Incorporating Invariances in Non-Linear Support Vector Machines

Part of Advances in Neural Information Processing Systems 14 (NIPS 2001)

Bibtex Metadata Paper


Olivier Chapelle, Bernhard Schölkopf


The choice of an SVM kernel corresponds to the choice of a rep(cid:173) resentation of the data in a feature space and, to improve per(cid:173) formance, it should therefore incorporate prior knowledge such as known transformation invariances. We propose a technique which extends earlier work and aims at incorporating invariances in non(cid:173) linear kernels. We show on a digit recognition task that the pro(cid:173) posed approach is superior to the Virtual Support Vector method, which previously had been the method of choice.