Cargando…

Improved Geometric Path Enumeration for Verifying ReLU Neural Networks

Neural networks provide quick approximations to complex functions, and have been increasingly used in perception as well as control tasks. For use in mission-critical and safety-critical applications, however, it is important to be able to analyze what a neural network can and cannot do. For feed-fo...

Descripción completa

Detalles Bibliográficos
Autores principales: Bak, Stanley, Tran, Hoang-Dung, Hobbs, Kerianne, Johnson, Taylor T.
Formato: Online Artículo Texto
Lenguaje:English
Publicado: 2020
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7363204/
http://dx.doi.org/10.1007/978-3-030-53288-8_4
_version_ 1783559623075168256
author Bak, Stanley
Tran, Hoang-Dung
Hobbs, Kerianne
Johnson, Taylor T.
author_facet Bak, Stanley
Tran, Hoang-Dung
Hobbs, Kerianne
Johnson, Taylor T.
author_sort Bak, Stanley
collection PubMed
description Neural networks provide quick approximations to complex functions, and have been increasingly used in perception as well as control tasks. For use in mission-critical and safety-critical applications, however, it is important to be able to analyze what a neural network can and cannot do. For feed-forward neural networks with ReLU activation functions, although exact analysis is NP-complete, recently-proposed verification methods can sometimes succeed. The main practical problem with neural network verification is excessive analysis runtime. Even on small networks, tools that are theoretically complete can sometimes run for days without producing a result. In this paper, we work to address the runtime problem by improving upon a recently-proposed geometric path enumeration method. Through a series of optimizations, several of which are new algorithmic improvements, we demonstrate significant speed improvement of exact analysis on the well-studied ACAS Xu benchmarks, sometimes hundreds of times faster than the original implementation. On more difficult benchmark instances, our optimized approach is often the fastest, even outperforming inexact methods that leverage overapproximation and refinement.
format Online
Article
Text
id pubmed-7363204
institution National Center for Biotechnology Information
language English
publishDate 2020
record_format MEDLINE/PubMed
spelling pubmed-73632042020-07-16 Improved Geometric Path Enumeration for Verifying ReLU Neural Networks Bak, Stanley Tran, Hoang-Dung Hobbs, Kerianne Johnson, Taylor T. Computer Aided Verification Article Neural networks provide quick approximations to complex functions, and have been increasingly used in perception as well as control tasks. For use in mission-critical and safety-critical applications, however, it is important to be able to analyze what a neural network can and cannot do. For feed-forward neural networks with ReLU activation functions, although exact analysis is NP-complete, recently-proposed verification methods can sometimes succeed. The main practical problem with neural network verification is excessive analysis runtime. Even on small networks, tools that are theoretically complete can sometimes run for days without producing a result. In this paper, we work to address the runtime problem by improving upon a recently-proposed geometric path enumeration method. Through a series of optimizations, several of which are new algorithmic improvements, we demonstrate significant speed improvement of exact analysis on the well-studied ACAS Xu benchmarks, sometimes hundreds of times faster than the original implementation. On more difficult benchmark instances, our optimized approach is often the fastest, even outperforming inexact methods that leverage overapproximation and refinement. 2020-06-13 /pmc/articles/PMC7363204/ http://dx.doi.org/10.1007/978-3-030-53288-8_4 Text en © The Author(s) 2020 Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
spellingShingle Article
Bak, Stanley
Tran, Hoang-Dung
Hobbs, Kerianne
Johnson, Taylor T.
Improved Geometric Path Enumeration for Verifying ReLU Neural Networks
title Improved Geometric Path Enumeration for Verifying ReLU Neural Networks
title_full Improved Geometric Path Enumeration for Verifying ReLU Neural Networks
title_fullStr Improved Geometric Path Enumeration for Verifying ReLU Neural Networks
title_full_unstemmed Improved Geometric Path Enumeration for Verifying ReLU Neural Networks
title_short Improved Geometric Path Enumeration for Verifying ReLU Neural Networks
title_sort improved geometric path enumeration for verifying relu neural networks
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7363204/
http://dx.doi.org/10.1007/978-3-030-53288-8_4
work_keys_str_mv AT bakstanley improvedgeometricpathenumerationforverifyingreluneuralnetworks
AT tranhoangdung improvedgeometricpathenumerationforverifyingreluneuralnetworks
AT hobbskerianne improvedgeometricpathenumerationforverifyingreluneuralnetworks
AT johnsontaylort improvedgeometricpathenumerationforverifyingreluneuralnetworks