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 linear lssvm

Submitted by reference

LS-SVM with a normalised linear. Model selection using scaled conjugate gradient descent, minimising the cross-entropy criterion.

Performance Estimates:

ADA - BER = 0.193500 : AUC = 0.888845

GINA - BER = 0.133080 : AUC = 0.936280

HIVA - BER = 0.284592 : AUC = 0.762422

NOVA - BER = 0.048260 : AUC = 0.988070

SYLVA - BER = 0.012925 : AUC = 0.998219

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1919 0.214 0.1962 0.8902 0.8627 0.8892 agnostic
gina 0.1014 0.1273 0.126 0.9608 0.9462 0.9461 agnostic
hiva 0.0375 0.3836 0.2983 0.9948 0.6008 0.7321 agnostic
nova 0.0004 0.044 0.0465 1 0.9968 0.99 agnostic
sylva 0.0114 0.0077 0.0147 0.9986 0.9984 0.9979 agnostic
Overall 0.0685 0.1553 0.1363 0.9689 0.881 0.9111 agnostic