leon.bottou.org

Large Scale Transductive SVMs

Abstract: We show how the Concave-Convex Procedure can be applied to Transductive SVMs, which traditionally require solving a combinatorial search problem. This provides for the first time a highly scalable algorithm in the nonlinear case. Detailed experiments verify the utility of our approach.

Ronan Collobert, Fabian Sinz, Jason Weston and Léon Bottou: Large Scale Transductive SVMs, Journal of Machine Learning Research, 7:1687-1712, September 2006.

JMLR Link jmlr-2006.djvu jmlr-2006.pdf jmlr-2006.ps.gz

@article{collobert-sinz-weston-bottou-2006,
  author = {Collobert, Ronan and Sinz, Fabian and Weston, Jason and Bottou, L\'{e}on},
  title = {Large Scale Transductive SVMs},
  year = {2006},
  journal = {Journal of Machine Learning Research},
  volume = {7},
  pages = {1687-1712},
  month = {September},
  url = {http://leon.bottou.org/papers/collobert-sinz-weston-bottou-2006},
}

Links

The UniverSVM software is based on a modified version of the SVQP2 optimizer.
About the DC algorithms http://lita.sciences.univ-metz.fr/~lethi/DCA.html
Home page of Jason Weston
Home page of Ronan Collobert
Home page of Fabian Sinz