Guarantees for Approximate Incremental SVMs

Nicolas Usunier, Antoine Bordes, Léon Bottou
Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, PMLR 9:884-891, 2010.

Abstract

Assume a teacher provides examples one by one. An approximate incremental SVM computes a sequence of classifiers that are close to the true SVM solutions computed on the successive incremental training sets. We show that simple algorithms can satisfy an averaged accuracy criterion with a computational cost that scales as well as the best SVM algorithms with the number of examples. Finally, we exhibit some experiments highlighting the benefits of joining fast incremental optimization and curriculum and active learning (Schon and Cohn, 2000; Bordes et al., 2005; Bengio et al., 2009).

Cite this Paper

BibTeX
@InProceedings{pmlr-v9-usunier10a, title = {Guarantees for Approximate Incremental SVMs}, author = {Usunier, Nicolas and Bordes, Antoine and Bottou, Léon}, booktitle = {Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics}, pages = {884--891}, year = {2010}, editor = {Teh, Yee Whye and Titterington, Mike}, volume = {9}, series = {Proceedings of Machine Learning Research}, address = {Chia Laguna Resort, Sardinia, Italy}, month = {13--15 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v9/usunier10a/usunier10a.pdf}, url = {https://proceedings.mlr.press/v9/usunier10a.html}, abstract = {Assume a teacher provides examples one by one. An approximate incremental SVM computes a sequence of classifiers that are close to the true SVM solutions computed on the successive incremental training sets. We show that simple algorithms can satisfy an averaged accuracy criterion with a computational cost that scales as well as the best SVM algorithms with the number of examples. Finally, we exhibit some experiments highlighting the benefits of joining fast incremental optimization and curriculum and active learning (Schon and Cohn, 2000; Bordes et al., 2005; Bengio et al., 2009).} }
Endnote
%0 Conference Paper %T Guarantees for Approximate Incremental SVMs %A Nicolas Usunier %A Antoine Bordes %A Léon Bottou %B Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2010 %E Yee Whye Teh %E Mike Titterington %F pmlr-v9-usunier10a %I PMLR %P 884--891 %U https://proceedings.mlr.press/v9/usunier10a.html %V 9 %X Assume a teacher provides examples one by one. An approximate incremental SVM computes a sequence of classifiers that are close to the true SVM solutions computed on the successive incremental training sets. We show that simple algorithms can satisfy an averaged accuracy criterion with a computational cost that scales as well as the best SVM algorithms with the number of examples. Finally, we exhibit some experiments highlighting the benefits of joining fast incremental optimization and curriculum and active learning (Schon and Cohn, 2000; Bordes et al., 2005; Bengio et al., 2009).
RIS
TY - CPAPER TI - Guarantees for Approximate Incremental SVMs AU - Nicolas Usunier AU - Antoine Bordes AU - Léon Bottou BT - Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics DA - 2010/03/31 ED - Yee Whye Teh ED - Mike Titterington ID - pmlr-v9-usunier10a PB - PMLR DP - Proceedings of Machine Learning Research VL - 9 SP - 884 EP - 891 L1 - http://proceedings.mlr.press/v9/usunier10a/usunier10a.pdf UR - https://proceedings.mlr.press/v9/usunier10a.html AB - Assume a teacher provides examples one by one. An approximate incremental SVM computes a sequence of classifiers that are close to the true SVM solutions computed on the successive incremental training sets. We show that simple algorithms can satisfy an averaged accuracy criterion with a computational cost that scales as well as the best SVM algorithms with the number of examples. Finally, we exhibit some experiments highlighting the benefits of joining fast incremental optimization and curriculum and active learning (Schon and Cohn, 2000; Bordes et al., 2005; Bengio et al., 2009). ER -
APA
Usunier, N., Bordes, A. & Bottou, L.. (2010). Guarantees for Approximate Incremental SVMs. Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 9:884-891 Available from https://proceedings.mlr.press/v9/usunier10a.html.

Related Material