List your company
ResearchFormerly WizWand

Image Classification on CIFAR-10H

2.01Error Rate (%)MAP TS
1.63924.14216.6459.1479Sep 29, 2021
Updated 4mo ago

Evaluation results

MethodLinks
MAP TSBackbone=DenseNet-BC2021.092.01——
IMax BinningBackbone=DenseNet-BC2021.092.04——
Ensemble TSBackbone=DenseNet-BC2021.092.17——
P+LBackbone=DenseNet-BC, Calibration=Uncalibrated2021.092.22——
MAP TSBackbone=PreResNet-1642021.092.43——
IMax BinningBackbone=PreResNet-1642021.092.43——
Ensemble TSBackbone=PreResNet-1642021.092.46——
MAP TSBackbone=ResNet-1642021.092.78——
Ensemble TSBackbone=ResNet-1642021.092.82——
IMax BinningBackbone=ResNet-1642021.092.85——
P+LBackbone=PreResNet-164, Calibration=Uncalibrated2021.092.9——
P+LBackbone=ResNet-164, Calibration=Uncalibrated2021.093.05——
Base ModelBackbone=DenseNet-BC2021.093.25——
L+LBackbone=DenseNet-BC2021.093.39——
Ensemble TSBackbone=ResNet-1102021.093.76——
IMax BinningBackbone=ResNet-1102021.093.8——
MAP TSBackbone=ResNet-1102021.093.83——
L+LBackbone=PreResNet-1642021.094.36——
P+LBackbone=ResNet-110, Calibration=Uncalibrated2021.094.4——
Human2021.094.62——
L+LBackbone=ResNet-1102021.094.7——
L+LBackbone=ResNet-1642021.094.71——
Base ModelBackbone=PreResNet-1642021.095——
Base ModelBackbone=ResNet-1642021.096.1——
Base ModelBackbone=ResNet-1102021.0911.28——
DS-FLMethod Category=KD-based2022.12—65.27—
Fed-ETMethod Category=KD-based2022.12—78.66—
FedDFMethod Category=KD-based2022.12—73.81—
Federated DropoutMethod Category=PT-based2022.12—46.64—
FedRolexMethod Category=PT-based2022.12—69.44—
FGRTemperature Scaling=false2026.06—91.95.1
FGRTemperature Scaling=true2026.06—91.94.5
FGR + RobustTemperature Scaling=false2026.06—90.38.3
FGR + RobustTemperature Scaling=true2026.06—90.32.8
HeteroFLMethod Category=PT-based2022.12—63.9—
Homogeneous (largest)Method Category=Homogeneous, Model Capacity=largest2022.12—75.74—
Homogeneous (smallest)Method Category=Homogeneous, Model Capacity=smallest2022.12—38.82—
MoCaETemperature Scaling=false2026.06—924.9
MoCaETemperature Scaling=true2026.06—924.2
Robust FilteredTemperature Scaling=false2026.06—911.3
Robust FilteredTemperature Scaling=true2026.06—910.7
Robust MoETemperature Scaling=false2026.06—90.49
Robust MoETemperature Scaling=true2026.06—90.43.6
Single ExpertTemperature Scaling=false2026.06—92.25.4
Single ExpertTemperature Scaling=true2026.06—92.24.8
Vanilla MoETemperature Scaling=false2026.06—925.5
Vanilla MoETemperature Scaling=true2026.06—924.9