English

Efficient Certification of Spatial Robustness

Machine Learning 2021-02-02 v2 Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning

Abstract

Recent work has exposed the vulnerability of computer vision models to vector field attacks. Due to the widespread usage of such models in safety-critical applications, it is crucial to quantify their robustness against such spatial transformations. However, existing work only provides empirical robustness quantification against vector field deformations via adversarial attacks, which lack provable guarantees. In this work, we propose novel convex relaxations, enabling us, for the first time, to provide a certificate of robustness against vector field transformations. Our relaxations are model-agnostic and can be leveraged by a wide range of neural network verifiers. Experiments on various network architectures and different datasets demonstrate the effectiveness and scalability of our method.

Keywords

Cite

@article{arxiv.2009.09318,
  title  = {Efficient Certification of Spatial Robustness},
  author = {Anian Ruoss and Maximilian Baader and Mislav Balunović and Martin Vechev},
  journal= {arXiv preprint arXiv:2009.09318},
  year   = {2021}
}

Comments

Conference Paper at AAAI 2021

R2 v1 2026-06-23T18:39:55.510Z