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.

Boost mix

Submitted by Jorge Sueiras

ADA(prior) --> Boost tree and NN.
GINA(prior) --> NN.
HIVA(agnostic) --> Boost reg.
NOVA(prior) --> Parsing and Boost tree.
SYLVA(prior) --> Boost tree.

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1587 0.1854 0.1836 0.917 0.8844 0.9055 prior
gina 0.0025 0.0537 0.0896 0.9998 0.9781 0.9655 prior
hiva 0.1885 0.2485 0.3257 0.8743 0.8125 0.7127 agnostic
nova 0.0272 0.06 0.0659 0.9911 0.9624 0.9712 prior
sylva 0.0014 0.0041 0.0157 0.9999 0.9981 0.9996 prior
Overall 0.0757 0.1103 0.1361 0.9564 0.9271 0.9109 prior

This entry is a complete prior knowledge entry.