Benchmarks

Benchmarks with DAM for Imbalanced Datasets


Benchmarks

METHOD ↓ TESTING AUC ↓ DATASET ↓ IMRATIO ↓ PARAMS PRETRAINING DATE REFERENCE
AUCM+ResNet20 0.756 Cat&Dog 1% 2.7M FALSE 04/08/2021 RobustAUC(2020)
AUCM+ResNet20 0.715 CIFAR10 1% 2.7M FALSE 04/08/2021 RobustAUC(2020)
AUCM+ResNet20 0.659 STL10 1% 2.7M FALSE 04/08/2021 RobustAUC(2020)
AUCM+ResNet20 0.715 CIFAR100 1% 2.7M FALSE 04/08/2021 RobustAUC(2020)
AUCM+ResNet20 0.920 Cat&Dog 10% 2.7M FALSE 04/08/2021 RobustAUC(2020)
AUCM+ResNet20 0.898 CIFAR10 10% 2.7M FALSE 04/08/2021 RobustAUC(2020)
AUCM+ResNet20 0.821 STL10 10% 2.7M FALSE 04/08/2021 RobustAUC(2020)
AUCM+ResNet20 0.695 CIFAR100 10% 2.7M FALSE 04/08/2021 RobustAUC(2020)

MEDICAL DATASETS

METHOD ↓ TESTING AUC ↓ DATASET ↓ IMRATIO ↓ PARAMS PRETRAINING DATE REFERENCE
DeepAUC-v1(Ensemble) 0.9305 CheXpert 20.25% - True 04/08/2021 RobustAUC(2020)
AUCM+EfficientNets(Ensemble) 0.9505 Melanoma 1.76% - True 04/08/2021 RobustAUC(2020)
AUCM+DenseNet121 0.8896 PatchCamelyon 1% - True 04/08/2021 RobustAUC(2020)