonné Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Model-agnostic out-of-distribution detection using combined statistical tests

Résumé

We present simple methods for out-of-distribution detection using a trained generative model. These techniques, based on classical statistical tests, are model-agnostic in the sense that they can be applied to any differentiable generative model. The idea is to combine a classical parametric test (Rao's score test) with the recently introduced typicality test. These two test statistics are both theoretically well-founded and exploit different sources of information based on the likelihood for the typicality test and its gradient for the score test. We show that combining them using Fisher's method overall leads to a more accurate out-of-distribution test. We also discuss the benefits of casting out-of-distribution detection as a statistical testing problem, noting in particular that false positive rate control can be valuable for practical out-of-distribution detection. Despite their simplicity and generality, these methods can be competitive with model-specific out-of-distribution detection algorithms without any assumptions on the out-distribution.

Dates et versions

hal-03600783 , version 1 (07-03-2022)

Identifiants

Citer

Federico Bergamin, Pierre-Alexandre Mattei, Jakob Havtorn, Hugo Senetaire, Hugo Schmutz, et al.. Model-agnostic out-of-distribution detection using combined statistical tests. AISTATS 2022 - 25th International Conference on Artificial Intelligence and Statistics, Mar 2022, Valence, Spain. ⟨hal-03600783⟩
88 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More