English

Robust-by-Design Classification via Unitary-Gradient Neural Networks

Machine Learning 2022-09-12 v1 Artificial Intelligence Differential Geometry Metric Geometry

Abstract

The use of neural networks in safety-critical systems requires safe and robust models, due to the existence of adversarial attacks. Knowing the minimal adversarial perturbation of any input x, or, equivalently, knowing the distance of x from the classification boundary, allows evaluating the classification robustness, providing certifiable predictions. Unfortunately, state-of-the-art techniques for computing such a distance are computationally expensive and hence not suited for online applications. This work proposes a novel family of classifiers, namely Signed Distance Classifiers (SDCs), that, from a theoretical perspective, directly output the exact distance of x from the classification boundary, rather than a probability score (e.g., SoftMax). SDCs represent a family of robust-by-design classifiers. To practically address the theoretical requirements of a SDC, a novel network architecture named Unitary-Gradient Neural Network is presented. Experimental results show that the proposed architecture approximates a signed distance classifier, hence allowing an online certifiable classification of x at the cost of a single inference.

Keywords

Cite

@article{arxiv.2209.04293,
  title  = {Robust-by-Design Classification via Unitary-Gradient Neural Networks},
  author = {Fabio Brau and Giulio Rossolini and Alessandro Biondi and Giorgio Buttazzo},
  journal= {arXiv preprint arXiv:2209.04293},
  year   = {2022}
}

Comments

Under review

R2 v1 2026-06-28T01:00:51.025Z