Skip to main navigation Skip to search Skip to main content

nnenum: Verification of ReLU Neural Networks with Optimized Abstraction Refinement

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

51 Scopus citations

Abstract

The surge of interest in applications of deep neural networks has led to a surge of interest in verification methods for such architectures. In summer 2020, the first international competition on neural network verification was held. This paper presents and evaluates the main optimizations used in the nnenum tool, which outperformed all other tools in the ACAS Xu benchmark category, sometimes by orders of magnitude. The method uses fast abstractions for speed, combined with refinement through ReLU splitting to increase accuracy when properties cannot be proven. Although the abstraction refinement process is a classic approach in formal methods, directly applying it to the neural network verification problem actually reduces performance, due to a cascade of overapproxmation error when using abstraction. This makes optimizations and their systematic evaluation essential for high performance.

Original languageEnglish
Title of host publicationNASA Formal Methods - 13th International Symposium, NFM 2021, Proceedings
EditorsAaron Dutle, César A. Muñoz, Mariano M. Moscato, Laura Titolo, Ivan Perez
PublisherSpringer Science and Business Media Deutschland GmbH
Pages19-36
Number of pages18
ISBN (Print)9783030763831
DOIs
StatePublished - 2021
Event13th International Symposium on NASA Formal Methods, NFM 2021 - Virtual, Online
Duration: May 24 2021May 28 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12673 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference13th International Symposium on NASA Formal Methods, NFM 2021
CityVirtual, Online
Period05/24/2105/28/21

Keywords

  • ACAS Xu
  • Neural network verification
  • ReLU

Fingerprint

Dive into the research topics of 'nnenum: Verification of ReLU Neural Networks with Optimized Abstraction Refinement'. Together they form a unique fingerprint.

Cite this