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.

xent rbf lssvm

Submitted by reference

LS-SVM with a radial basis function kernel. Model selection using scaled conjugate gradient descent, minimising the cross-entropy criterion.

Performance Estimates:

ADA - BER = 0.188366 : AUC = 0.899357

GINA - BER = 0.057766 : AUC = 0.985230

HIVA - BER = 0.311404 : AUC = 0.722657

NOVA - BER = 0.067778 : AUC = 0.980191

SYLVA - BER = 0.011380 : AUC = 0.998757

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1629 0.2079 0.1884 0.9235 0.8746 0.8999 agnostic
gina 0 0.0442 0.05 1 0.9957 0.9892 agnostic
hiva 0 0.2838 0.3382 1 0.7485 0.6991 agnostic
nova 0 0.058 0.0674 1 0.9917 0.9832 agnostic
sylva 0.009 0.0061 0.0119 0.9991 0.9989 0.9984 agnostic
Overall 0.0344 0.12 0.1312 0.9845 0.9219 0.914 agnostic