\proptoSVM for Learning with Label Proportions

Felix Yu, Dong Liu, Sanjiv Kumar, Jebara Tony, Shih-Fu Chang
Proceedings of the 30th International Conference on Machine Learning, PMLR 28(3):504-512, 2013.

Abstract

We study the problem of learning with label proportions in which the training data is provided in groups and only the proportion of each class in each group is known. We propose a new method called proportion-SVM, or \proptoSVM, which explicitly models the latent unknown instance labels together with the known group label proportions in a large-margin framework. Unlike the existing works, our approach avoids making restrictive assumptions about the data. The \proptoSVM model leads to a non-convex integer programming problem. In order to solve it efficiently, we propose two algorithms: one based on simple alternating optimization and the other based on a convex relaxation. Extensive experiments on standard datasets show that \proptoSVM outperforms the state-of-the-art, especially for larger group sizes.

Cite this Paper

BibTeX
@InProceedings{pmlr-v28-yu13a, title = {$\propto$SVM for Learning with Label Proportions}, author = {Yu, Felix and Liu, Dong and Kumar, Sanjiv and Tony, Jebara and Chang, Shih-Fu}, booktitle = {Proceedings of the 30th International Conference on Machine Learning}, pages = {504--512}, 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/yu13a.pdf}, url = {https://proceedings.mlr.press/v28/yu13a.html}, abstract = {We study the problem of learning with label proportions in which the training data is provided in groups and only the proportion of each class in each group is known. We propose a new method called proportion-SVM, or \proptoSVM, which explicitly models the latent unknown instance labels together with the known group label proportions in a large-margin framework. Unlike the existing works, our approach avoids making restrictive assumptions about the data. The \proptoSVM model leads to a non-convex integer programming problem. In order to solve it efficiently, we propose two algorithms: one based on simple alternating optimization and the other based on a convex relaxation. Extensive experiments on standard datasets show that \proptoSVM outperforms the state-of-the-art, especially for larger group sizes.} }
Endnote
%0 Conference Paper %T \proptoSVM for Learning with Label Proportions %A Felix Yu %A Dong Liu %A Sanjiv Kumar %A Jebara Tony %A Shih-Fu Chang %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-yu13a %I PMLR %P 504--512 %U https://proceedings.mlr.press/v28/yu13a.html %V 28 %N 3 %X We study the problem of learning with label proportions in which the training data is provided in groups and only the proportion of each class in each group is known. We propose a new method called proportion-SVM, or \proptoSVM, which explicitly models the latent unknown instance labels together with the known group label proportions in a large-margin framework. Unlike the existing works, our approach avoids making restrictive assumptions about the data. The \proptoSVM model leads to a non-convex integer programming problem. In order to solve it efficiently, we propose two algorithms: one based on simple alternating optimization and the other based on a convex relaxation. Extensive experiments on standard datasets show that \proptoSVM outperforms the state-of-the-art, especially for larger group sizes.
RIS
TY - CPAPER TI - \proptoSVM for Learning with Label Proportions AU - Felix Yu AU - Dong Liu AU - Sanjiv Kumar AU - Jebara Tony AU - Shih-Fu Chang BT - Proceedings of the 30th International Conference on Machine Learning DA - 2013/05/26 ED - Sanjoy Dasgupta ED - David McAllester ID - pmlr-v28-yu13a PB - PMLR DP - Proceedings of Machine Learning Research VL - 28 IS - 3 SP - 504 EP - 512 L1 - http://proceedings.mlr.press/v28/yu13a.pdf UR - https://proceedings.mlr.press/v28/yu13a.html AB - We study the problem of learning with label proportions in which the training data is provided in groups and only the proportion of each class in each group is known. We propose a new method called proportion-SVM, or \proptoSVM, which explicitly models the latent unknown instance labels together with the known group label proportions in a large-margin framework. Unlike the existing works, our approach avoids making restrictive assumptions about the data. The \proptoSVM model leads to a non-convex integer programming problem. In order to solve it efficiently, we propose two algorithms: one based on simple alternating optimization and the other based on a convex relaxation. Extensive experiments on standard datasets show that \proptoSVM outperforms the state-of-the-art, especially for larger group sizes. ER -
APA
Yu, F., Liu, D., Kumar, S., Tony, J. & Chang, S.. (2013). \proptoSVM for Learning with Label Proportions. Proceedings of the 30th International Conference on Machine Learning, in Proceedings of Machine Learning Research 28(3):504-512 Available from https://proceedings.mlr.press/v28/yu13a.html.

Related Material