home · mobile · calendar · bactac · 2006-2007 · 

BACTAC - Reid

Ensemble Selection
PhD Candidate
11/28/2006
3:30pm-4:30pm

In machine learning, ensemble techniques nearly always outperform single-model techniques. Traditionally, ensembles have been constructed with the same base learner. Recently, however, Caruana et al (2004) developed and studied an ensemble technique (Ensemble Selection) that uses a wide variety of different base learners, and constructs the ensemble through greedy forward stepwise selection. Ensemble Selection attains state-of-the-art performance on a variety of problems, and can optimize to any easily computable performance metric. Some disadvantages of Ensemble Selection (e.g. the size of the resulting ensemble) can be partially overcome.

Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
webmaster@cs.colorado.edu
www.cs.colorado.edu
May 5, 2012 (14:24)
XHTML 1.0/CSS2
©2012