Output Reachable Set Estimation and Verification for Multilayer Neural Networks

Weiming Xiang, Hoang Dung Tran, Taylor T. Johnson

Research output: Contribution to journalArticle

Abstract

In this brief, the output reachable estimation and safety verification problems for multilayer perceptron (MLP) neural networks are addressed. First, a conception called maximum sensitivity is introduced, and for a class of MLPs whose activation functions are monotonic functions, the maximum sensitivity can be computed via solving convex optimization problems. Then, using a simulation-based method, the output reachable set estimation problem for neural networks is formulated into a chain of optimization problems. Finally, an automated safety verification is developed based on the output reachable set estimation result. An application to the safety verification for a robotic arm model with two joints is presented to show the effectiveness of the proposed approaches.

Original languageEnglish (US)
Article number8318388
Pages (from-to)5777-5783
Number of pages7
JournalIEEE Transactions on Neural Networks and Learning Systems
Volume29
Issue number11
DOIs
StatePublished - Nov 2018
Externally publishedYes

Fingerprint

Multilayer neural networks
Neural networks
Robotic arms
Convex optimization
Chemical activation

Keywords

  • Multilayer perceptron (MLP)
  • reachable set estimation
  • simulation
  • verification

ASJC Scopus subject areas

  • Software
  • Computer Science Applications
  • Computer Networks and Communications
  • Artificial Intelligence

Cite this

Output Reachable Set Estimation and Verification for Multilayer Neural Networks. / Xiang, Weiming; Tran, Hoang Dung; Johnson, Taylor T.

In: IEEE Transactions on Neural Networks and Learning Systems, Vol. 29, No. 11, 8318388, 11.2018, p. 5777-5783.

Research output: Contribution to journalArticle

Xiang, Weiming ; Tran, Hoang Dung ; Johnson, Taylor T. / Output Reachable Set Estimation and Verification for Multilayer Neural Networks. In: IEEE Transactions on Neural Networks and Learning Systems. 2018 ; Vol. 29, No. 11. pp. 5777-5783.
@article{a208322f809d447a8d52533787ab9ef9,
title = "Output Reachable Set Estimation and Verification for Multilayer Neural Networks",
abstract = "In this brief, the output reachable estimation and safety verification problems for multilayer perceptron (MLP) neural networks are addressed. First, a conception called maximum sensitivity is introduced, and for a class of MLPs whose activation functions are monotonic functions, the maximum sensitivity can be computed via solving convex optimization problems. Then, using a simulation-based method, the output reachable set estimation problem for neural networks is formulated into a chain of optimization problems. Finally, an automated safety verification is developed based on the output reachable set estimation result. An application to the safety verification for a robotic arm model with two joints is presented to show the effectiveness of the proposed approaches.",
keywords = "Multilayer perceptron (MLP), reachable set estimation, simulation, verification",
author = "Weiming Xiang and Tran, {Hoang Dung} and Johnson, {Taylor T.}",
year = "2018",
month = "11",
doi = "10.1109/TNNLS.2018.2808470",
language = "English (US)",
volume = "29",
pages = "5777--5783",
journal = "IEEE Transactions on Neural Networks and Learning Systems",
issn = "2162-237X",
publisher = "IEEE Computational Intelligence Society",
number = "11",

}

TY - JOUR

T1 - Output Reachable Set Estimation and Verification for Multilayer Neural Networks

AU - Xiang, Weiming

AU - Tran, Hoang Dung

AU - Johnson, Taylor T.

PY - 2018/11

Y1 - 2018/11

N2 - In this brief, the output reachable estimation and safety verification problems for multilayer perceptron (MLP) neural networks are addressed. First, a conception called maximum sensitivity is introduced, and for a class of MLPs whose activation functions are monotonic functions, the maximum sensitivity can be computed via solving convex optimization problems. Then, using a simulation-based method, the output reachable set estimation problem for neural networks is formulated into a chain of optimization problems. Finally, an automated safety verification is developed based on the output reachable set estimation result. An application to the safety verification for a robotic arm model with two joints is presented to show the effectiveness of the proposed approaches.

AB - In this brief, the output reachable estimation and safety verification problems for multilayer perceptron (MLP) neural networks are addressed. First, a conception called maximum sensitivity is introduced, and for a class of MLPs whose activation functions are monotonic functions, the maximum sensitivity can be computed via solving convex optimization problems. Then, using a simulation-based method, the output reachable set estimation problem for neural networks is formulated into a chain of optimization problems. Finally, an automated safety verification is developed based on the output reachable set estimation result. An application to the safety verification for a robotic arm model with two joints is presented to show the effectiveness of the proposed approaches.

KW - Multilayer perceptron (MLP)

KW - reachable set estimation

KW - simulation

KW - verification

UR - http://www.scopus.com/inward/record.url?scp=85044071608&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85044071608&partnerID=8YFLogxK

U2 - 10.1109/TNNLS.2018.2808470

DO - 10.1109/TNNLS.2018.2808470

M3 - Article

C2 - 29993822

AN - SCOPUS:85044071608

VL - 29

SP - 5777

EP - 5783

JO - IEEE Transactions on Neural Networks and Learning Systems

JF - IEEE Transactions on Neural Networks and Learning Systems

SN - 2162-237X

IS - 11

M1 - 8318388

ER -