Robust sketching for multiple square-root LASSO problems

Vu Pham, Laurent El Ghaoui
Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, PMLR 38:753-761, 2015.

Abstract

Many learning tasks, such as cross-validation, parameter search, or leave-one-out analysis, involve multiple instances of similar problems, each instance sharing a large part of learning data with the others. We introduce a robust framework for solving multiple square-root LASSO problems, based on a sketch of the learning data that uses low-rank approximations. Our approach allows a dramatic reduction in computational effort, in effect reducing the number of observations from m (the number of observations to start with) to k (the number of singular values retained in the low-rank model), while not sacrificing—sometimes even improving—the statistical performance. Theoretical analysis, as well as numerical experiments on both synthetic and real data, illustrate the efficiency of the method in large scale applications.

Cite this Paper


BibTeX
@InProceedings{pmlr-v38-pham15, title = {{Robust sketching for multiple square-root LASSO problems}}, author = {Vu Pham and Laurent El Ghaoui}, booktitle = {Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics}, pages = {753--761}, year = {2015}, editor = {Guy Lebanon and S. V. N. Vishwanathan}, 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/pham15.pdf}, url = { http://proceedings.mlr.press/v38/pham15.html }, abstract = {Many learning tasks, such as cross-validation, parameter search, or leave-one-out analysis, involve multiple instances of similar problems, each instance sharing a large part of learning data with the others. We introduce a robust framework for solving multiple square-root LASSO problems, based on a sketch of the learning data that uses low-rank approximations. Our approach allows a dramatic reduction in computational effort, in effect reducing the number of observations from m (the number of observations to start with) to k (the number of singular values retained in the low-rank model), while not sacrificing—sometimes even improving—the statistical performance. Theoretical analysis, as well as numerical experiments on both synthetic and real data, illustrate the efficiency of the method in large scale applications.} }
Endnote
%0 Conference Paper %T Robust sketching for multiple square-root LASSO problems %A Vu Pham %A Laurent El Ghaoui %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-pham15 %I PMLR %P 753--761 %U http://proceedings.mlr.press/v38/pham15.html %V 38 %X Many learning tasks, such as cross-validation, parameter search, or leave-one-out analysis, involve multiple instances of similar problems, each instance sharing a large part of learning data with the others. We introduce a robust framework for solving multiple square-root LASSO problems, based on a sketch of the learning data that uses low-rank approximations. Our approach allows a dramatic reduction in computational effort, in effect reducing the number of observations from m (the number of observations to start with) to k (the number of singular values retained in the low-rank model), while not sacrificing—sometimes even improving—the statistical performance. Theoretical analysis, as well as numerical experiments on both synthetic and real data, illustrate the efficiency of the method in large scale applications.
RIS
TY - CPAPER TI - Robust sketching for multiple square-root LASSO problems AU - Vu Pham AU - Laurent El Ghaoui 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-pham15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 38 SP - 753 EP - 761 L1 - http://proceedings.mlr.press/v38/pham15.pdf UR - http://proceedings.mlr.press/v38/pham15.html AB - Many learning tasks, such as cross-validation, parameter search, or leave-one-out analysis, involve multiple instances of similar problems, each instance sharing a large part of learning data with the others. We introduce a robust framework for solving multiple square-root LASSO problems, based on a sketch of the learning data that uses low-rank approximations. Our approach allows a dramatic reduction in computational effort, in effect reducing the number of observations from m (the number of observations to start with) to k (the number of singular values retained in the low-rank model), while not sacrificing—sometimes even improving—the statistical performance. Theoretical analysis, as well as numerical experiments on both synthetic and real data, illustrate the efficiency of the method in large scale applications. ER -
APA
Pham, V. & El Ghaoui, L.. (2015). Robust sketching for multiple square-root LASSO problems. Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 38:753-761 Available from http://proceedings.mlr.press/v38/pham15.html .

Related Material