Predicting the Generalization Ability of a Few-Shot Classifier - IMT Atlantique Accéder directement au contenu
Article Dans Une Revue Information Année : 2021

Predicting the Generalization Ability of a Few-Shot Classifier

Résumé

In the context of few-shot learning, one cannot measure the generalization ability of a trained classifier using validation sets, due to the small number of labeled samples. In this paper, we are interested in finding alternatives to answer the question: is my classifier generalizing well to new data? We investigate the case of transfer-based few-shot learning solutions, and consider three settings: (i) supervised where we only have access to a few labeled samples, (ii) semi-supervised where we have access to both a few labeled samples and a set of unlabeled samples and (iii) unsupervised where we only have access to unlabeled samples. For each setting, we propose reasonable measures that we empirically demonstrate to be correlated with the generalization ability of the considered classifiers. We also show that these simple measures can predict the generalization ability up to a certain confidence. We conduct our experiments on standard few-shot vision datasets.
Fichier principal
Vignette du fichier
information-12-00029-1.pdf (1.27 Mo) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-03241656 , version 1 (31-05-2021)

Licence

Paternité

Identifiants

Citer

Myriam Bontonou, Louis Béthune, Vincent Gripon. Predicting the Generalization Ability of a Few-Shot Classifier. Information, 2021, 12 (1), pp.29. ⟨10.3390/info12010029⟩. ⟨hal-03241656⟩
44 Consultations
86 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More