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-1a

Submitted by Juha Reunanen

Same models as in cross-indexing-prior-1, but using also validation data for training.

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1606 0.1836 0.1838 0.924 0.9026 0.9034 agnostic
gina 0 0 0.025 1 1 0.9966 prior
hiva 0.0851 0.1544 0.2863 0.9674 0.9533 0.7662 agnostic
nova 0.0004 0 0.048 1 1 0.9908 agnostic
sylva 0.0026 0.0029 0.0067 0.9998 0.9999 0.9989 prior
Overall 0.0497 0.0682 0.11 0.9782 0.9712 0.9312 prior

This entry is a complete prior knowledge entry.