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.

RF-CLOP

Submitted by H. Jair Escalante

Logitboost with trees, CLOP implementation. Parameters fixed according Roman Lutz's paper.

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1538 0.1578 0.169 0.9258 0.9063 0.9151 agnostic
gina 0 0 0.0366 1 1 0.9878 agnostic
hiva 0.2477 0.3073 0.3131 0.8195 0.7467 0.7475 agnostic
nova 0 0 0.0713 1 1 0.9788 agnostic
sylva 0.0045 0.0057 0.0065 0.9997 0.9995 0.9991 agnostic
Overall 0.0812 0.0942 0.1193 0.949 0.9305 0.9257 agnostic

This entry is a complete agnostic learning entry.