Model selection for multi-class SVMs

Yann Guermeur (1) , Myriam Maumy (2) , Frédéric Sur (1)

1 MODBIO - Computational models in molecular biology
2 IRMA - Institut de Recherche Mathématique Avancée

In the framework of statistical learning, fitting a model to a given problem is usually done in two steps. First, model selection is performed, to set the values of the hyperparameters. Second, training results in the selection, for this set of values, of a function performing satisfactorily on the problem. Choosing the values of the hyperparameters remains a difficult task, which has only been addressed so far in the case of bi-class SVMs. We derive here a solution dedicated to M-SVMs. It is based on a new bound on the risk of large margin classifiers.

Mots clés: Multi-class SVMs; hyperparameters; soft margin parameter

Domaines: Traitement des images [eess.IV]

Yann Guermeur, Myriam Maumy, Frédéric Sur. Model selection for multi-class SVMs. International Symposium on Applied Stochastic Models and Data Analysis - ASMDA 2005, May 2005, Brest, France. ⟨inria-00104299⟩