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.

liknon feature selection+ state of art classifiers

Submitted by Erinija

Feature selection performed by liknon. The final rule selected based on minimum error on the independent test set. The rules selected: Ensemble of linear discriminants for HIVA, NOVA and ADA. 3 Nearest neighbors for GINA, quadratic classifier for SYLVA.

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1904 0.1964 0.1955 0.8485 0.8383 0.8511 agnostic
gina 0.0332 0.0287 0.0598 0.9963 0.9898 0.9727 agnostic
hiva 0.1545 0.2664 0.2939 0.9332 0.7842 0.7589 agnostic
nova 0.0466 0.104 0.0967 0.9915 0.9531 0.9651 agnostic
sylva 0.0369 0.037 0.0386 0.5209 0.4705 0.5 agnostic
Overall 0.0923 0.1265 0.1369 0.8581 0.8072 0.8096 agnostic

This entry is a complete agnostic learning entry.