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.

ensemble

Submitted by reference (gcc)

ensemble of 100 LS-SVMs, model selection via PRESS, RBF kernel.

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1598 0.2062 0.1863 0.9248 0.8737 0.9004 agnostic
gina 0 0.0379 0.0504 1 0.995 0.9888 agnostic
hiva 0.0322 0.2697 0.2692 0.9981 0.7616 0.7767 agnostic
nova 0.0004 0.044 0.0617 1 0.9938 0.9852 agnostic
sylva 0.0021 0.0045 0.0084 0.9999 0.9992 0.9988 agnostic
Overall 0.0389 0.1125 0.1152 0.9846 0.9246 0.93 agnostic

This entry is a complete agnostic learning entry.