Improving normalizing flows with the approximate mass for out-of-distribution detection

Samy Chali (1) , Inna Kucher (1) , Marc Duranton (2) , Jacques-Olivier Klein (3)

1 LIAE - Laboratoire Intelligence Artificielle Embarquée
2 LIST (CEA) - Laboratoire d'Intégration des Systèmes et des Technologies
3 C2N - Centre de Nanosciences et de Nanotechnologies

Normalizing flows are generative models that show poor performance on out-of-distribution (OOD) detection tasks with a likelihood-based test. In this study we focus on the "approximate mass" metric. We show that while it improves OOD detection performance, it has limitations under a maximum likelihood training. To solve this limitation we modify the training objective by incorporating the approximate mass. It smooths the learnt distribution in the vicinity of training in-distribution data. We measure an average of 97.6% AUROC in our experiments on different benchmarks, showing an improvement of 16% with respect to the best baseline we tested against.

Domaines: Intelligence artificielle [cs.AI]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Réseau de neurones [cs.NE]

Samy Chali, Inna Kucher, Marc Duranton, Jacques-Olivier Klein. Improving normalizing flows with the approximate mass for out-of-distribution detection. CVPRW 2023 - 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, IEEE, Jun 2023, Vancouver, Canada. pp.750-758, ⟨10.1109/CVPRW59228.2023.00082⟩. ⟨cea-04191592⟩