Certified geometric robustness -- Super-DeepG

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cohen, Noémie, Ducoffe, Mélanie, Gabreau, Christophe, Pagetti, Claire, Pucel, Xavier
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915960892424192
author Cohen, Noémie
Ducoffe, Mélanie
Gabreau, Christophe
Pagetti, Claire
Pucel, Xavier
author_facet Cohen, Noémie
Ducoffe, Mélanie
Gabreau, Christophe
Pagetti, Claire
Pucel, Xavier
contents Safety-critical applications are required to perform as expected in normal operations. Image processing functions are often required to be insensitive to small geometric perturbations such as rotation, scaling, shearing or translation. This paper addresses the formal verification of neural networks against geometric perturbations on their image dataset. Our method Super-DeepG improves the reasoning used in linear relaxation techniques and Lipschitz optimization, and provides an implementation that leverages GPU hardware. By doing so, Super-DeepG achieves both precision and computational efficiency of robustness certification, to an extent that outperforms prior work. Super-DeepG is shared as an open-source tool on GitHub.
format Preprint
id arxiv_https___arxiv_org_abs_2604_24379
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Certified geometric robustness -- Super-DeepG
Cohen, Noémie
Ducoffe, Mélanie
Gabreau, Christophe
Pagetti, Claire
Pucel, Xavier
Artificial Intelligence
Machine Learning
Symbolic Computation
Safety-critical applications are required to perform as expected in normal operations. Image processing functions are often required to be insensitive to small geometric perturbations such as rotation, scaling, shearing or translation. This paper addresses the formal verification of neural networks against geometric perturbations on their image dataset. Our method Super-DeepG improves the reasoning used in linear relaxation techniques and Lipschitz optimization, and provides an implementation that leverages GPU hardware. By doing so, Super-DeepG achieves both precision and computational efficiency of robustness certification, to an extent that outperforms prior work. Super-DeepG is shared as an open-source tool on GitHub.
title Certified geometric robustness -- Super-DeepG
topic Artificial Intelligence
Machine Learning
Symbolic Computation
url https://arxiv.org/abs/2604.24379