Deep learning and system identification

Lennart Ljung; Carl Andersson; Koen Tiels; Thomas B. Schön

doi:10.1016/j.ifacol.2020.12.1329

Deep learning and system identification

Lennart Ljung, Carl Andersson, Koen Tiels, Thomas B. Schön

Research output: Contribution to journal › Conference article › peer-review

105 Citations (Scopus)

324 Downloads (Pure)

Abstract

Deep learning is a topic of considerable interest today. Since it deals with estimating - or learning - models, there are connections to the area of System Identification developed in the Automatic Control community. Such connections are explored and exploited in this contribution. It is stressed that common deep nets such as feedforward and cascadeforward nets are nonlinear ARX (NARX) models, and can thus be easily incorporated in System Identification code and practice. The case of LSTM nets is an example of NonLinear State-Space (NLSS) models.

Original language	English
Pages (from-to)	1175-1181
Number of pages	7
Journal	IFAC-PapersOnLine
Volume	53
Issue number	2
DOIs	https://doi.org/10.1016/j.ifacol.2020.12.1329
Publication status	Published - Nov 2020
Event	21st World Congress of the International Federation of Aufomatic Control (IFAC 2020 World Congress) - Berlin, Germany Duration: 12 Jul 2020 → 17 Jul 2020 Conference number: 21 https://www.ifac2020.org/

Funding

This research was financially supported by the Swedish Foundation for Strategic Research (SSF) via the project ASSEMBLE (contract number: RIT15-0012) and by the Swedish Research Council via the projects Learning flexible models for nonlinear dynamics (contract number: 2017-03807) and NewLEADS - New Directions in Learning Dynamical Systems (contract number: 621-2016-06079). Ljung's work was supported by Vinnova's Competence Center LinkSic.

Funders	Funder number
Vinnova's Competence Center LinkSic
Stiftelsen för Strategisk Forskning	RIT15-0012
Vetenskapsrådet	2017-03807, 621-2016-06079

Keywords

Model structure
Bias/variance trade-off
Model validation
LSTM
Cascadeforwardnet
Deep nets
Bias/Variance Trade-off

Access to Document

10.1016/j.ifacol.2020.12.1329

IFAC20_1054_FIFinal author version , 971 KB

Cite this

@article{729a018a087b41b2b44b495a9a4bb7c2,

title = "Deep learning and system identification",

abstract = "Deep learning is a topic of considerable interest today. Since it deals with estimating - or learning - models, there are connections to the area of System Identification developed in the Automatic Control community. Such connections are explored and exploited in this contribution. It is stressed that common deep nets such as feedforward and cascadeforward nets are nonlinear ARX (NARX) models, and can thus be easily incorporated in System Identification code and practice. The case of LSTM nets is an example of NonLinear State-Space (NLSS) models.",

keywords = "Model structure, Bias/variance trade-off, Model validation, LSTM, Cascadeforwardnet, Deep nets, Bias/Variance Trade-off",

author = "Lennart Ljung and Carl Andersson and Koen Tiels and Sch{\"o}n, {Thomas B.}",

year = "2020",

month = nov,

doi = "10.1016/j.ifacol.2020.12.1329",

language = "English",

volume = "53",

pages = "1175--1181",

journal = "IFAC-PapersOnLine",

issn = "2405-8963",

publisher = "Elsevier",

number = "2",

note = "21st World Congress of the International Federation of Aufomatic Control (IFAC 2020 World Congress), IFAC 2020 ; Conference date: 12-07-2020 Through 17-07-2020",

url = "https://www.ifac2020.org/",

}

TY - JOUR

T1 - Deep learning and system identification

AU - Ljung, Lennart

AU - Andersson, Carl

AU - Tiels, Koen

AU - Schön, Thomas B.

N1 - Conference code: 21

PY - 2020/11

Y1 - 2020/11

N2 - Deep learning is a topic of considerable interest today. Since it deals with estimating - or learning - models, there are connections to the area of System Identification developed in the Automatic Control community. Such connections are explored and exploited in this contribution. It is stressed that common deep nets such as feedforward and cascadeforward nets are nonlinear ARX (NARX) models, and can thus be easily incorporated in System Identification code and practice. The case of LSTM nets is an example of NonLinear State-Space (NLSS) models.

AB - Deep learning is a topic of considerable interest today. Since it deals with estimating - or learning - models, there are connections to the area of System Identification developed in the Automatic Control community. Such connections are explored and exploited in this contribution. It is stressed that common deep nets such as feedforward and cascadeforward nets are nonlinear ARX (NARX) models, and can thus be easily incorporated in System Identification code and practice. The case of LSTM nets is an example of NonLinear State-Space (NLSS) models.

KW - Model structure

KW - Bias/variance trade-off

KW - Model validation

KW - LSTM

KW - Cascadeforwardnet

KW - Deep nets

KW - Bias/Variance Trade-off

UR - http://www.scopus.com/inward/record.url?scp=85094769243&partnerID=8YFLogxK

U2 - 10.1016/j.ifacol.2020.12.1329

DO - 10.1016/j.ifacol.2020.12.1329

M3 - Conference article

SN - 2405-8963

VL - 53

SP - 1175

EP - 1181

JO - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

IS - 2

T2 - 21st World Congress of the International Federation of Aufomatic Control (IFAC 2020 World Congress)

Y2 - 12 July 2020 through 17 July 2020

ER -

Deep learning and system identification

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this