Kernel Partial Least Squares is Universally Consistent

Gilles Blanchard, Nicole Krämer
Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, PMLR 9:57-64, 2010.

Abstract

We prove the statistical consistency of kernel Partial Least Squares Regression applied to a bounded regression learning problem on a reproducing kernel Hilbert space. Partial Least Squares stands out of well-known classical approaches as e.g. Ridge Regression or Principal Components Regression, as it is not defined as the solution of a global cost minimization procedure over a fixed model nor is it a linear estimator. Instead, approximate solutions are constructed by projections onto a nested set of data-dependent subspaces. To prove consistency, we exploit the known fact that Partial Least Squares is equivalent to the conjugate gradient algorithm in combination with early stopping. The choice of the stopping rule (number of iterations) is a crucial point. We study two empirical stopping rules. The first one monitors the estimation error in each iteration step of Partial Least Squares, and the second one estimates the empirical complexity in terms of a condition number. Both stopping rules lead to universally consistent estimators provided the kernel is universal.

Cite this Paper


BibTeX
@InProceedings{pmlr-v9-blanchard10a, title = {Kernel Partial Least Squares is Universally Consistent}, author = {Blanchard, Gilles and Krämer, Nicole}, booktitle = {Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics}, pages = {57--64}, year = {2010}, editor = {Teh, Yee Whye and Titterington, Mike}, volume = {9}, series = {Proceedings of Machine Learning Research}, address = {Chia Laguna Resort, Sardinia, Italy}, month = {13--15 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v9/blanchard10a/blanchard10a.pdf}, url = {https://proceedings.mlr.press/v9/blanchard10a.html}, abstract = {We prove the statistical consistency of kernel Partial Least Squares Regression applied to a bounded regression learning problem on a reproducing kernel Hilbert space. Partial Least Squares stands out of well-known classical approaches as e.g. Ridge Regression or Principal Components Regression, as it is not defined as the solution of a global cost minimization procedure over a fixed model nor is it a linear estimator. Instead, approximate solutions are constructed by projections onto a nested set of data-dependent subspaces. To prove consistency, we exploit the known fact that Partial Least Squares is equivalent to the conjugate gradient algorithm in combination with early stopping. The choice of the stopping rule (number of iterations) is a crucial point. We study two empirical stopping rules. The first one monitors the estimation error in each iteration step of Partial Least Squares, and the second one estimates the empirical complexity in terms of a condition number. Both stopping rules lead to universally consistent estimators provided the kernel is universal.} }
Endnote
%0 Conference Paper %T Kernel Partial Least Squares is Universally Consistent %A Gilles Blanchard %A Nicole Krämer %B Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2010 %E Yee Whye Teh %E Mike Titterington %F pmlr-v9-blanchard10a %I PMLR %P 57--64 %U https://proceedings.mlr.press/v9/blanchard10a.html %V 9 %X We prove the statistical consistency of kernel Partial Least Squares Regression applied to a bounded regression learning problem on a reproducing kernel Hilbert space. Partial Least Squares stands out of well-known classical approaches as e.g. Ridge Regression or Principal Components Regression, as it is not defined as the solution of a global cost minimization procedure over a fixed model nor is it a linear estimator. Instead, approximate solutions are constructed by projections onto a nested set of data-dependent subspaces. To prove consistency, we exploit the known fact that Partial Least Squares is equivalent to the conjugate gradient algorithm in combination with early stopping. The choice of the stopping rule (number of iterations) is a crucial point. We study two empirical stopping rules. The first one monitors the estimation error in each iteration step of Partial Least Squares, and the second one estimates the empirical complexity in terms of a condition number. Both stopping rules lead to universally consistent estimators provided the kernel is universal.
RIS
TY - CPAPER TI - Kernel Partial Least Squares is Universally Consistent AU - Gilles Blanchard AU - Nicole Krämer BT - Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics DA - 2010/03/31 ED - Yee Whye Teh ED - Mike Titterington ID - pmlr-v9-blanchard10a PB - PMLR DP - Proceedings of Machine Learning Research VL - 9 SP - 57 EP - 64 L1 - http://proceedings.mlr.press/v9/blanchard10a/blanchard10a.pdf UR - https://proceedings.mlr.press/v9/blanchard10a.html AB - We prove the statistical consistency of kernel Partial Least Squares Regression applied to a bounded regression learning problem on a reproducing kernel Hilbert space. Partial Least Squares stands out of well-known classical approaches as e.g. Ridge Regression or Principal Components Regression, as it is not defined as the solution of a global cost minimization procedure over a fixed model nor is it a linear estimator. Instead, approximate solutions are constructed by projections onto a nested set of data-dependent subspaces. To prove consistency, we exploit the known fact that Partial Least Squares is equivalent to the conjugate gradient algorithm in combination with early stopping. The choice of the stopping rule (number of iterations) is a crucial point. We study two empirical stopping rules. The first one monitors the estimation error in each iteration step of Partial Least Squares, and the second one estimates the empirical complexity in terms of a condition number. Both stopping rules lead to universally consistent estimators provided the kernel is universal. ER -
APA
Blanchard, G. & Krämer, N.. (2010). Kernel Partial Least Squares is Universally Consistent. Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 9:57-64 Available from https://proceedings.mlr.press/v9/blanchard10a.html.

Related Material