A Scalable Algorithm for Structured Kernel Feature Selection

Shaogang Ren, Shuai Huang, John Onofrey, Xenios Papademetris, Xiaoning Qian
Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, PMLR 38:781-789, 2015.

Abstract

Kernel methods are powerful tools for nonlinear feature representation. Incorporated with structured LASSO, the kernelized structured LASSO is an effective feature selection approach that can preserve the nonlinear input-output relationships as well as the structured sparseness. But as the data dimension increases, the method can quickly become computationally prohibitive. In this paper we propose a stochastic optimization algorithm that can efficiently address this computational problem on account of the redundant kernel representations of the given data. Experiments on simulation data and PET 3D brain image data show that our method can achieve superior accuracy with less computational cost than existing methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v38-ren15, title = {{A Scalable Algorithm for Structured Kernel Feature Selection}}, author = {Ren, Shaogang and Huang, Shuai and Onofrey, John and Papademetris, Xenios and Qian, Xiaoning}, booktitle = {Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics}, pages = {781--789}, 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/ren15.pdf}, url = {https://proceedings.mlr.press/v38/ren15.html}, abstract = {Kernel methods are powerful tools for nonlinear feature representation. Incorporated with structured LASSO, the kernelized structured LASSO is an effective feature selection approach that can preserve the nonlinear input-output relationships as well as the structured sparseness. But as the data dimension increases, the method can quickly become computationally prohibitive. In this paper we propose a stochastic optimization algorithm that can efficiently address this computational problem on account of the redundant kernel representations of the given data. Experiments on simulation data and PET 3D brain image data show that our method can achieve superior accuracy with less computational cost than existing methods.} }
Endnote
%0 Conference Paper %T A Scalable Algorithm for Structured Kernel Feature Selection %A Shaogang Ren %A Shuai Huang %A John Onofrey %A Xenios Papademetris %A Xiaoning Qian %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-ren15 %I PMLR %P 781--789 %U https://proceedings.mlr.press/v38/ren15.html %V 38 %X Kernel methods are powerful tools for nonlinear feature representation. Incorporated with structured LASSO, the kernelized structured LASSO is an effective feature selection approach that can preserve the nonlinear input-output relationships as well as the structured sparseness. But as the data dimension increases, the method can quickly become computationally prohibitive. In this paper we propose a stochastic optimization algorithm that can efficiently address this computational problem on account of the redundant kernel representations of the given data. Experiments on simulation data and PET 3D brain image data show that our method can achieve superior accuracy with less computational cost than existing methods.
RIS
TY - CPAPER TI - A Scalable Algorithm for Structured Kernel Feature Selection AU - Shaogang Ren AU - Shuai Huang AU - John Onofrey AU - Xenios Papademetris AU - Xiaoning Qian 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-ren15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 38 SP - 781 EP - 789 L1 - http://proceedings.mlr.press/v38/ren15.pdf UR - https://proceedings.mlr.press/v38/ren15.html AB - Kernel methods are powerful tools for nonlinear feature representation. Incorporated with structured LASSO, the kernelized structured LASSO is an effective feature selection approach that can preserve the nonlinear input-output relationships as well as the structured sparseness. But as the data dimension increases, the method can quickly become computationally prohibitive. In this paper we propose a stochastic optimization algorithm that can efficiently address this computational problem on account of the redundant kernel representations of the given data. Experiments on simulation data and PET 3D brain image data show that our method can achieve superior accuracy with less computational cost than existing methods. ER -
APA
Ren, S., Huang, S., Onofrey, J., Papademetris, X. & Qian, X.. (2015). A Scalable Algorithm for Structured Kernel Feature Selection. Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 38:781-789 Available from https://proceedings.mlr.press/v38/ren15.html.

Related Material