Guerin Joris auteur aut IRD Delmas K. auteur aut IRD Ferreira R. auteur aut IRD Guiochet J. auteur aut IRD text bookSection eng text/pdf born digital access The usage of deep neural networks in safety-critical systems is limited by our ability to guarantee their correct behavior. Runtime monitors are components aiming to identify unsafe predictions and discard them before they can lead to catastrophic consequences. Several recent works on runtime monitoring have focused on out-of-distribution (OOD) detection, i.e., identifying inputs that are different from the training data. In this work, we argue that OOD detection is not a well-suited framework to design efficient runtime monitors and that it is more relevant to evaluate monitors based on their ability to discard incorrect predictions. We call this setting out-of-model-scope detection and discuss the conceptual differences with OOD. We also conduct extensive experiments on popular datasets from the literature to show that studying monitors in the OOD setting can be misleading: 1. very good OOD results can give a false impression of safety, 2. comparison under the OOD setting does not allow identifying the best monitor to detect errors. Finally, we also show that removing erroneous training data samples helps to train better monitors. specialized 122 Williams B. ed. edt IRD Chen Y. ed. edt IRD Neville J. ed. edt IRD 14829-14837 Washington AAAI Press 2023 AAAI Conference on Artificial Intelligence ; Conference on Innovative Applications of Artificial Intelligence ; Symposium on Educational Advances in Artificial Intelligence, 37. ; 35. ; 13., Washington (USA), 2023/02/07-14 https://www.documentation.ird.fr/hor/fdi:010090489 10.1609/aaai.v37i12.26732 978-1-57735-880-0 [F B010090489] https://www.documentation.ird.fr/hor/fdi:010090489 https://www.documentation.ird.fr/intranet/publi/2023-11/010090489.pdf Accès réservé (Intranet de l'IRD) IRD - Base Horizon / Pleins textes 2023-11-03 2025-02-24 fdi:010090489 fre