List your company
ResearchFormerly WizWand

Image Classification on MNIST Origin

96.7Accuracy (Perturbation 0.01)DRAUC-Da
53.43664.66875.987.132Nov 6, 2023
Updated 8mo ago

Evaluation results

MethodLinks
DRAUC-DaModel=ResNet202023.1196.798.3798.5798.79
DRAUC-DaModel=SmallCNN2023.1196.3598.0498.5998.92
DRAUC-DfModel=SmallCNN2023.1196.0698.3898.6998.84
WDROModel=SmallCNN2023.1195.9897.6298.4898.4
DRAUC-DfModel=ResNet202023.1195.9698.2198.4498.8
GLOTModel=SmallCNN2023.1195.7897.8197.7698.56
CEModel=SmallCNN2023.1195.5497.0198.0998.41
AUCMLossModel=SmallCNN2023.1195.5298.1698.0498.6
ADVShiftModel=SmallCNN2023.1194.0697.6698.0998.09
AUCDROModel=SmallCNN2023.119497.897.7698.54
WDROModel=ResNet202023.1193.8797.8197.6698.47
CEModel=ResNet202023.1191.8897.4997.1497.88
AUCDROModel=ResNet202023.1189.1194.495.7196.73
AUCMLossModel=ResNet202023.1189.0997.8296.2697.74
DROLTModel=SmallCNN2023.1188.992.3696.3398.29
DROLTModel=ResNet202023.1188.8294.4996.1797.97
ADVShiftModel=ResNet202023.1187.4390.1296.7497.36
GLOTModel=ResNet202023.1184.4695.997.4697.07
FocalLossModel=ResNet202023.1170.7894.4595.8397.28
FocalLossModel=SmallCNN2023.1155.191.3992.6196.35