Hyperplane queries in a feature-space M-tree for speeding up active learning - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2007

Hyperplane queries in a feature-space M-tree for speeding up active learning

Résumé

In content based retrieval, relevance feedback RF is a noticeable method for reducing the "semantic gap" between the low-level features describing the content and the usually higher-level meaning of user's target. While recent RF methods based on active learning are able to identify complex target classes after relatively few iterations, they can be quite slow on very large databases. To address this scalability issue for active RF, we put forward a method that consists in the construction of an M-tree in the feature space associated to a kernet function and in performing approximate kNN hyperplane queries with the feature space M-tree. The experiments performed on two image databases show that a significant speedup can be achieved, at the expense of a limited increase in the number of feedback rounds.

Mots clés

Fichier non déposé

Dates et versions

hal-00419592 , version 1 (24-09-2009)

Identifiants

  • HAL Id : hal-00419592 , version 1

Citer

Michel Crucianu, Daniel Estevez, Vincent Oria, Jean Philippe Tarel. Hyperplane queries in a feature-space M-tree for speeding up active learning. 23èmes Journées Bases de Données Avancées (BDA'07), Oct 2007, Marseille, France. 12p. ⟨hal-00419592⟩
54 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More