A Zonotopic Dempster-Shafer Approach to the Quantitative Verification of Neural Networks

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The reliability and usefulness of verification depend on the ability to represent appropriately the uncertainty. Most existing work on neural network verification relies on the hypothesis of either set-based or probabilistic information on the inputs. In this work, we rely on the framework of imprecise probabilities, specifically p-boxes, to propose a quantitative verification of ReLU neural networks, which can account for both probabilistic information and epistemic uncertainty on inputs. On classical benchmarks, including the ACAS Xu examples, we demonstrate that our approach improves the tradeoff between tightness and efficiency compared to related work on probabilistic network verification, while handling much more general classes of uncertainties on the inputs and providing fully guaranteed results.

Original languageEnglish
Title of host publicationFormal Methods - 26th International Symposium, FM 2024, Proceedings
EditorsAndré Platzer, Kristin Yvonne Rozier, Matteo Pradella, Matteo Rossi
PublisherSpringer Science and Business Media Deutschland GmbH
Pages324-342
Number of pages19
ISBN (Print)9783031711619
DOIs
Publication statusPublished - 1 Jan 2025
Event26th International Symposium on Formal Methods, FM 2024 - Milan, Italy
Duration: 9 Sept 202413 Sept 2024

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14933 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference26th International Symposium on Formal Methods, FM 2024
Country/TerritoryItaly
CityMilan
Period9/09/2413/09/24

Keywords

  • Dempster-Shafer structures
  • neural networks
  • p-boxes
  • probability bounds
  • zonotopes

Fingerprint

Dive into the research topics of 'A Zonotopic Dempster-Shafer Approach to the Quantitative Verification of Neural Networks'. Together they form a unique fingerprint.

Cite this