Study of the effective number of parameters in nonlinear identification benchmarks

Anna Marconato; Maarten Schoukens; Yves Rolain; J.F.M. Schoukens

doi:10.1109/CDC.2013.6760552

Study of the effective number of parameters in nonlinear identification benchmarks

Anna Marconato, Maarten Schoukens, Yves Rolain, J.F.M. Schoukens

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

14 Citations (Scopus)

2 Downloads (Pure)

Abstract

This paper discusses the importance of the notion of effective number of parameters as a measure of model complexity. Exploiting this concept allows a fair comparison of models obtained from different model classes. Several illustrative examples of linear and nonlinear models are presented to provide more insight in the problem. As one possible way of showing that model complexity can be reduced without having to pull any parameters to zero, an approach for rank reduced estimation based on the truncated SVD is also discussed. These ideas are then applied to two nonlinear real world problems: The Wiener-Hammerstein and the Silverbox benchmarks.

Original language	English
Title of host publication	2013 IEEE 52nd Annual Conference on Decision and Control, CDC 2013
Publisher	Institute of Electrical and Electronics Engineers
Pages	4308-4313
Number of pages	6
ISBN (Print)	9781467357173
DOIs	https://doi.org/10.1109/CDC.2013.6760552
Publication status	Published - 1 Jan 2013
Externally published	Yes
Event	52nd IEEE Conference on Decision and Control (CDC 2013) - Florence, Italy Duration: 10 Dec 2013 → 13 Dec 2013 Conference number: 52

Conference

Conference	52nd IEEE Conference on Decision and Control (CDC 2013)
Abbreviated title	CDC 2013
Country/Territory	Italy
City	Florence
Period	10/12/13 → 13/12/13

Access to Document

10.1109/CDC.2013.6760552

Cite this

@inproceedings{b0f5bad9cd0a4b49b73fd787eb573ab9,

title = "Study of the effective number of parameters in nonlinear identification benchmarks",

abstract = "This paper discusses the importance of the notion of effective number of parameters as a measure of model complexity. Exploiting this concept allows a fair comparison of models obtained from different model classes. Several illustrative examples of linear and nonlinear models are presented to provide more insight in the problem. As one possible way of showing that model complexity can be reduced without having to pull any parameters to zero, an approach for rank reduced estimation based on the truncated SVD is also discussed. These ideas are then applied to two nonlinear real world problems: The Wiener-Hammerstein and the Silverbox benchmarks.",

author = "Anna Marconato and Maarten Schoukens and Yves Rolain and J.F.M. Schoukens",

year = "2013",

month = jan,

day = "1",

doi = "10.1109/CDC.2013.6760552",

language = "English",

isbn = "9781467357173",

pages = "4308--4313",

booktitle = "2013 IEEE 52nd Annual Conference on Decision and Control, CDC 2013",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

note = "52nd IEEE Conference on Decision and Control (CDC 2013), CDC 2013 ; Conference date: 10-12-2013 Through 13-12-2013",

}

Marconato, A, Schoukens, M, Rolain, Y & Schoukens, JFM 2013, Study of the effective number of parameters in nonlinear identification benchmarks. in 2013 IEEE 52nd Annual Conference on Decision and Control, CDC 2013., 6760552, Institute of Electrical and Electronics Engineers, pp. 4308-4313, 52nd IEEE Conference on Decision and Control (CDC 2013), Florence, Italy, 10/12/13. https://doi.org/10.1109/CDC.2013.6760552

Study of the effective number of parameters in nonlinear identification benchmarks. / Marconato, Anna; Schoukens, Maarten; Rolain, Yves et al.
2013 IEEE 52nd Annual Conference on Decision and Control, CDC 2013. Institute of Electrical and Electronics Engineers, 2013. p. 4308-4313 6760552.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Study of the effective number of parameters in nonlinear identification benchmarks

AU - Marconato, Anna

AU - Schoukens, Maarten

AU - Rolain, Yves

AU - Schoukens, J.F.M.

N1 - Conference code: 52

PY - 2013/1/1

Y1 - 2013/1/1

N2 - This paper discusses the importance of the notion of effective number of parameters as a measure of model complexity. Exploiting this concept allows a fair comparison of models obtained from different model classes. Several illustrative examples of linear and nonlinear models are presented to provide more insight in the problem. As one possible way of showing that model complexity can be reduced without having to pull any parameters to zero, an approach for rank reduced estimation based on the truncated SVD is also discussed. These ideas are then applied to two nonlinear real world problems: The Wiener-Hammerstein and the Silverbox benchmarks.

AB - This paper discusses the importance of the notion of effective number of parameters as a measure of model complexity. Exploiting this concept allows a fair comparison of models obtained from different model classes. Several illustrative examples of linear and nonlinear models are presented to provide more insight in the problem. As one possible way of showing that model complexity can be reduced without having to pull any parameters to zero, an approach for rank reduced estimation based on the truncated SVD is also discussed. These ideas are then applied to two nonlinear real world problems: The Wiener-Hammerstein and the Silverbox benchmarks.

UR - http://www.scopus.com/inward/record.url?scp=84902354693&partnerID=8YFLogxK

U2 - 10.1109/CDC.2013.6760552

DO - 10.1109/CDC.2013.6760552

M3 - Conference contribution

AN - SCOPUS:84902354693

SN - 9781467357173

SP - 4308

EP - 4313

BT - 2013 IEEE 52nd Annual Conference on Decision and Control, CDC 2013

PB - Institute of Electrical and Electronics Engineers

T2 - 52nd IEEE Conference on Decision and Control (CDC 2013)

Y2 - 10 December 2013 through 13 December 2013

ER -

Study of the effective number of parameters in nonlinear identification benchmarks

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this