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-2

Submitted by Juha Reunanen

Same as 'cross-indexing-1' for all datasets except HIVA, the performance guess for which is now 0.2502.

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1773 0.2161 0.1964 0.9086 0.8624 0.8934 agnostic
gina 0 0.0285 0.0436 1 0.9967 0.9909 agnostic
hiva 0.3031 0.3942 0.3689 0.9643 0.7164 0.7654 agnostic
nova 0.0004 0.054 0.0534 1 0.9949 0.9904 agnostic
sylva 0.0048 0.0041 0.011 0.9994 0.9992 0.9987 agnostic
Overall 0.0971 0.1394 0.1347 0.9745 0.9139 0.9278 agnostic

This entry is a Model Selection Game entry.