PASCAL Agnostic Learning
vs.
Prior Knowledge
IJCNN07

The challenge is now over. But it remains open for post-challenge submissions!


IMPORTANT: Entries made since February 1st 2007 might be using validation data, now available for training.

LogitBoost with trees

Submitted by Roman Lutz

http://stat.ethz.ch/~lutz/publ/WCCIlogitboost.php

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.155 0.19 0.166 0.9271 0.8935 0.9168 agnostic
gina 0 0.0127 0.0353 1 0.9994 0.9938 agnostic
hiva 0.1357 0.2651 0.3018 0.9433 0.7734 0.7512 agnostic
nova 0 0.048 0.0469 1 0.996 0.9905 agnostic
sylva 0.0016 0.0123 0.0078 0.9998 0.999 0.999 agnostic
Overall 0.0585 0.1056 0.1115 0.974 0.9323 0.9303 agnostic

This entry is a complete agnostic learning entry.