Neural Network Robustness as a Verification Property: A Principled Case Study

Marco Casadio*, Ekaterina Komendantskaya, Matthew L. Daggitt, Wen Kokke, Guy Katz, Guy Amir, Idan Refaeli

*Corresponding author for this work

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

16 Scopus citations

Abstract

Neural networks are very successful at detecting patterns in noisy data, and have become the technology of choice in many fields. However, their usefulness is hampered by their susceptibility to adversarial attacks. Recently, many methods for measuring and improving a network’s robustness to adversarial perturbations have been proposed, and this growing body of research has given rise to numerous explicit or implicit notions of robustness. Connections between these notions are often subtle, and a systematic comparison between them is missing in the literature. In this paper we begin addressing this gap, by setting up general principles for the empirical analysis and evaluation of a network’s robustness as a mathematical property—during the network’s training phase, its verification, and after its deployment. We then apply these principles and conduct a case study that showcases the practical benefits of our general approach.

Original languageEnglish
Title of host publicationComputer Aided Verification - 34th International Conference, CAV 2022, Proceedings
EditorsSharon Shoham, Yakir Vizel
PublisherSpringer Science and Business Media Deutschland GmbH
Pages219-231
Number of pages13
ISBN (Print)9783031131844
DOIs
StatePublished - 2022
Event34th International Conference on Computer Aided Verification, CAV 2022 - Haifa, Israel
Duration: 7 Aug 202210 Aug 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13371 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference34th International Conference on Computer Aided Verification, CAV 2022
Country/TerritoryIsrael
CityHaifa
Period7/08/2210/08/22

Bibliographical note

Publisher Copyright:
© 2022, The Author(s).

Keywords

  • Adversarial Training
  • Neural Networks
  • Robustness
  • Verification

Fingerprint

Dive into the research topics of 'Neural Network Robustness as a Verification Property: A Principled Case Study'. Together they form a unique fingerprint.

Cite this