Noisy Learning for Neural ODEs Acts as a Robustness Locus Widening

Martin Gonzalez (1) , Hatem Hajri (1) , Loic Cantat (1) , Mihaly Petreczky (2)

1 IRT SystemX
2 Centrale Lille

We investigate the problems and challenges of evaluating the robustness of Differential Equationbased (DE) networks against synthetic distribution shifts. We propose a novel and simple accuracy metric which can be used to evaluate intrinsic robustness and to validate dataset corruption simulators. We also propose methodology recommendations, destined for evaluating the many faces of neural DEs' robustness and for comparing them with their discrete counterparts rigorously. We then use this criteria to evaluate a cheap data augmentation technique as a reliable way for demonstrating the natural robustness of neural ODEs against simulated image corruptions across multiple datasets.

Domaines: Apprentissage [cs.LG]

Martin Gonzalez, Hatem Hajri, Loic Cantat, Mihaly Petreczky. Noisy Learning for Neural ODEs Acts as a Robustness Locus Widening. Workshop on Principles of Distribution Shift (PODS-ICML), Jul 2022, Baltimore, United States. ⟨hal-03782513⟩