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.

mixed ensembles #2

Submitted by Joerg Wichard

a long run of mixed ensembles
balanced data sets
ratio test:training = 0.25:0.75

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1676 0.211 0.1768 0.9161 0.8824 0.9075 agnostic
gina 0 0.0509 0.0608 1 0.9913 0.9846 agnostic
hiva 0.1011 0.321 0.2913 0.978 0.7959 0.7619 agnostic
nova 0.1091 0.124 0.1334 0.9697 0.9309 0.9453 agnostic
sylva 0.0037 0.0107 0.007 0.9998 0.9989 0.9988 agnostic
Overall 0.0763 0.1435 0.1339 0.9727 0.9199 0.9196 agnostic

This entry is a complete agnostic learning entry.