The Marabou Framework for Verification and Analysis of Deep Neural Networks

Guy Katz; Derek A. Huang; Duligur Ibeling; Kyle Julian; Christopher Lazarus; Rachel Lim; Parth Shah; Shantanu Thakoor; Haoze Wu; Aleksandar Zeljić; David L. Dill; Mykel J. Kochenderfer; Clark Barrett

doi:https://doi.org/10.1007/978-3-030-25540-4_26

The Marabou Framework for Verification and Analysis of Deep Neural Networks

Guy Katz, Derek A. Huang, Duligur Ibeling, Kyle Julian, Christopher Lazarus, Rachel Lim, Parth Shah, Shantanu Thakoor, Haoze Wu, Aleksandar Zeljić, David L. Dill, Mykel J. Kochenderfer, Clark Barrett

The Rachel and Selim Benin School of Engineering and Computer Science

The Hebrew University of Jerusalem

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Deep neural networks are revolutionizing the way complex systems are designed. Consequently, there is a pressing need for tools and techniques for network analysis and certification. To help in addressing that need, we present Marabou, a framework for verifying deep neural networks. Marabou is an SMT-based tool that can answer queries about a network’s properties by transforming these queries into constraint satisfaction problems. It can accommodate networks with different activation functions and topologies, and it performs high-level reasoning on the network that can curtail the search space and improve performance. It also supports parallel execution to further enhance scalability. Marabou accepts multiple input formats, including protocol buffer files generated by the popular TensorFlow framework for neural networks. We describe the system architecture and main components, evaluate the technique and discuss ongoing work.

Original language	English
Title of host publication	Computer Aided Verification - 31st International Conference, CAV 2019, Proceedings
Editors	Isil Dillig, Serdar Tasiran
Publisher	Springer Verlag
Pages	443-452
Number of pages	10
ISBN (Print)	9783030255398
DOIs	https://doi.org/10.1007/978-3-030-25540-4_26
State	Published - 2019
Event	31st International Conference on Computer Aided Verification, CAV 2019 - New York City, United States Duration: 15 Jul 2019 → 18 Jul 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11561 LNCS

Conference

Conference	31st International Conference on Computer Aided Verification, CAV 2019
Country/Territory	United States
City	New York City
Period	15/07/19 → 18/07/19

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

Access to Document

https://doi.org/10.1007/978-3-030-25540-4_26

Cite this

Katz, G., Huang, D. A., Ibeling, D., Julian, K., Lazarus, C., Lim, R., Shah, P., Thakoor, S., Wu, H., Zeljić, A., Dill, D. L., Kochenderfer, M. J., & Barrett, C. (2019). The Marabou Framework for Verification and Analysis of Deep Neural Networks. In I. Dillig, & S. Tasiran (Eds.), Computer Aided Verification - 31st International Conference, CAV 2019, Proceedings (pp. 443-452). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11561 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-25540-4_26

Katz, Guy ; Huang, Derek A. ; Ibeling, Duligur et al. / The Marabou Framework for Verification and Analysis of Deep Neural Networks. Computer Aided Verification - 31st International Conference, CAV 2019, Proceedings. editor / Isil Dillig ; Serdar Tasiran. Springer Verlag, 2019. pp. 443-452 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{093616df9526440f9e6f652b4f55178a,

title = "The Marabou Framework for Verification and Analysis of Deep Neural Networks",

abstract = "Deep neural networks are revolutionizing the way complex systems are designed. Consequently, there is a pressing need for tools and techniques for network analysis and certification. To help in addressing that need, we present Marabou, a framework for verifying deep neural networks. Marabou is an SMT-based tool that can answer queries about a network{\textquoteright}s properties by transforming these queries into constraint satisfaction problems. It can accommodate networks with different activation functions and topologies, and it performs high-level reasoning on the network that can curtail the search space and improve performance. It also supports parallel execution to further enhance scalability. Marabou accepts multiple input formats, including protocol buffer files generated by the popular TensorFlow framework for neural networks. We describe the system architecture and main components, evaluate the technique and discuss ongoing work.",

author = "Guy Katz and Huang, {Derek A.} and Duligur Ibeling and Kyle Julian and Christopher Lazarus and Rachel Lim and Parth Shah and Shantanu Thakoor and Haoze Wu and Aleksandar Zelji{\'c} and Dill, {David L.} and Kochenderfer, {Mykel J.} and Clark Barrett",

note = "Publisher Copyright: {\textcopyright} The Author(s). 2019.; 31st International Conference on Computer Aided Verification, CAV 2019 ; Conference date: 15-07-2019 Through 18-07-2019",

year = "2019",

doi = "https://doi.org/10.1007/978-3-030-25540-4_26",

language = "الإنجليزيّة",

isbn = "9783030255398",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "443--452",

editor = "Isil Dillig and Serdar Tasiran",

booktitle = "Computer Aided Verification - 31st International Conference, CAV 2019, Proceedings",

address = "ألمانيا",

}

Katz, G, Huang, DA, Ibeling, D, Julian, K, Lazarus, C, Lim, R, Shah, P, Thakoor, S, Wu, H, Zeljić, A, Dill, DL, Kochenderfer, MJ & Barrett, C 2019, The Marabou Framework for Verification and Analysis of Deep Neural Networks. in I Dillig & S Tasiran (eds), Computer Aided Verification - 31st International Conference, CAV 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11561 LNCS, Springer Verlag, pp. 443-452, 31st International Conference on Computer Aided Verification, CAV 2019, New York City, United States, 15/07/19. https://doi.org/10.1007/978-3-030-25540-4_26

The Marabou Framework for Verification and Analysis of Deep Neural Networks. / Katz, Guy; Huang, Derek A.; Ibeling, Duligur et al.
Computer Aided Verification - 31st International Conference, CAV 2019, Proceedings. ed. / Isil Dillig; Serdar Tasiran. Springer Verlag, 2019. p. 443-452 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11561 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - The Marabou Framework for Verification and Analysis of Deep Neural Networks

AU - Katz, Guy

AU - Huang, Derek A.

AU - Ibeling, Duligur

AU - Julian, Kyle

AU - Lazarus, Christopher

AU - Lim, Rachel

AU - Shah, Parth

AU - Thakoor, Shantanu

AU - Wu, Haoze

AU - Zeljić, Aleksandar

AU - Dill, David L.

AU - Kochenderfer, Mykel J.

AU - Barrett, Clark

PY - 2019

Y1 - 2019

N2 - Deep neural networks are revolutionizing the way complex systems are designed. Consequently, there is a pressing need for tools and techniques for network analysis and certification. To help in addressing that need, we present Marabou, a framework for verifying deep neural networks. Marabou is an SMT-based tool that can answer queries about a network’s properties by transforming these queries into constraint satisfaction problems. It can accommodate networks with different activation functions and topologies, and it performs high-level reasoning on the network that can curtail the search space and improve performance. It also supports parallel execution to further enhance scalability. Marabou accepts multiple input formats, including protocol buffer files generated by the popular TensorFlow framework for neural networks. We describe the system architecture and main components, evaluate the technique and discuss ongoing work.

AB - Deep neural networks are revolutionizing the way complex systems are designed. Consequently, there is a pressing need for tools and techniques for network analysis and certification. To help in addressing that need, we present Marabou, a framework for verifying deep neural networks. Marabou is an SMT-based tool that can answer queries about a network’s properties by transforming these queries into constraint satisfaction problems. It can accommodate networks with different activation functions and topologies, and it performs high-level reasoning on the network that can curtail the search space and improve performance. It also supports parallel execution to further enhance scalability. Marabou accepts multiple input formats, including protocol buffer files generated by the popular TensorFlow framework for neural networks. We describe the system architecture and main components, evaluate the technique and discuss ongoing work.

UR - http://www.scopus.com/inward/record.url?scp=85069853231&partnerID=8YFLogxK

U2 - https://doi.org/10.1007/978-3-030-25540-4_26

DO - https://doi.org/10.1007/978-3-030-25540-4_26

M3 - منشور من مؤتمر

SN - 9783030255398

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 443

EP - 452

BT - Computer Aided Verification - 31st International Conference, CAV 2019, Proceedings

A2 - Dillig, Isil

A2 - Tasiran, Serdar

PB - Springer Verlag

T2 - 31st International Conference on Computer Aided Verification, CAV 2019

Y2 - 15 July 2019 through 18 July 2019

ER -

Katz G, Huang DA, Ibeling D, Julian K, Lazarus C, Lim R et al. The Marabou Framework for Verification and Analysis of Deep Neural Networks. In Dillig I, Tasiran S, editors, Computer Aided Verification - 31st International Conference, CAV 2019, Proceedings. Springer Verlag. 2019. p. 443-452. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: https://doi.org/10.1007/978-3-030-25540-4_26

The Marabou Framework for Verification and Analysis of Deep Neural Networks

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this