408 search results - page 38 / 82 » Learning Preferences for Multiclass Problems |
ALT
2001
Springer
14 years 6 months ago
2001
Springer
While there has been a significant amount of theoretical and empirical research on the multiple-instance learning model, most of this research is for concept learning. However, f...
ECML
2007
Springer
14 years 28 days ago
2007
Springer
Class binarizations are effective methods for improving weak learners by decomposing multi-class problems into several two-class problems. This paper analyzes how these methods can...
NIPS
2003
13 years 10 months ago
2003
We present a family of margin based online learning algorithms for various prediction tasks. In particular we derive and analyze algorithms for binary and multiclass categorizatio...
CVPR
2010
IEEE
14 years 5 months ago
2010
IEEE
Several object categorization algorithms use kernel methods over multiple cues, as they offer a principled approach to combine multiple cues, and to obtain state-of-theart perform...
ECAI
2004
Springer
14 years 23 days ago
2004
Springer
Functional genomics is a typical scientific discovery domain characterized by a very large number of attributes (genes) relative to the number of examples (observations). The dang...