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.

RS1

Submitted by Vladimir Nikulin

Random Subsets (RS) of features (was motivated by RF).
No prior knowledge used

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1831 0.2173 0.19 0.8098 0.7995 0.8146 agnostic
gina 0 0.0413 0.0548 1 0.9562 0.9419 agnostic
hiva 0.1856 0.3384 0.2948 0.7369 0.6301 0.6712 agnostic
nova 0.012 0.046 0.0626 0.9882 0.9466 0.9422 agnostic
sylva 0.0077 0.0053 0.0091 0.9925 0.9983 0.9927 agnostic
Overall 0.0777 0.1297 0.1223 0.9055 0.8661 0.8725 agnostic

This entry is a complete agnostic learning entry.