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.

SVM+GbO+trees

Submitted by Vladimir Nikulin

This submission includes all previous and some new stuff,
for example, RS, RVM with regularization & Naive-Bayes technique.
Now, I will concentrate on further CV-testing and JMLR-paper..

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1728 0.1917 0.1878 0.8142 0.7988 0.7938 prior
gina 0 0.0064 0.0269 1 0.9958 0.9737 prior
hiva 0 0.2322 0.282 1 0.7861 0.7104 agnostic
nova 0.0114 0.044 0.063 0.9901 0.964 0.9407 agnostic
sylva 0.0066 0.0049 0.0096 0.9962 0.9983 0.9916 agnostic
Overall 0.0382 0.0958 0.1139 0.9601 0.9086 0.882 prior

This entry is a complete prior knowledge entry.