Falsification of cyber-physical systems with reinforcement learning

Koki Kato, Fuyuki Ishikawa, Shinichi Honiden

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

We propose a novel framework for testing configurable cyber-physical systems over a given specification represented as metric temporal logic formula. Given a system model with configurable properties and a specification, our approach first learns to falsify the model by using reinforcement learning technique under a certain variety of configurations. After the training phase, it is expected that the experienced falsification agent can quickly find an input signal such that the output violates the specification, even though the specific configuration is not known to the agent. Thus we can use this agent again and again when different configurations are investigated for a product family or for trials and errors of configuration design. We performed a preliminary experiment to validate our hypothesis that the reinforcement learning technique can be applied for falsification problems.

Original languageEnglish
Title of host publicationProceedings - 2018 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages5-6
Number of pages2
ISBN (Print)9781538667484
DOIs
Publication statusPublished - 2018 Aug 7
Externally publishedYes
Event3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018 - Porto, Portugal
Duration: 2018 Apr 10 → …

Other

Other3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018
CountryPortugal
CityPorto
Period18/4/10 → …

Fingerprint

Reinforcement learning
Specifications
Temporal logic
Testing
Cyber Physical System
Experiments

Keywords

  • Cyber-physical-system
  • Falsification
  • Model-based-development
  • Reinforcement-learning

ASJC Scopus subject areas

  • Safety, Risk, Reliability and Quality
  • Artificial Intelligence

Cite this

Kato, K., Ishikawa, F., & Honiden, S. (2018). Falsification of cyber-physical systems with reinforcement learning. In Proceedings - 2018 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018 (pp. 5-6). [8429476] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/MT-CPS.2018.00009

Falsification of cyber-physical systems with reinforcement learning. / Kato, Koki; Ishikawa, Fuyuki; Honiden, Shinichi.

Proceedings - 2018 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 5-6 8429476.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kato, K, Ishikawa, F & Honiden, S 2018, Falsification of cyber-physical systems with reinforcement learning. in Proceedings - 2018 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018., 8429476, Institute of Electrical and Electronics Engineers Inc., pp. 5-6, 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018, Porto, Portugal, 18/4/10. https://doi.org/10.1109/MT-CPS.2018.00009
Kato K, Ishikawa F, Honiden S. Falsification of cyber-physical systems with reinforcement learning. In Proceedings - 2018 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 5-6. 8429476 https://doi.org/10.1109/MT-CPS.2018.00009
Kato, Koki ; Ishikawa, Fuyuki ; Honiden, Shinichi. / Falsification of cyber-physical systems with reinforcement learning. Proceedings - 2018 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 5-6
@inproceedings{04e93815c5a04bf49154080349071b7c,
title = "Falsification of cyber-physical systems with reinforcement learning",
abstract = "We propose a novel framework for testing configurable cyber-physical systems over a given specification represented as metric temporal logic formula. Given a system model with configurable properties and a specification, our approach first learns to falsify the model by using reinforcement learning technique under a certain variety of configurations. After the training phase, it is expected that the experienced falsification agent can quickly find an input signal such that the output violates the specification, even though the specific configuration is not known to the agent. Thus we can use this agent again and again when different configurations are investigated for a product family or for trials and errors of configuration design. We performed a preliminary experiment to validate our hypothesis that the reinforcement learning technique can be applied for falsification problems.",
keywords = "Cyber-physical-system, Falsification, Model-based-development, Reinforcement-learning",
author = "Koki Kato and Fuyuki Ishikawa and Shinichi Honiden",
year = "2018",
month = "8",
day = "7",
doi = "10.1109/MT-CPS.2018.00009",
language = "English",
isbn = "9781538667484",
pages = "5--6",
booktitle = "Proceedings - 2018 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Falsification of cyber-physical systems with reinforcement learning

AU - Kato, Koki

AU - Ishikawa, Fuyuki

AU - Honiden, Shinichi

PY - 2018/8/7

Y1 - 2018/8/7

N2 - We propose a novel framework for testing configurable cyber-physical systems over a given specification represented as metric temporal logic formula. Given a system model with configurable properties and a specification, our approach first learns to falsify the model by using reinforcement learning technique under a certain variety of configurations. After the training phase, it is expected that the experienced falsification agent can quickly find an input signal such that the output violates the specification, even though the specific configuration is not known to the agent. Thus we can use this agent again and again when different configurations are investigated for a product family or for trials and errors of configuration design. We performed a preliminary experiment to validate our hypothesis that the reinforcement learning technique can be applied for falsification problems.

AB - We propose a novel framework for testing configurable cyber-physical systems over a given specification represented as metric temporal logic formula. Given a system model with configurable properties and a specification, our approach first learns to falsify the model by using reinforcement learning technique under a certain variety of configurations. After the training phase, it is expected that the experienced falsification agent can quickly find an input signal such that the output violates the specification, even though the specific configuration is not known to the agent. Thus we can use this agent again and again when different configurations are investigated for a product family or for trials and errors of configuration design. We performed a preliminary experiment to validate our hypothesis that the reinforcement learning technique can be applied for falsification problems.

KW - Cyber-physical-system

KW - Falsification

KW - Model-based-development

KW - Reinforcement-learning

UR - http://www.scopus.com/inward/record.url?scp=85052490039&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85052490039&partnerID=8YFLogxK

U2 - 10.1109/MT-CPS.2018.00009

DO - 10.1109/MT-CPS.2018.00009

M3 - Conference contribution

AN - SCOPUS:85052490039

SN - 9781538667484

SP - 5

EP - 6

BT - Proceedings - 2018 3rd Workshop on Monitoring and Testing of Cyber-Physical Systems, MT-CPS 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -