Similarity Learning for Nearest Neighbor Classification

Abstract : In this paper, we propose an algorithm for learning a general class of similarity measures for kNN classification. This class encompasses, among others, the standard cosine measure, as well as the Dice and Jaccard coefficients. The algorithm we propose is an extension of the voted perceptron algorithm and allows one to learn different types of similarity functions (either based on diagonal, symmetric or asymmetric similarity matrices). The results we obtained show that learning similarity measures yields significant improvements on several collections, for two prediction rules: the standard kNN rule, which was our primary goal, and a symmetric version of it.
Type de document :
Communication dans un congrès
8th IEEE International Conference on Data Mining (ICDM) 2008, 2008, Pisa, Italy. pp.983-988, 2008
Liste complète des métadonnées

https://hal.inria.fr/hal-00953867
Contributeur : Marie-Christine Fauvet <>
Soumis le : vendredi 28 février 2014 - 16:02:45
Dernière modification le : jeudi 11 janvier 2018 - 06:23:23

Identifiants

  • HAL Id : hal-00953867, version 1

Collections

Citation

Ali Mustafa Qamar, Eric Gaussier, Jean-Pierre Chevallet, Joo-Hwee Lim. Similarity Learning for Nearest Neighbor Classification. 8th IEEE International Conference on Data Mining (ICDM) 2008, 2008, Pisa, Italy. pp.983-988, 2008. 〈hal-00953867〉

Partager

Métriques

Consultations de la notice

220