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.

boosted trees

Submitted by Foreseer

Boosted trees on
agnostic ADA, agnostic HIVA, precomputed SYLVA and precomputed GINA

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1831 0.2083 0.2035 0.8581 0.8436 0.8419 agnostic
gina 0 0 0.0354 1 1 0.9646 prior
hiva 0.1782 0.221 0.3234 0.8647 0.7894 0.7325 agnostic
nova 0.1781 0.19 0.2005 0.9657 0.9685 0.9574 agnostic
sylva 0.0019 0 0.0245 0.9976 1 0.976 prior
Overall 0.1083 0.1239 0.1575 0.9372 0.9203 0.8945 prior

This entry is a complete prior knowledge entry.