A step towards understanding why classification helps regression

Silvia L. Pintea; Yancong Lin; Jouke Dijkstra; Jan C. van Gemert

doi:10.1109/ICCV51070.2023.01828

A step towards understanding why classification helps regression

Silvia L. Pintea, Yancong Lin, Jouke Dijkstra, Jan C. van Gemert

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

Abstract

A number of computer vision deep regression approaches report improved results when adding a classification loss to the regression loss. Here, we explore why this is useful in practice and when it is beneficial. To do so, we start from precisely controlled dataset variations and data samplings and find that the effect of adding a classification loss is the most pronounced for regression with imbalanced data. We explain these empirical findings by formalizing the relation between the balanced and imbalanced regression losses. Finally, we show that our findings hold on two real imbalanced image datasets for depth estimation (NYUD2-DIR), and age estimation (IMDB-WIKI-DIR), and on the problem of imbalanced video progress prediction (Breakfast). Our main takeaway is: for a regression task, if the data sampling is imbalanced, then add a classification loss.

Original language	English
Title of host publication	Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV)
Editors	Cristina Ceballos
Place of Publication	Piscataway
Publisher	IEEE
Pages	19915-19924
Number of pages	10
ISBN (Electronic)	979-8-3503-0718-4
ISBN (Print)	979-8-3503-0719-1
DOIs	https://doi.org/10.1109/ICCV51070.2023.01828
Publication status	Published - 2023
Event	2023 IEEE/CVF International Conference on Computer Vision (ICCV) - Paris, France Duration: 1 Oct 2023 → 6 Oct 2023

Publication series

Name	Proceedings of the IEEE International Conference on Computer Vision
ISSN (Print)	1550-5499

Conference

Conference	2023 IEEE/CVF International Conference on Computer Vision (ICCV)
Country/Territory	France
City	Paris
Period	1/10/23 → 6/10/23

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.

Access to Document

10.1109/ICCV51070.2023.01828

Cite this

Pintea, S. L., Lin, Y., Dijkstra, J., & van Gemert, J. C. (2023). A step towards understanding why classification helps regression. In C. Ceballos (Ed.), Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV) (pp. 19915-19924). (Proceedings of the IEEE International Conference on Computer Vision). IEEE. https://doi.org/10.1109/ICCV51070.2023.01828

@inproceedings{380cfc42b50d4f5ca5ad76f31978096f,

title = "A step towards understanding why classification helps regression",

abstract = "A number of computer vision deep regression approaches report improved results when adding a classification loss to the regression loss. Here, we explore why this is useful in practice and when it is beneficial. To do so, we start from precisely controlled dataset variations and data samplings and find that the effect of adding a classification loss is the most pronounced for regression with imbalanced data. We explain these empirical findings by formalizing the relation between the balanced and imbalanced regression losses. Finally, we show that our findings hold on two real imbalanced image datasets for depth estimation (NYUD2-DIR), and age estimation (IMDB-WIKI-DIR), and on the problem of imbalanced video progress prediction (Breakfast). Our main takeaway is: for a regression task, if the data sampling is imbalanced, then add a classification loss.",

author = "Pintea, {Silvia L.} and Yancong Lin and Jouke Dijkstra and {van Gemert}, {Jan C.}",

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.; 2023 IEEE/CVF International Conference on Computer Vision (ICCV) ; Conference date: 01-10-2023 Through 06-10-2023",

year = "2023",

doi = "10.1109/ICCV51070.2023.01828",

language = "English",

isbn = "979-8-3503-0719-1",

series = "Proceedings of the IEEE International Conference on Computer Vision",

publisher = "IEEE",

pages = "19915--19924",

editor = "Ceballos, {Cristina }",

booktitle = "Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV)",

address = "United States",

}

Pintea, SL , Lin, Y, Dijkstra, J & van Gemert, JC 2023, A step towards understanding why classification helps regression. in C Ceballos (ed.), Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV). Proceedings of the IEEE International Conference on Computer Vision, IEEE, Piscataway, pp. 19915-19924, 2023 IEEE/CVF International Conference on Computer Vision (ICCV), Paris, France, 1/10/23. https://doi.org/10.1109/ICCV51070.2023.01828

A step towards understanding why classification helps regression. / Pintea, Silvia L.; Lin, Yancong; Dijkstra, Jouke et al.
Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV). ed. / Cristina Ceballos. Piscataway: IEEE, 2023. p. 19915-19924 (Proceedings of the IEEE International Conference on Computer Vision).

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

TY - GEN

T1 - A step towards understanding why classification helps regression

AU - Pintea, Silvia L.

AU - Lin, Yancong

AU - Dijkstra, Jouke

AU - van Gemert, Jan C.

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 - A number of computer vision deep regression approaches report improved results when adding a classification loss to the regression loss. Here, we explore why this is useful in practice and when it is beneficial. To do so, we start from precisely controlled dataset variations and data samplings and find that the effect of adding a classification loss is the most pronounced for regression with imbalanced data. We explain these empirical findings by formalizing the relation between the balanced and imbalanced regression losses. Finally, we show that our findings hold on two real imbalanced image datasets for depth estimation (NYUD2-DIR), and age estimation (IMDB-WIKI-DIR), and on the problem of imbalanced video progress prediction (Breakfast). Our main takeaway is: for a regression task, if the data sampling is imbalanced, then add a classification loss.

AB - A number of computer vision deep regression approaches report improved results when adding a classification loss to the regression loss. Here, we explore why this is useful in practice and when it is beneficial. To do so, we start from precisely controlled dataset variations and data samplings and find that the effect of adding a classification loss is the most pronounced for regression with imbalanced data. We explain these empirical findings by formalizing the relation between the balanced and imbalanced regression losses. Finally, we show that our findings hold on two real imbalanced image datasets for depth estimation (NYUD2-DIR), and age estimation (IMDB-WIKI-DIR), and on the problem of imbalanced video progress prediction (Breakfast). Our main takeaway is: for a regression task, if the data sampling is imbalanced, then add a classification loss.

UR - http://www.scopus.com/inward/record.url?scp=85185878467&partnerID=8YFLogxK

U2 - 10.1109/ICCV51070.2023.01828

DO - 10.1109/ICCV51070.2023.01828

M3 - Conference contribution

SN - 979-8-3503-0719-1

T3 - Proceedings of the IEEE International Conference on Computer Vision

SP - 19915

EP - 19924

BT - Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV)

A2 - Ceballos, Cristina

PB - IEEE

CY - Piscataway

T2 - 2023 IEEE/CVF International Conference on Computer Vision (ICCV)

Y2 - 1 October 2023 through 6 October 2023

ER -

A step towards understanding why classification helps regression

Abstract

Publication series

Conference

Bibliographical note

Access to Document

Other files and links

Embargoed Document

Fingerprint

Cite this