A Metric Learning Reality Check

Kevin Musgrave; Serge Belongie; Ser Nam Lim

doi:10.1007/978-3-030-58595-2_41

A Metric Learning Reality Check

Kevin Musgrave^*, Serge Belongie, Ser Nam Lim

^*Corresponding author af dette arbejde

Publikation: Bidrag til tidsskrift › Konferenceartikel › Forskning › peer review

228 Citationer (Scopus)

Abstract

Deep metric learning papers from the past four years have consistently claimed great advances in accuracy, often more than doubling the performance of decade-old methods. In this paper, we take a closer look at the field to see if this is actually true. We find flaws in the experimental methodology of numerous metric learning papers, and show that the actual improvements over time have been marginal at best. Code is available at github.com/KevinMusgrave/powerful-benchmarker.

Originalsprog	Engelsk
Tidsskrift	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Sider (fra-til)	681-699
Antal sider	19
ISSN	0302-9743
DOI	https://doi.org/10.1007/978-3-030-58595-2_41
Status	Udgivet - 2020
Udgivet eksternt	Ja
Begivenhed	16th European Conference on Computer Vision, ECCV 2020 - Glasgow, Storbritannien Varighed: 23 aug. 2020 → 28 aug. 2020

Konference

Konference	16th European Conference on Computer Vision, ECCV 2020
Land/Område	Storbritannien
By	Glasgow
Periode	23/08/2020 → 28/08/2020

Bibliografisk note

Publisher Copyright:
© 2020, Springer Nature Switzerland AG.

Adgang til dokumentet

10.1007/978-3-030-58595-2_41

Andre filer og links

Link to publication in Scopus

Citationsformater

@inproceedings{006b19f527754fad81f6058f115e5f09,

title = "A Metric Learning Reality Check",

abstract = "Deep metric learning papers from the past four years have consistently claimed great advances in accuracy, often more than doubling the performance of decade-old methods. In this paper, we take a closer look at the field to see if this is actually true. We find flaws in the experimental methodology of numerous metric learning papers, and show that the actual improvements over time have been marginal at best. Code is available at github.com/KevinMusgrave/powerful-benchmarker.",

keywords = "Deep metric learning",

author = "Kevin Musgrave and Serge Belongie and Lim, {Ser Nam}",

note = "Publisher Copyright: {\textcopyright} 2020, Springer Nature Switzerland AG.; 16th European Conference on Computer Vision, ECCV 2020 ; Conference date: 23-08-2020 Through 28-08-2020",

year = "2020",

doi = "10.1007/978-3-030-58595-2_41",

language = "English",

pages = "681--699",

journal = "Lecture Notes in Computer Science",

issn = "0302-9743",

publisher = "Springer Verlag",

}

TY - GEN

T1 - A Metric Learning Reality Check

AU - Musgrave, Kevin

AU - Belongie, Serge

AU - Lim, Ser Nam

PY - 2020

Y1 - 2020

N2 - Deep metric learning papers from the past four years have consistently claimed great advances in accuracy, often more than doubling the performance of decade-old methods. In this paper, we take a closer look at the field to see if this is actually true. We find flaws in the experimental methodology of numerous metric learning papers, and show that the actual improvements over time have been marginal at best. Code is available at github.com/KevinMusgrave/powerful-benchmarker.

AB - Deep metric learning papers from the past four years have consistently claimed great advances in accuracy, often more than doubling the performance of decade-old methods. In this paper, we take a closer look at the field to see if this is actually true. We find flaws in the experimental methodology of numerous metric learning papers, and show that the actual improvements over time have been marginal at best. Code is available at github.com/KevinMusgrave/powerful-benchmarker.

KW - Deep metric learning

UR - http://www.scopus.com/inward/record.url?scp=85097421278&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-58595-2_41

DO - 10.1007/978-3-030-58595-2_41

M3 - Conference article

AN - SCOPUS:85097421278

SP - 681

EP - 699

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

T2 - 16th European Conference on Computer Vision, ECCV 2020

Y2 - 23 August 2020 through 28 August 2020

ER -