A View on Model Misspecification in Uncertainty Quantification

Yuko Kato; David M.J. Tax; Marco Loog

doi:10.1007/978-3-031-39144-6_5

A View on Model Misspecification in Uncertainty Quantification

^*Corresponding author for this work

Pattern Recognition and Bioinformatics

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

44 Downloads (Pure)

Abstract

Estimating uncertainty of machine learning models is essential to assess the quality of the predictions that these models provide. However, there are several factors that influence the quality of uncertainty estimates, one of which is the amount of model misspecification. Model misspecification always exists as models are mere simplifications or approximations to reality. The question arises whether the estimated uncertainty under model misspecification is reliable or not. In this paper, we argue that model misspecification should receive more attention, by providing thought experiments and contextualizing these with relevant literature.

Original language	English
Title of host publication	Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers
Editors	Toon Calders, Bart Goethals, Celine Vens, Jefrey Lijffijt
Publisher	Springer
Pages	65-77
Number of pages	13
ISBN (Print)	9783031391439
DOIs	https://doi.org/10.1007/978-3-031-39144-6_5
Publication status	Published - 2023
Event	Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers - Mechelen, Belgium Duration: 7 Nov 2022 → 9 Nov 2022

Publication series

Name	Communications in Computer and Information Science
Volume	1805 CCIS
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Conference

Conference	Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers
Country/Territory	Belgium
City	Mechelen
Period	7/11/22 → 9/11/22

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

Epistemic and Aleatoric uncertainty
Model misspecification
Uncertainty quantification

Access to Document

10.1007/978-3-031-39144-6_5

978-3-031-39144-6_5Final published version, 548 KB

Cite this

Kato, Y., Tax, D. M. J., & Loog, M. (2023). A View on Model Misspecification in Uncertainty Quantification. In T. Calders, B. Goethals, C. Vens, & J. Lijffijt (Eds.), Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers (pp. 65-77). (Communications in Computer and Information Science; Vol. 1805 CCIS). Springer. https://doi.org/10.1007/978-3-031-39144-6_5

Kato, Yuko ; Tax, David M.J. ; Loog, Marco. / A View on Model Misspecification in Uncertainty Quantification. Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers. editor / Toon Calders ; Bart Goethals ; Celine Vens ; Jefrey Lijffijt. Springer, 2023. pp. 65-77 (Communications in Computer and Information Science).

@inproceedings{d81c6379b27640eeb21af5f71c350d84,

title = "A View on Model Misspecification in Uncertainty Quantification",

abstract = "Estimating uncertainty of machine learning models is essential to assess the quality of the predictions that these models provide. However, there are several factors that influence the quality of uncertainty estimates, one of which is the amount of model misspecification. Model misspecification always exists as models are mere simplifications or approximations to reality. The question arises whether the estimated uncertainty under model misspecification is reliable or not. In this paper, we argue that model misspecification should receive more attention, by providing thought experiments and contextualizing these with relevant literature.",

keywords = "Epistemic and Aleatoric uncertainty, Model misspecification, Uncertainty quantification",

author = "Yuko Kato and Tax, {David M.J.} and Marco Loog",

note = "Green Open Access added to TU Delft Institutional Repository {\textquoteleft}You share, we take care!{\textquoteright} – 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. ; Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers ; Conference date: 07-11-2022 Through 09-11-2022",

year = "2023",

doi = "10.1007/978-3-031-39144-6_5",

language = "English",

isbn = "9783031391439",

series = "Communications in Computer and Information Science",

publisher = "Springer",

pages = "65--77",

editor = "Toon Calders and Bart Goethals and Celine Vens and Jefrey Lijffijt",

booktitle = "Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers",

}

Kato, Y , Tax, DMJ & Loog, M 2023, A View on Model Misspecification in Uncertainty Quantification. in T Calders, B Goethals, C Vens & J Lijffijt (eds), Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers. Communications in Computer and Information Science, vol. 1805 CCIS, Springer, pp. 65-77, Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers, Mechelen, Belgium, 7/11/22. https://doi.org/10.1007/978-3-031-39144-6_5

A View on Model Misspecification in Uncertainty Quantification. / Kato, Yuko ; Tax, David M.J.; Loog, Marco.
Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers. ed. / Toon Calders; Bart Goethals; Celine Vens; Jefrey Lijffijt. Springer, 2023. p. 65-77 (Communications in Computer and Information Science; Vol. 1805 CCIS).

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

TY - GEN

T1 - A View on Model Misspecification in Uncertainty Quantification

AU - Kato, Yuko

AU - Tax, David M.J.

AU - Loog, Marco

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 - 2023

Y1 - 2023

N2 - Estimating uncertainty of machine learning models is essential to assess the quality of the predictions that these models provide. However, there are several factors that influence the quality of uncertainty estimates, one of which is the amount of model misspecification. Model misspecification always exists as models are mere simplifications or approximations to reality. The question arises whether the estimated uncertainty under model misspecification is reliable or not. In this paper, we argue that model misspecification should receive more attention, by providing thought experiments and contextualizing these with relevant literature.

AB - Estimating uncertainty of machine learning models is essential to assess the quality of the predictions that these models provide. However, there are several factors that influence the quality of uncertainty estimates, one of which is the amount of model misspecification. Model misspecification always exists as models are mere simplifications or approximations to reality. The question arises whether the estimated uncertainty under model misspecification is reliable or not. In this paper, we argue that model misspecification should receive more attention, by providing thought experiments and contextualizing these with relevant literature.

KW - Epistemic and Aleatoric uncertainty

KW - Model misspecification

KW - Uncertainty quantification

UR - http://www.scopus.com/inward/record.url?scp=85172136515&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-39144-6_5

DO - 10.1007/978-3-031-39144-6_5

M3 - Conference contribution

AN - SCOPUS:85172136515

SN - 9783031391439

T3 - Communications in Computer and Information Science

SP - 65

EP - 77

BT - Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers

A2 - Calders, Toon

A2 - Goethals, Bart

A2 - Vens, Celine

A2 - Lijffijt, Jefrey

PB - Springer

T2 - Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers

Y2 - 7 November 2022 through 9 November 2022

ER -

Kato Y , Tax DMJ , Loog M. A View on Model Misspecification in Uncertainty Quantification. In Calders T, Goethals B, Vens C, Lijffijt J, editors, Artificial Intelligence and Machine Learning - 34th Joint Benelux Conference, BNAIC/Benelearn 2022, Revised Selected Papers. Springer. 2023. p. 65-77. (Communications in Computer and Information Science). doi: 10.1007/978-3-031-39144-6_5