Learning Coordination Classifiers
Full Text:
FINAL.pdf
We present a new approach to ensemble classification that requires learning only a single base classifier. The idea is to learn a classifier that simultaneouslypredicts pairs of test labels -- as opposed tolearning multiple predictors for single test labels --then coordinating the assignment of individual labelsby propagating beliefs on a graph over the data.We argue that the approach is statistically well motivated,even for independent identically distributed(iid) data. In fact, we present experimental resultsthat show improvements in classification accuracyover single-example classifiers, across a range ofiid data sets and over a set of base classifiers. Likeboosting, the technique increases representationalcapacity while controlling variance through a principledform of classifier combination.
Citation
Y. Guo,
R. Greiner,
D. Schuurmans.
"Learning Coordination Classifiers".
International Joint Conference on Artificial Intelligence (IJCAI), Edinburgh, Scotland, August 2005.
Keywords: |
machine learning, ensemble methods |
Category: |
In Conference |
BibTeX
@incollection{Guo+al:IJCAI05,
author = {Yuhong Guo and Russ Greiner and Dale Schuurmans},
title = {Learning Coordination Classifiers},
booktitle = {International Joint Conference on Artificial Intelligence
(IJCAI)},
year = 2005,
}
Last Updated: April 23, 2007
Submitted by Nelson Loyola