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.

cross-indexing-prior-2

Submitted by Juha Reunanen

Like cross-indexing-prior-1 (and -1a), but model selection performed using validation data as well.
Still using only CLOP models...

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1395 0.1659 0.1807 0.9444 0.9267 0.907 agnostic
gina 0 0 0.0344 1 1 0.9945 prior
hiva 0.0639 0.0763 0.2977 0.9792 0.9793 0.7663 agnostic
nova 0.0004 0 0.0499 1 1 0.9907 agnostic
sylva 0.0036 0.0029 0.0079 0.9996 0.9996 0.999 prior
Overall 0.0415 0.049 0.1141 0.9846 0.9811 0.9315 prior

This entry is a complete prior knowledge entry.