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

Parameter-free Online Test-time Adaptation

  • École de technologie supérieure
  • FiveAI Ltd.

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

224 Citations (Scopus)

Résumé

Training state-of-the-art vision models has become prohibitively expensive for researchers and practitioners. For the sake of accessibility and resource reuse, it is important to focus on adapting these models to a variety of down-stream scenarios. An interesting and practical paradigm is online test-time adaptation, according to which training data is inaccessible, no labelled data from the test distribution is available, and adaptation can only happen at test time and on a handful of samples. In this paper, we investigate how test-time adaptation methods fare for a number of pre-trained models on a variety of real-world scenarios, significantly extending the way they have been originally evaluated. We show that they perform well only in narrowly-defined experimental setups and sometimes fail catastrophically when their hyperparameters are not selected for the same scenario in which they are being tested. Motivated by the inherent uncertainty around the conditions that will ultimately be encountered at test time, we propose a particularly 'conservative' approach, which addresses the problem with a Laplacian Adjusted Maximum-likelihood Estimation (LAME) objective. By adapting the model's output (not its parameters), and solving our objective with an efficient concave-convex procedure, our approach exhibits a much higher average accuracy across scenarios than existing methods, while being notably faster and have a much lower memory footprint. The code is available at https://github.com/fiveai/LAME.

Langue originaleAnglais
titreProceedings - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022
EditeurIEEE Computer Society
Pages8334-8343
Nombre de pages10
ISBN (Electronique)9781665469463
Les DOIs
ÉtatPublié - 2022
Evénement2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022 - New Orleans, Etats-Unis
Durée: 19 juin 2022 → 24 juin 2022

Série de publications

NomProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
Volume2022-June
ISSN (imprimé)1063-6919

Conférence

Conférence2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022
Pays/TerritoireEtats-Unis
La villeNew Orleans
Période19/06/22 → 24/06/22

Classification disciplinaire ASJC Scopus

  • Logiciel
  • Vision par ordinateur et reconnaissance de formes

Empreinte digitale

Voici les principaux termes ou expressions associés à « Parameter-free Online Test-time Adaptation ». 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