Supervision Strategies for the Online Learning of an Evolving Classifier for Gesture Commands

Manuel Bouillon 1, * Eric Anquetil 1
* Auteur correspondant
1 IntuiDoc - intuitive user interaction for document
IRISA-D6 - MEDIA ET INTERACTIONS
Abstract : Touch sensitive interfaces enable new interaction methods like using gesture commands. To easily memorize more than a dozen of gesture commands, it is important to be able to customize them. The classifier used to recognize drawn symbols must hence be customisable, able to learn from very few data, and evolving, able to learn and improve during its use. This work studies different supervision strategies for the online training of the evolving classifier. We compare six supervision strategies, depending on user interaction (solicitation by the system), and self-evaluation capacities (notion of reject). In particular, there is a trade-off between the number of user interactions, to supervise the online training, and the error rate of the classifier. We show in this paper that the strategy giving the best results is to learn from data validated by the user, when the confidence of the recognition is too low, and from data implicitly validated.
Type de document :
Communication dans un congrès
22nd International Conference on Pattern Recognition (ICPR), 2014, Stockholm, Sweden. pp.2029-2034, 2014
Liste complète des métadonnées

Littérature citée [12 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01062587
Contributeur : Manuel Bouillon <>
Soumis le : mercredi 10 septembre 2014 - 10:10:03
Dernière modification le : jeudi 5 avril 2018 - 12:30:15
Document(s) archivé(s) le : jeudi 11 décembre 2014 - 12:11:43

Fichier

ICPR_2014_Bouillon_Anquetil.pd...
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01062587, version 1

Citation

Manuel Bouillon, Eric Anquetil. Supervision Strategies for the Online Learning of an Evolving Classifier for Gesture Commands. 22nd International Conference on Pattern Recognition (ICPR), 2014, Stockholm, Sweden. pp.2029-2034, 2014. 〈hal-01062587〉

Partager

Métriques

Consultations de la notice

350

Téléchargements de fichiers

116