Ho–Kashyap with Early Stopping Versus Soft Margin SVM for Linear Classifiers –An Application

Fabien Lauer (1) , Mohamed Bentoumi (1) , Gérard Bloch (1) , Gilles Millérioux (1) , Patrice Aknin (2)

1 CRAN - Centre de Recherche en Automatique de Nancy
2 INRETS/LTN - Laboratoire des Technologies Nouvelles

In a classification problem, hard margin SVMs tend to minimize the generalization error by maximizing the margin. Regularization is obtained with soft margin SVMs which improve performances by relaxing the constraints on the margin maximization. This article shows that comparable performances can be obtained in the linearly separable case with the Ho–Kashyap learning rule associated to early stopping methods. These methods are applied on a non-destructive control application for a 4-class problem of rail defect classification.

Keywords: classification; SVM; Ho-Kashyap; non-destructive control

Domains: Machine Learning [cs.LG]; Neural and Evolutionary Computing [cs.NE]; Signal and Image processing; Signal and Image Processing

Fabien Lauer, Mohamed Bentoumi, Gérard Bloch, Gilles Millérioux, Patrice Aknin. Ho–Kashyap with Early Stopping Versus Soft Margin SVM for Linear Classifiers –An Application. International Symposium on Neural Networks, Aug 2004, Dalian, China. pp.524-530, ⟨10.1007/b99834⟩. ⟨hal-00120606⟩