Incremental Multi-Classifier Learning Algorithm on Grid'5000 for Large Scale Image Annotation

Abstract : With our previous research, active learning with multi-classifier showed considering performance in large scale data but much calculation was involved. In this paper, we proposed an incremental multi-classifier (SVM classifiers were used) learning algorithm for large scale imbalanced image annotation. For further accelerating the training and predicting process, Grid'5000, French National Grid, was adopted. The result show that the best performance was reached with only 15-30% of the corpus annotated and our new method could achieve almost the same precision while save nearly 50-60% or even more than 94% of the calculation time when parallel multi-threads were used. Our proposed method will be much potential on very large scale data for less processing time.
Type de document :
Communication dans un congrès
VLS-MCMR '10 Proceedings of the international workshop on Very-large-scale multimedia corpus, mining and retrieval, Oct 2010, Firenze, Italy. 2010, 〈10.1145/1878137.1878139〉
Liste complète des métadonnées

https://hal.inria.fr/hal-00687165
Contributeur : Ist Rennes <>
Soumis le : jeudi 12 avril 2012 - 15:00:19
Dernière modification le : jeudi 11 janvier 2018 - 06:21:05

Identifiants

Collections

Citation

Yubing Tong, Bahjat Safadi, Georges Quenot. Incremental Multi-Classifier Learning Algorithm on Grid'5000 for Large Scale Image Annotation. VLS-MCMR '10 Proceedings of the international workshop on Very-large-scale multimedia corpus, mining and retrieval, Oct 2010, Firenze, Italy. 2010, 〈10.1145/1878137.1878139〉. 〈hal-00687165〉

Partager

Métriques

Consultations de la notice

70