On semiseparable kernels and efficient computation of regularized system identification and function estimation

Tianshi Chen; Martin S. Andersen

doi:10.1016/j.ifacol.2020.12.222

On semiseparable kernels and efficient computation of regularized system identification and function estimation

Tianshi Chen^*
, Martin S. Andersen

^*Corresponding author for this work

The Chinese University of Hong Kong, Shenzhen

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

176 Downloads (Orbit)

Abstract

A long-standing problem for kernel-based regularization methods is their high computational complexity O(N³), where N is the number of data points. In this paper, we show that for semiseparable kernels and some typical input signals, their computational complexity can be lowered to O(Nq²), where q is the output kernel's semiseparability rank that only depends on the chosen kernel and the input signal.

Original language	English
Book series	IFAC-PapersOnLine
Volume	53
Issue number	2
Pages (from-to)	462-467
ISSN	2405-8963
DOIs	https://doi.org/10.1016/j.ifacol.2020.12.222
Publication status	Published - 2020
Event	21^st IFAC World Congress 2020 - Berlin, Germany Duration: 12 Jul 2020 → 17 Jul 2020

Conference

Conference	21^st IFAC World Congress 2020
Country/Territory	Germany
City	Berlin
Period	12/07/2020 → 17/07/2020

Bibliographical note

Funding Information:
ChineseUniversityofHongKong,Shenzhen.Sthairnte-suepUgnriavnetrsiutyndoefrHcoonngtrKaoctngN,Soh.e2n0z1h4e.n0.003.23 funded by the a flernel. Unfortunately, it did not worfl. TΩen we find 2405-8963 Copyright © 2020 The Authors. This is an open access article under the CC BY-NC-ND license. Peer review under responsibility of International Federation of Automatic Control. 10.1016/j.ifacol.2020.12.222

Publisher Copyright:
Copyright © 2020 The Authors. This is an open access article under the CC BY-NC-ND license

Keywords

Efficient computation
Kernel design
Kernel-based regularization
Semiseparable kernels
System identification

Access to Document

10.1016/j.ifacol.2020.12.222

FulltextFinal published version, 469 KB

OpenUrl availability

Check availability in DTU Findit

Cite this

@inproceedings{63b66fd75cba443fbec1566f545889b2,

title = "On semiseparable kernels and efficient computation of regularized system identification and function estimation",

abstract = "A long-standing problem for kernel-based regularization methods is their high computational complexity O(N3), where N is the number of data points. In this paper, we show that for semiseparable kernels and some typical input signals, their computational complexity can be lowered to O(Nq2), where q is the output kernel's semiseparability rank that only depends on the chosen kernel and the input signal.",

keywords = "Efficient computation, Kernel design, Kernel-based regularization, Semiseparable kernels, System identification",

author = "Tianshi Chen and Andersen, \{Martin S.\}",

note = "Funding Information: ChineseUniversityofHongKong,Shenzhen.Sthairnte-suepUgnriavnetrsiutyndoefrHcoonngtrKaoctngN,Soh.e2n0z1h4e.n0.003.23 funded by the a flernel. Unfortunately, it did not worfl. TΩen we find 2405-8963 Copyright {\textcopyright} 2020 The Authors. This is an open access article under the CC BY-NC-ND license. Peer review under responsibility of International Federation of Automatic Control. 10.1016/j.ifacol.2020.12.222 Publisher Copyright: Copyright {\textcopyright} 2020 The Authors. This is an open access article under the CC BY-NC-ND license; 21<sup>st</sup> IFAC World Congress 2020 ; Conference date: 12-07-2020 Through 17-07-2020",

year = "2020",

doi = "10.1016/j.ifacol.2020.12.222",

language = "English",

volume = "53",

pages = "462--467",

journal = "IFAC-PapersOnLine",

issn = "2405-8963",

publisher = "Elsevier",

number = "2",

}

TY - GEN

T1 - On semiseparable kernels and efficient computation of regularized system identification and function estimation

AU - Chen, Tianshi

AU - Andersen, Martin S.

N1 - Funding Information: ChineseUniversityofHongKong,Shenzhen.Sthairnte-suepUgnriavnetrsiutyndoefrHcoonngtrKaoctngN,Soh.e2n0z1h4e.n0.003.23 funded by the a flernel. Unfortunately, it did not worfl. TΩen we find 2405-8963 Copyright © 2020 The Authors. This is an open access article under the CC BY-NC-ND license. Peer review under responsibility of International Federation of Automatic Control. 10.1016/j.ifacol.2020.12.222 Publisher Copyright: Copyright © 2020 The Authors. This is an open access article under the CC BY-NC-ND license

PY - 2020

Y1 - 2020

N2 - A long-standing problem for kernel-based regularization methods is their high computational complexity O(N3), where N is the number of data points. In this paper, we show that for semiseparable kernels and some typical input signals, their computational complexity can be lowered to O(Nq2), where q is the output kernel's semiseparability rank that only depends on the chosen kernel and the input signal.

AB - A long-standing problem for kernel-based regularization methods is their high computational complexity O(N3), where N is the number of data points. In this paper, we show that for semiseparable kernels and some typical input signals, their computational complexity can be lowered to O(Nq2), where q is the output kernel's semiseparability rank that only depends on the chosen kernel and the input signal.

KW - Efficient computation

KW - Kernel design

KW - Kernel-based regularization

KW - Semiseparable kernels

KW - System identification

U2 - 10.1016/j.ifacol.2020.12.222

DO - 10.1016/j.ifacol.2020.12.222

M3 - Conference article

AN - SCOPUS:85105099203

SN - 2405-8963

VL - 53

SP - 462

EP - 467

JO - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

IS - 2

T2 - 21<sup>st</sup> IFAC World Congress 2020

Y2 - 12 July 2020 through 17 July 2020

ER -

On semiseparable kernels and efficient computation of regularized system identification and function estimation

Abstract

Conference

Bibliographical note

Keywords

Access to Document

OpenUrl availability

Fingerprint

Cite this