List your company
ResearchFormerly WizWand

Near-OOD Detection on CIFAR-100 (test)

89.78AUROCNAC-UE
44.571256.308168.04579.7819Jun 5, 2023Nov 22, 2023May 11, 2024Oct 28, 2024Apr 17, 2025Oct 4, 2025Mar 24, 2026
Updated 6mo ago

Evaluation results

MethodLinks
NAC-UEBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0689.7835.06
KNNBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0689.7337.64
RMDSBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0688.8343.86
OECategory=With outliers2025.1188.330.73
OEMethod category=Training methods with outliers2025.1188.330.73
VIMBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0687.7549.19
GENBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0687.2158.75
MSPBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0687.1953.08
TempScaleBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0687.1755.81
OpenMaxBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0686.9148.06
EBOBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0686.3666.6
MLSBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0686.3166.59
ReActBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0685.9367.4
MDSBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0683.5952.81
ODINBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0682.1877
GENArchitecture=ResNet2026.0381.3154.42
GENCategory=Post-hoc inference2025.1181.3154.42
GENMethod category=Post-hoc inference methods2025.1181.3154.42
MLSCategory=Post-hoc inference2025.1181.0555.47
MLSMethod category=Post-hoc inference methods2025.1181.0555.47
SCALEArchitecture=ResNet2026.0380.99—
MixOECategory=With outliers2025.1180.9555.22
MixOEMethod category=Training methods with outliers2025.1180.9555.22
EBOCategory=Post-hoc inference2025.1180.9155.62
EBOMethod category=Post-hoc inference methods2025.1180.9155.62
TempScaleCategory=Post-hoc inference2025.1180.954.49
CRAFTCategory=Without outliers2025.1180.953.73
TempScaleMethod category=Post-hoc inference methods2025.1180.954.49
CRAFTMethod category=Training methods without outliers2025.1180.953.73
ReActCategory=Post-hoc inference2025.1180.7756.39
ReActMethod category=Post-hoc inference methods2025.1180.7756.39
ExCeLCategory=Post-hoc inference2025.1180.755.21
ExCeLMethod category=Post-hoc inference methods2025.1180.755.21
RankOODCategory=Without outliers2025.1180.6752.59
RankOODMethod category=Training methods without outliers2025.1180.6752.59
MOSCategory=Without outliers2025.1180.456.05
SHEBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0680.3181
MSPArchitecture=ResNet2026.0380.2754.8
MSPCategory=Post-hoc inference2025.1180.2754.8
MSPMethod category=Post-hoc inference methods2025.1180.2754.8
AggBBAS-EDArchitecture=ResNet2026.0380.2657.26
KNNArchitecture=ResNet2026.0380.1861.22
KNNCategory=Post-hoc inference2025.1180.1861.22
KNNMethod category=Post-hoc inference methods2025.1180.1861.22
RMDSArchitecture=ResNet2026.0380.1555.46
RMDSCategory=Post-hoc inference2025.1180.1555.46
RMDSMethod category=Post-hoc inference methods2025.1180.1555.46
AggBBAS-ECArchitecture=ResNet2026.0379.9658.23
ODINArchitecture=ResNet2026.0379.957.91
ODINCategory=Post-hoc inference2025.1179.957.91
DICECategory=Post-hoc inference2025.1179.3857.95
SHEArchitecture=ResNet2026.0378.9559.07
SHECategory=Post-hoc inference2025.1178.9559.07
LogitNormCategory=Without outliers2025.1178.4762.89
LogitNormMethod category=Training methods without outliers2025.1178.4762.89
ASHArchitecture=ResNet2026.0378.265.71
ASHCategory=Post-hoc inference2025.1178.265.71
ASHMethod category=Post-hoc inference methods2025.1178.265.71
UDGCategory=With outliers2025.1178.0261.42
UDGMethod category=Training methods with outliers2025.1178.0261.42
RankFeatBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0677.9865.32
KLMBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0677.8990.55
BBAS-ECArchitecture=ResNet2026.0377.4560.74
BBAS-EDArchitecture=ResNet2026.0377.3460.58
G-ODINCategory=Without outliers2025.1177.1567.58
G-ODINMethod category=Training methods without outliers2025.1177.1567.58
MCDCategory=With outliers2025.1177.0755.88
MCDMethod category=Training methods with outliers2025.1177.0755.88
DICEBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0677.0173.71
KLMCategory=Post-hoc inference2025.1176.5677.92
OpenMaxCategory=Post-hoc inference2025.1176.4156.58
VIMArchitecture=ResNet2026.0374.9862.63
VIMCategory=Post-hoc inference2025.1174.9862.63
VIMMethod category=Post-hoc inference methods2025.1174.9862.63
ARPLCategory=Without outliers2025.1174.9461.56
ASHBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0674.1187.31
CIDERCategory=Without outliers2025.1173.172.02
CIDERMethod category=Training methods without outliers2025.1173.172.02
ConfBranchCategory=Without outliers2025.1171.670.21
ConfBranchMethod category=Training methods without outliers2025.1171.670.21
CSICategory=Without outliers2025.1171.4570.26
GradNormCategory=Post-hoc inference2025.1170.1385.58
OpenGANCategory=Post-hoc inference2025.1165.9876.52
RankFeatCategory=Post-hoc inference2025.1161.8880.59
MDSEnsBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0661.2991.87
MDSArchitecture=ResNet2026.0358.6983.53
MDSCategory=Post-hoc inference2025.1158.6983.53
GramBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0658.3391.68
GradNormBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0654.4394.54
OpenGANBackbone=ResNet-18, In-Distribution Dataset=CIFAR-102023.0652.8194.84
GramCategory=Post-hoc inference2025.1151.6692.28
MDSEnsArchitecture=ResNet2026.0346.3195.88
MDSEnsCategory=Post-hoc inference2025.1146.3195.88