System Identification for Linear Dynamics with Bilinear Observation Models: An Expectation–Maximization Approach

Diyou Liu; Mohammad Khosravi

doi:10.1109/CDC56724.2024.10886575

System Identification for Linear Dynamics with Bilinear Observation Models: An Expectation–Maximization Approach

Team Khosravi

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

Abstract

In this paper, we study the system identification problem for linear time-invariant dynamics with bilinear observation models. Accordingly, we consider a suitable parametric description for the system model and formulate the identification problem as estimating the parameters characterizing the mathematical representation of the system through input-output measurement data. To this end, we employ a probabilistic framework aiming to obtain the maximum likelihood estimates of the parameters. Accordingly, we propose utilizing the expectation-maximization approach to improve the tractability of the identification procedure. Through the numerical experiments, we verify the efficacy of the proposed scheme and demonstrate its performance.

Original language	English
Title of host publication	Proceedings of the IEEE 63rd Conference on Decision and Control, CDC 2024
Publisher	IEEE
Pages	7190-7195
Number of pages	6
ISBN (Electronic)	979-8-3503-1633-9
DOIs	https://doi.org/10.1109/CDC56724.2024.10886575
Publication status	Published - 2025
Event	63rd IEEE Conference on Decision and Control, CDC 2024 - Milan, Italy Duration: 16 Dec 2024 → 19 Dec 2024

Publication series

Name	Proceedings of the IEEE Conference on Decision and Control
ISSN (Print)	0743-1546
ISSN (Electronic)	2576-2370

Conference

Conference	63rd IEEE Conference on Decision and Control, CDC 2024
Country/Territory	Italy
City	Milan
Period	16/12/24 → 19/12/24

Bibliographical note

Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care
Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Keywords

Maximum likelihood estimation
Monte Carlo methods
Probabilistic logic
Mathematical models
Data models
System identification
Numerical models
Dynamical systems

Access to Document

10.1109/CDC56724.2024.10886575

Cite this

@inproceedings{526f5bcb9fe041ee9334ac813242b8c1,

title = "System Identification for Linear Dynamics with Bilinear Observation Models: An Expectation–Maximization Approach",

abstract = "In this paper, we study the system identification problem for linear time-invariant dynamics with bilinear observation models. Accordingly, we consider a suitable parametric description for the system model and formulate the identification problem as estimating the parameters characterizing the mathematical representation of the system through input-output measurement data. To this end, we employ a probabilistic framework aiming to obtain the maximum likelihood estimates of the parameters. Accordingly, we propose utilizing the expectation-maximization approach to improve the tractability of the identification procedure. Through the numerical experiments, we verify the efficacy of the proposed scheme and demonstrate its performance.",

keywords = "Maximum likelihood estimation, Monte Carlo methods, Probabilistic logic, Mathematical models, Data models, System identification, Numerical models, Dynamical systems",

author = "Diyou Liu and Mohammad Khosravi",

note = "Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.; 63rd IEEE Conference on Decision and Control, CDC 2024 ; Conference date: 16-12-2024 Through 19-12-2024",

year = "2025",

doi = "10.1109/CDC56724.2024.10886575",

language = "English",

series = "Proceedings of the IEEE Conference on Decision and Control",

publisher = "IEEE",

pages = "7190--7195",

booktitle = "Proceedings of the IEEE 63rd Conference on Decision and Control, CDC 2024",

address = "United States",

}

Liu, D & Khosravi, M 2025, System Identification for Linear Dynamics with Bilinear Observation Models: An Expectation–Maximization Approach. in Proceedings of the IEEE 63rd Conference on Decision and Control, CDC 2024. Proceedings of the IEEE Conference on Decision and Control, IEEE, pp. 7190-7195, 63rd IEEE Conference on Decision and Control, CDC 2024, Milan, Italy, 16/12/24. https://doi.org/10.1109/CDC56724.2024.10886575

System Identification for Linear Dynamics with Bilinear Observation Models: An Expectation–Maximization Approach. / Liu, Diyou ; Khosravi, Mohammad.
Proceedings of the IEEE 63rd Conference on Decision and Control, CDC 2024. IEEE, 2025. p. 7190-7195 (Proceedings of the IEEE Conference on Decision and Control).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - System Identification for Linear Dynamics with Bilinear Observation Models: An Expectation–Maximization Approach

AU - Liu, Diyou

AU - Khosravi, Mohammad

N1 - Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

PY - 2025

Y1 - 2025

N2 - In this paper, we study the system identification problem for linear time-invariant dynamics with bilinear observation models. Accordingly, we consider a suitable parametric description for the system model and formulate the identification problem as estimating the parameters characterizing the mathematical representation of the system through input-output measurement data. To this end, we employ a probabilistic framework aiming to obtain the maximum likelihood estimates of the parameters. Accordingly, we propose utilizing the expectation-maximization approach to improve the tractability of the identification procedure. Through the numerical experiments, we verify the efficacy of the proposed scheme and demonstrate its performance.

AB - In this paper, we study the system identification problem for linear time-invariant dynamics with bilinear observation models. Accordingly, we consider a suitable parametric description for the system model and formulate the identification problem as estimating the parameters characterizing the mathematical representation of the system through input-output measurement data. To this end, we employ a probabilistic framework aiming to obtain the maximum likelihood estimates of the parameters. Accordingly, we propose utilizing the expectation-maximization approach to improve the tractability of the identification procedure. Through the numerical experiments, we verify the efficacy of the proposed scheme and demonstrate its performance.

KW - Maximum likelihood estimation

KW - Monte Carlo methods

KW - Probabilistic logic

KW - Mathematical models

KW - Data models

KW - System identification

KW - Numerical models

KW - Dynamical systems

UR - http://www.scopus.com/inward/record.url?scp=86000544220&partnerID=8YFLogxK

U2 - 10.1109/CDC56724.2024.10886575

DO - 10.1109/CDC56724.2024.10886575

M3 - Conference contribution

T3 - Proceedings of the IEEE Conference on Decision and Control

SP - 7190

EP - 7195

BT - Proceedings of the IEEE 63rd Conference on Decision and Control, CDC 2024

PB - IEEE

T2 - 63rd IEEE Conference on Decision and Control, CDC 2024

Y2 - 16 December 2024 through 19 December 2024

ER -

System Identification for Linear Dynamics with Bilinear Observation Models: An Expectation–Maximization Approach

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Embargoed Document

Fingerprint

Cite this