List your company

OOD Detection on Dirty-MNIST vs Fashion MNIST (test)

0.271NLLENSEMBLE
0.266760.295380.3240.35262Oct 31, 2024

Evaluation results

MethodLinks
ENSEMBLEBackbone=LeNet, Runs=52024.100.27183.9151.30686.09596.065
SINGLEBackbone=LeNet, Runs=52024.100.27282.2992.90865.93550
ENSEMBLE+OOD HE-CKABackbone=LeNet, Runs=5, OOD Tuning=OOD Likelihood Minimization2024.100.27784.091.71299.99699.742
DDUBackbone=LeNet, Runs=5, OOD Metric=Feature space density (for PE)2024.100.27882.1773.95294.168—
HYPERNETBackbone=LeNet, Runs=52024.100.27881.1574.25346.39364.856
SVGD+HE-CKABackbone=LeNet, Runs=5, Kernel=Cosine similarity feature kernel2024.100.29883.8792.84694.3899.213
SVGD+RBFBackbone=LeNet, Runs=5, Kernel=RBF2024.100.30483.563.17891.00398.083
ENSEMBLE+HE-CKABackbone=LeNet, Runs=52024.100.30683.6843.17494.65698.866
HYPERNET+OOD HE-CKABackbone=LeNet, Runs=5, OOD Tuning=OOD Likelihood Minimization2024.100.32582.3983.05898.07377.548
SVGD+CKApwBackbone=LeNet, Runs=5, Kernel=Cosine similarity feature kernel2024.100.37782.3517.35989.19599.207