List your company
ResearchFormerly WizWand

Out-of-Distribution Detection on FashionMNIST (In) vs MNIST (Out) (test)

0.9718AUROCLikelihood Ratio
0.762240.8166450.871050.925455Feb 10, 2026Mar 2, 2026Mar 22, 2026Apr 11, 2026May 1, 2026May 21, 2026Jun 11, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Likelihood RatioFlow Model=ResFlow2026.020.9718
LIDFlow Model=ResFlow2026.020.951
lambda (Similarity-only)2026.020.9445
SPEM-noiseFlow Model=ResFlow, Feature Extractor=ResNet-152, Hyperparameter alpha=0.12026.020.9432
VaLLA 200Inducing points=2002026.060.933
LLA KFACApproximation type=Kronecker-factorized2026.060.932
VaLLA 100Inducing points=1002026.060.923
SPEMFlow Model=ResFlow, Feature Extractor=ResNet-152, Hyperparameter alpha=0.42026.020.9207
LLA DiagApproximation type=Diagonal2026.060.914
LLA*Layer restriction=Last-layer2026.060.882
ComplexityFlow Model=ResFlow2026.020.8811
LLA* KFACLayer restriction=Last-layer, Approximation type=Kronecker-factorized2026.060.88
LikelihoodFlow Model=ResFlow2026.020.8797
MAP2026.060.874
ELLANumber of random points=20002026.060.874
GMMFlow Model=ResFlow2026.020.8486
Typicality (sqrt(d))Flow Model=ResFlow2026.020.7914
Typicality (Entropy)Flow Model=ResFlow2026.020.7703