Abstract

Models play an essential role in the design process of cyber-physical systems. They form the basis for simulation and analysis and help in identifying design problems as early as possible. However, the construction of models that comprise physical and digital behavior is challenging. Therefore, there is considerable interest in learning such hybrid behavior by means of machine learning which requires sufficient and representative training data covering the behavior of the physical system adequately. In this work, we exploit a combination of automata learning and model-based testing to generate sufficient training data fully automatically.

Experimental results on a platooning scenario show that recurrent neural networks learned with this data achieved significantly better results compared to models learned from randomly generated data. In particular, the classification error for crash detection is reduced by a factor of five and a similar F1-score is obtained with up to three orders of magnitude fewer training samples.

Original languageEnglish
Title of host publicationTesting Software and Systems
Subtitle of host publication31st IFIP WG 6.1 International Conference, ICTSS 2019, Paris, France, October 15–17, 2019, Proceedings
EditorsChristophe Gaston, Nikolai Kosmatov, Pascale Le Gall
Place of PublicationCham
PublisherSpringer International Publishing AG
Pages3-21
Number of pages19
ISBN (Print)978-3-030-31280-0
DOIs
Publication statusPublished - 2019
EventIFIP-ICTSS 2019: 31st IFIP International Conference on Testing Software and Systems - Paris, France
Duration: 15 Oct 201917 Oct 2019

Publication series

NameLecture Notes in Computer Science
Volume11812

Conference

ConferenceIFIP-ICTSS 2019
CountryFrance
CityParis
Period15/10/1917/10/19

Fingerprint

Hybrid systems
Learning systems
Testing
Recurrent neural networks

Keywords

  • Hybrid systems
  • Behavior modeling
  • Automata learning
  • Model-Based Testing
  • Machine learning
  • Autonomous vehicle
  • Platooning

Cite this

Aichernig, B. K., Bloem, R., Ebrahimi, M., Horn, M., Pernkopf, F., Roth, W., ... Tranninger, M. (2019). Learning a Behavior Model of Hybrid Systems Through Combining Model-Based Testing and Machine Learning. In C. Gaston, N. Kosmatov, & P. Le Gall (Eds.), Testing Software and Systems: 31st IFIP WG 6.1 International Conference, ICTSS 2019, Paris, France, October 15–17, 2019, Proceedings (pp. 3-21). (Lecture Notes in Computer Science; Vol. 11812). Cham: Springer International Publishing AG . https://doi.org/10.1007/978-3-030-31280-0_1

Learning a Behavior Model of Hybrid Systems Through Combining Model-Based Testing and Machine Learning. / Aichernig, Bernhard K.; Bloem, Roderick; Ebrahimi, Masoud; Horn, Martin; Pernkopf, Franz; Roth, Wolfgang; Rupp, Astrid; Tappler, Martin; Tranninger, Markus.

Testing Software and Systems: 31st IFIP WG 6.1 International Conference, ICTSS 2019, Paris, France, October 15–17, 2019, Proceedings. ed. / Christophe Gaston; Nikolai Kosmatov; Pascale Le Gall. Cham : Springer International Publishing AG , 2019. p. 3-21 (Lecture Notes in Computer Science; Vol. 11812).

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

Aichernig, BK, Bloem, R, Ebrahimi, M, Horn, M, Pernkopf, F, Roth, W, Rupp, A, Tappler, M & Tranninger, M 2019, Learning a Behavior Model of Hybrid Systems Through Combining Model-Based Testing and Machine Learning. in C Gaston, N Kosmatov & P Le Gall (eds), Testing Software and Systems: 31st IFIP WG 6.1 International Conference, ICTSS 2019, Paris, France, October 15–17, 2019, Proceedings. Lecture Notes in Computer Science, vol. 11812, Springer International Publishing AG , Cham, pp. 3-21, IFIP-ICTSS 2019, Paris, France, 15/10/19. https://doi.org/10.1007/978-3-030-31280-0_1
Aichernig BK, Bloem R, Ebrahimi M, Horn M, Pernkopf F, Roth W et al. Learning a Behavior Model of Hybrid Systems Through Combining Model-Based Testing and Machine Learning. In Gaston C, Kosmatov N, Le Gall P, editors, Testing Software and Systems: 31st IFIP WG 6.1 International Conference, ICTSS 2019, Paris, France, October 15–17, 2019, Proceedings. Cham: Springer International Publishing AG . 2019. p. 3-21. (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-030-31280-0_1
Aichernig, Bernhard K. ; Bloem, Roderick ; Ebrahimi, Masoud ; Horn, Martin ; Pernkopf, Franz ; Roth, Wolfgang ; Rupp, Astrid ; Tappler, Martin ; Tranninger, Markus. / Learning a Behavior Model of Hybrid Systems Through Combining Model-Based Testing and Machine Learning. Testing Software and Systems: 31st IFIP WG 6.1 International Conference, ICTSS 2019, Paris, France, October 15–17, 2019, Proceedings. editor / Christophe Gaston ; Nikolai Kosmatov ; Pascale Le Gall. Cham : Springer International Publishing AG , 2019. pp. 3-21 (Lecture Notes in Computer Science).
@inproceedings{2ef9be82021242378dd1d1040e1c4065,
title = "Learning a Behavior Model of Hybrid Systems Through Combining Model-Based Testing and Machine Learning",
abstract = "Models play an essential role in the design process of cyber-physical systems. They form the basis for simulation and analysis and help in identifying design problems as early as possible. However, the construction of models that comprise physical and digital behavior is challenging. Therefore, there is considerable interest in learning such hybrid behavior by means of machine learning which requires sufficient and representative training data covering the behavior of the physical system adequately. In this work, we exploit a combination of automata learning and model-based testing to generate sufficient training data fully automatically.Experimental results on a platooning scenario show that recurrent neural networks learned with this data achieved significantly better results compared to models learned from randomly generated data. In particular, the classification error for crash detection is reduced by a factor of five and a similar F1-score is obtained with up to three orders of magnitude fewer training samples.",
keywords = "Hybrid systems, Behavior modeling, Automata learning, Model-Based Testing, Machine learning, Autonomous vehicle, Platooning",
author = "Aichernig, {Bernhard K.} and Roderick Bloem and Masoud Ebrahimi and Martin Horn and Franz Pernkopf and Wolfgang Roth and Astrid Rupp and Martin Tappler and Markus Tranninger",
year = "2019",
doi = "10.1007/978-3-030-31280-0_1",
language = "English",
isbn = "978-3-030-31280-0",
series = "Lecture Notes in Computer Science",
publisher = "Springer International Publishing AG",
pages = "3--21",
editor = "Christophe Gaston and Nikolai Kosmatov and {Le Gall}, Pascale",
booktitle = "Testing Software and Systems",
address = "Switzerland",

}

TY - GEN

T1 - Learning a Behavior Model of Hybrid Systems Through Combining Model-Based Testing and Machine Learning

AU - Aichernig, Bernhard K.

AU - Bloem, Roderick

AU - Ebrahimi, Masoud

AU - Horn, Martin

AU - Pernkopf, Franz

AU - Roth, Wolfgang

AU - Rupp, Astrid

AU - Tappler, Martin

AU - Tranninger, Markus

PY - 2019

Y1 - 2019

N2 - Models play an essential role in the design process of cyber-physical systems. They form the basis for simulation and analysis and help in identifying design problems as early as possible. However, the construction of models that comprise physical and digital behavior is challenging. Therefore, there is considerable interest in learning such hybrid behavior by means of machine learning which requires sufficient and representative training data covering the behavior of the physical system adequately. In this work, we exploit a combination of automata learning and model-based testing to generate sufficient training data fully automatically.Experimental results on a platooning scenario show that recurrent neural networks learned with this data achieved significantly better results compared to models learned from randomly generated data. In particular, the classification error for crash detection is reduced by a factor of five and a similar F1-score is obtained with up to three orders of magnitude fewer training samples.

AB - Models play an essential role in the design process of cyber-physical systems. They form the basis for simulation and analysis and help in identifying design problems as early as possible. However, the construction of models that comprise physical and digital behavior is challenging. Therefore, there is considerable interest in learning such hybrid behavior by means of machine learning which requires sufficient and representative training data covering the behavior of the physical system adequately. In this work, we exploit a combination of automata learning and model-based testing to generate sufficient training data fully automatically.Experimental results on a platooning scenario show that recurrent neural networks learned with this data achieved significantly better results compared to models learned from randomly generated data. In particular, the classification error for crash detection is reduced by a factor of five and a similar F1-score is obtained with up to three orders of magnitude fewer training samples.

KW - Hybrid systems

KW - Behavior modeling

KW - Automata learning

KW - Model-Based Testing

KW - Machine learning

KW - Autonomous vehicle

KW - Platooning

U2 - 10.1007/978-3-030-31280-0_1

DO - 10.1007/978-3-030-31280-0_1

M3 - Conference contribution

SN - 978-3-030-31280-0

T3 - Lecture Notes in Computer Science

SP - 3

EP - 21

BT - Testing Software and Systems

A2 - Gaston, Christophe

A2 - Kosmatov, Nikolai

A2 - Le Gall, Pascale

PB - Springer International Publishing AG

CY - Cham

ER -