Bagging regularizes (2002)
by
Tomaso Poggio
,
Ryan Rifkin
,
Sayan Mukherjee
,
Alex Rakhlin
| Venue: | AI Memo 2002-003, CBCL Memo 214, MIT AI Lab |
| Citations: | 9 - 1 self |
BibTeX
@TECHREPORT{Poggio02baggingregularizes,
author = {Tomaso Poggio and Ryan Rifkin and Sayan Mukherjee and Alex Rakhlin},
title = {Bagging regularizes},
institution = {AI Memo 2002-003, CBCL Memo 214, MIT AI Lab},
year = {2002}
}
Bookmark
 |
 |
 |
 |
 |
 |
OpenURL
Abstract
Intuitively, we expect that averaging — or bagging — different regressors with low correlation should smooth their behavior and be somewhat similar to regularization. In this note we make this intuition precise. Using an almost classical definition of stability, we prove that a certain form of averaging provides generalization bounds with a rate of convergence of the same order as Tikhonov regularization — similar to fashionable RKHSbased learning algorithms.
