List your company
ResearchFormerly WizWand

Out-of-Distribution Detection on MNIST (ID) vs Fashion-MNIST (OOD) (test)

0.9955AUCDAPPr
0.1963640.4038320.61130.818768Oct 25, 2023Mar 26, 2024Aug 26, 2024Jan 26, 2025Jun 28, 2025Nov 28, 2025May 1, 2026
Updated 6mo ago

Evaluation results

MethodLinks
DAPPrBackbone=ConvNet (3 conv + 3 dense)2026.050.9955—
BNN-ARHTBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.99510.9947
BNN-ARHTIn-Distrib.=MNIST, Architecture=LeNet [22]2023.100.99510.9947
DPNBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.99410.9937
DPNIn-Distrib.=MNIST, Architecture=LeNet [22]2023.100.99410.9937
MC DropoutBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.99330.9927
MC DropoutIn-Distrib.=MNIST, Architecture=LeNet [22]2023.100.99330.9927
F-EDLBackbone=ConvNet (3 conv + 3 dense)2026.050.9931—
R-EDLBackbone=ConvNet (3 conv + 3 dense)2026.050.9929—
I-EDLBackbone=ConvNet (3 conv + 3 dense)2026.050.9886—
REGrad*layer-selective=true, perturbation=true2024.040.98780.9879
COMBOODBackbone=LeNet2026.020.98710.9863
PostNetBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.98590.947
PostNetIn-Distrib.=MNIST, Architecture=LeNet [22]2023.100.98590.947
I-EDLBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.98490.9889
KL-PNBackbone=ConvNet (3 conv + 3 dense)2026.050.9816—
Exgrad2024.040.98110.9798
EDLBackbone=ConvNet (3 conv + 3 dense)2026.050.9808—
Entropy2024.040.98040.9794
ExGrad V Term2024.040.98040.9795
Perturb θ2024.040.97990.9764
LA2024.040.97950.9737
DUQBackbone=ConvNet (3 conv + 3 dense)2026.050.9792—
MDSBackbone=LeNet2026.020.97610.9754
Reg. MahalanobisBackbone=LeNet2026.020.97570.9726
PostNetBackbone=ConvNet (3 conv + 3 dense)2026.050.9724—
Perturb x2024.040.97090.9539
Inserted Dropout2024.040.970.9575
MC DropoutBackbone=ConvNet (3 conv + 3 dense)2026.050.9656—
GramBackbone=LeNet2026.020.96410.9637
KNNBackbone=LeNet2026.020.95680.951
UNGrad2024.040.95410.9491
MC-AA2024.040.95280.9524
ODINBackbone=LeNet2026.020.94910.9446
Kendall and GalBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.92540.9277
Kendall and GalIn-Distrib.=MNIST, Architecture=LeNet [22]2023.100.92540.9277
Deep EnsemblesBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.9070.9108
Deep EnsemblesIn-Distrib.=MNIST, Architecture=LeNet [22]2023.100.9070.9108
DetectronBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.75570.8375
DetectronIn-Distrib.=MNIST, Architecture=LeNet [22]2023.100.75570.8375
EDLBackbone=LeNet, In-Distribution Dataset=MNIST2023.100.73430.8022
EDLIn-Distrib.=MNIST, Architecture=LeNet [22]2023.100.73430.8022
RKL-PNBackbone=ConvNet (3 conv + 3 dense)2026.050.7218—
NEGrad2024.040.40990.4441
GradNorm2024.040.22710.4088
RKL-PNBackbone=LeNet, In-Distribution Dataset=MNIST2023.10—0.7845