Convex Multi-Task Learning by Clustering

Aviad Barzilai, Koby Crammer
Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, PMLR 38:65-73, 2015.

Abstract

We consider the problem of multi-task learning in which tasks belong to hidden clusters. We formulate the learning problem as a novel convex optimization problem in which linear classifiers are combinations of (a small number of) some basis. Our formulation jointly learns both the basis and the linear combination. We propose a scalable optimization algorithm for finding the optimal solution. Our new methods outperform existing state-of-the-art methods on multi-task sentiment classification tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v38-barzilai15, title = {{Convex Multi-Task Learning by Clustering}}, author = {Barzilai, Aviad and Crammer, Koby}, booktitle = {Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics}, pages = {65--73}, year = {2015}, editor = {Lebanon, Guy and Vishwanathan, S. V. N.}, volume = {38}, series = {Proceedings of Machine Learning Research}, address = {San Diego, California, USA}, month = {09--12 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v38/barzilai15.pdf}, url = {https://proceedings.mlr.press/v38/barzilai15.html}, abstract = {We consider the problem of multi-task learning in which tasks belong to hidden clusters. We formulate the learning problem as a novel convex optimization problem in which linear classifiers are combinations of (a small number of) some basis. Our formulation jointly learns both the basis and the linear combination. We propose a scalable optimization algorithm for finding the optimal solution. Our new methods outperform existing state-of-the-art methods on multi-task sentiment classification tasks.} }
Endnote
%0 Conference Paper %T Convex Multi-Task Learning by Clustering %A Aviad Barzilai %A Koby Crammer %B Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2015 %E Guy Lebanon %E S. V. N. Vishwanathan %F pmlr-v38-barzilai15 %I PMLR %P 65--73 %U https://proceedings.mlr.press/v38/barzilai15.html %V 38 %X We consider the problem of multi-task learning in which tasks belong to hidden clusters. We formulate the learning problem as a novel convex optimization problem in which linear classifiers are combinations of (a small number of) some basis. Our formulation jointly learns both the basis and the linear combination. We propose a scalable optimization algorithm for finding the optimal solution. Our new methods outperform existing state-of-the-art methods on multi-task sentiment classification tasks.
RIS
TY - CPAPER TI - Convex Multi-Task Learning by Clustering AU - Aviad Barzilai AU - Koby Crammer BT - Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics DA - 2015/02/21 ED - Guy Lebanon ED - S. V. N. Vishwanathan ID - pmlr-v38-barzilai15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 38 SP - 65 EP - 73 L1 - http://proceedings.mlr.press/v38/barzilai15.pdf UR - https://proceedings.mlr.press/v38/barzilai15.html AB - We consider the problem of multi-task learning in which tasks belong to hidden clusters. We formulate the learning problem as a novel convex optimization problem in which linear classifiers are combinations of (a small number of) some basis. Our formulation jointly learns both the basis and the linear combination. We propose a scalable optimization algorithm for finding the optimal solution. Our new methods outperform existing state-of-the-art methods on multi-task sentiment classification tasks. ER -
APA
Barzilai, A. & Crammer, K.. (2015). Convex Multi-Task Learning by Clustering. Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 38:65-73 Available from https://proceedings.mlr.press/v38/barzilai15.html.

Related Material