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.

serate quadratic lssvm

Submitted by reference

LS-SVM with a normalised linear kernel, model selection via scaled conjugate gradient descent, minimising a smoothed error rate criterion.

Performance Estimates:

ADA - BER = 0.188416 : AUC = 0.898034

GINA - BER = 0.056555 : AUC = 0.984516

HIVA - BER = 0.258698 : AUC = 0.789676

NOVA - BER = 0.052420 : AUC = 0.986467

SYLVA - BER = 0.007641 : AUC = 0.999033

Dataset Balanced Error Area Under Curve  
Train Valid Test Train Valid Test
ada 0.1655 0.2077 0.1856 0.9207 0.8719 0.897 agnostic
gina 0 0.0348 0.0514 1 0.9933 0.9878 agnostic
hiva 0.0124 0.2878 0.2772 0.9997 0.7191 0.7667 agnostic
nova 0.0004 0.074 0.0491 1 0.9957 0.99 agnostic
sylva 0.002 0.0053 0.0078 0.9999 0.999 0.9988 agnostic
Overall 0.0361 0.1219 0.1142 0.9841 0.9158 0.9281 agnostic