A Posteriori Lipschitz Bounds for Deep Neural Networks
SIAM Journal on Numerical Analysis · 30.06.2026 · 64 (3) · 779-803
DOI: 10.1137/25m1742503 ↗Kurz gefasst
Fachartikel in der Zeitschrift SIAM Journal on Numerical Analysis, online erschienen am 02.06.2026. Verfasst von Moreno Pintore und 1 weiteren Autor:innen. Die Arbeit nennt 51 Literaturangaben. Gefördert von: PEPR/IA.
Abstract (Originalsprache)
Deriving sharp a posteriori upper bounds of the Lipschitz constant of deep neural networks is crucial to formally guarantee the robustness of neural network based models. We analyze three existing upper bounds written for the [Formula: see text] norm. We highlight the interest of working with the [Formula: see text] and [Formula: see text] norms and we propose two novel bounds for feed-forward fully connected neural networks, which can be extended to convolutional neural networks. Several numerical tests confirm the theoretical results, help to quantify the relationship between the presented bounds, and establish the better accuracy of the new bounds. Four numerical tests are studied: one with random matrices; two where the output has an analytical closed form; and one for convolutional neural networks trained on the MNIST dataset. The last numerical test is an application to a reference convolutional neural network. One of our bounds is optimal in the sense that it is exact for the second test uniformly with respect to the depth of the neural network and it is always better than the other ones.
Förderung
- PEPR/IA
Zitieren
APA
Pintore, M., & Després, B. (2026). A Posteriori Lipschitz Bounds for Deep Neural Networks. SIAM Journal on Numerical Analysis, 64(3), 779-803. https://doi.org/10.1137/25m1742503
Vancouver
Pintore M, Després B. A Posteriori Lipschitz Bounds for Deep Neural Networks. SIAM Journal on Numerical Analysis. 2026;64(3):779-803. doi:10.1137/25m1742503
BibTeX
@article{Pintore2026,
author = {Pintore, Moreno and Després, Bruno},
title = {A Posteriori Lipschitz Bounds for Deep Neural Networks},
journal = {SIAM Journal on Numerical Analysis},
year = {2026},
volume = {64},
number = {3},
pages = {779--803},
doi = {10.1137/25m1742503}
}
EUVIMED