Safe Screening of Non-Support Vectors in Pathwise SVM Computation

Kohei Ogawa, Yoshiki Suzuki, Ichiro Takeuchi
Proceedings of the 30th International Conference on Machine Learning, PMLR 28(3):1382-1390, 2013.

Abstract

In this paper, we claim that some of the non-support vectors (non-SVs) that have no influence on the classifier can be screened out prior to the training phase in pathwise SVM computation scenario, in which one is asked to train a sequence (or path) of SVM classifiers for different regularization parameters. Based on a recently proposed framework so-called safe screening rule, we derive a rule for screening out non-SVs in advance, and discuss how we can exploit the advantage of the rule in pathwise SVM computation scenario. Experiments indicate that our approach often substantially reduce the total pathwise computation cost.

Cite this Paper


BibTeX
@InProceedings{pmlr-v28-ogawa13b, title = {Safe Screening of Non-Support Vectors in Pathwise SVM Computation}, author = {Ogawa, Kohei and Suzuki, Yoshiki and Takeuchi, Ichiro}, booktitle = {Proceedings of the 30th International Conference on Machine Learning}, pages = {1382--1390}, year = {2013}, editor = {Dasgupta, Sanjoy and McAllester, David}, volume = {28}, number = {3}, series = {Proceedings of Machine Learning Research}, address = {Atlanta, Georgia, USA}, month = {17--19 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v28/ogawa13b.pdf}, url = {https://proceedings.mlr.press/v28/ogawa13b.html}, abstract = {In this paper, we claim that some of the non-support vectors (non-SVs) that have no influence on the classifier can be screened out prior to the training phase in pathwise SVM computation scenario, in which one is asked to train a sequence (or path) of SVM classifiers for different regularization parameters. Based on a recently proposed framework so-called safe screening rule, we derive a rule for screening out non-SVs in advance, and discuss how we can exploit the advantage of the rule in pathwise SVM computation scenario. Experiments indicate that our approach often substantially reduce the total pathwise computation cost.} }
Endnote
%0 Conference Paper %T Safe Screening of Non-Support Vectors in Pathwise SVM Computation %A Kohei Ogawa %A Yoshiki Suzuki %A Ichiro Takeuchi %B Proceedings of the 30th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2013 %E Sanjoy Dasgupta %E David McAllester %F pmlr-v28-ogawa13b %I PMLR %P 1382--1390 %U https://proceedings.mlr.press/v28/ogawa13b.html %V 28 %N 3 %X In this paper, we claim that some of the non-support vectors (non-SVs) that have no influence on the classifier can be screened out prior to the training phase in pathwise SVM computation scenario, in which one is asked to train a sequence (or path) of SVM classifiers for different regularization parameters. Based on a recently proposed framework so-called safe screening rule, we derive a rule for screening out non-SVs in advance, and discuss how we can exploit the advantage of the rule in pathwise SVM computation scenario. Experiments indicate that our approach often substantially reduce the total pathwise computation cost.
RIS
TY - CPAPER TI - Safe Screening of Non-Support Vectors in Pathwise SVM Computation AU - Kohei Ogawa AU - Yoshiki Suzuki AU - Ichiro Takeuchi BT - Proceedings of the 30th International Conference on Machine Learning DA - 2013/05/26 ED - Sanjoy Dasgupta ED - David McAllester ID - pmlr-v28-ogawa13b PB - PMLR DP - Proceedings of Machine Learning Research VL - 28 IS - 3 SP - 1382 EP - 1390 L1 - http://proceedings.mlr.press/v28/ogawa13b.pdf UR - https://proceedings.mlr.press/v28/ogawa13b.html AB - In this paper, we claim that some of the non-support vectors (non-SVs) that have no influence on the classifier can be screened out prior to the training phase in pathwise SVM computation scenario, in which one is asked to train a sequence (or path) of SVM classifiers for different regularization parameters. Based on a recently proposed framework so-called safe screening rule, we derive a rule for screening out non-SVs in advance, and discuss how we can exploit the advantage of the rule in pathwise SVM computation scenario. Experiments indicate that our approach often substantially reduce the total pathwise computation cost. ER -
APA
Ogawa, K., Suzuki, Y. & Takeuchi, I.. (2013). Safe Screening of Non-Support Vectors in Pathwise SVM Computation. Proceedings of the 30th International Conference on Machine Learning, in Proceedings of Machine Learning Research 28(3):1382-1390 Available from https://proceedings.mlr.press/v28/ogawa13b.html.

Related Material