Passer à la navigation principale Passer à la recherche Passer au contenu principal

A Unifying Mutual Information View of Metric Learning: Cross-Entropy vs. Pairwise Losses

  • École de technologie supérieure
  • Université Paris-Saclay

Résultats de recherche: Chapitre dans un livre, rapport, actes de conférence › Participation à un ouvrage collectif lié à un colloque ou une conférence › Revue par des pairs

142 Citations (Scopus)

Résumé

Recently, substantial research efforts in Deep Metric Learning (DML) focused on designing complex pairwise-distance losses, which require convoluted schemes to ease optimization, such as sample mining or pair weighting. The standard cross-entropy loss for classification has been largely overlooked in DML. On the surface, the cross-entropy may seem unrelated and irrelevant to metric learning as it does not explicitly involve pairwise distances. However, we provide a theoretical analysis that links the cross-entropy to several well-known and recent pairwise losses. Our connections are drawn from two different perspectives: one based on an explicit optimization insight; the other on discriminative and generative views of the mutual information between the labels and the learned features. First, we explicitly demonstrate that the cross-entropy is an upper bound on a new pairwise loss, which has a structure similar to various pairwise losses: it minimizes intra-class distances while maximizing inter-class distances. As a result, minimizing the cross-entropy can be seen as an approximate bound-optimization (or Majorize-Minimize) algorithm for minimizing this pairwise loss. Second, we show that, more generally, minimizing the cross-entropy is actually equivalent to maximizing the mutual information, to which we connect several well-known pairwise losses. Furthermore, we show that various standard pairwise losses can be explicitly related to one another via bound relationships. Our findings indicate that the cross-entropy represents a proxy for maximizing the mutual information – as pairwise losses do – without the need for convoluted sample-mining heuristics. Our experiments (Code available at: https://github.com/jeromerony/dml_cross_entropy) over four standard DML benchmarks strongly support our findings. We obtain state-of-the-art results, outperforming recent and complex DML methods.

langue originaleAnglais
titreComputer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings
rédacteurs en chefAndrea Vedaldi, Horst Bischof, Thomas Brox, Jan-Michael Frahm
EditeurSpringer Science and Business Media Deutschland GmbH
Pages548-564
Nombre de pages17
ISBN (imprimé)9783030585389
Les DOIs
étatPublié - 2020
Modification externeOui
Evénement16th European Conference on Computer Vision, ECCV 2020 - Glasgow, Royaume-Uni
Durée: 23 août 2020 → 28 août 2020

Série de publications

NomLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12351 LNCS
ISSN (imprimé)0302-9743
ISSN (Electronique)1611-3349

Conférence

Conférence16th European Conference on Computer Vision, ECCV 2020
Pays/TerritoireRoyaume-Uni
La villeGlasgow
Période23/08/20 → 28/08/20

Empreinte digitale

Voici les principaux termes ou expressions associés à « A Unifying Mutual Information View of Metric Learning: Cross-Entropy vs. Pairwise Losses ». Ces libellés thématiques sont générés à partir du titre et du résumé de la publication. Ensemble, ils forment une empreinte digitale unique.

Citer cette ressource