List your company
ResearchFormerly WizWand

Image Classification on ImageNet noise level 95 16H (test)

7.89Error RateP+L (MAP TS)
7.64289.311410.9812.6486Sep 29, 2021
Updated 8mo ago

Evaluation results

MethodLinks
P+L (MAP TS)Backbone=VGG-19, Number of seeds=25, Labeled data points=5067-5152, Calibration=MAP TS2021.097.89
P+L (Uncalibrated)Backbone=VGG-19, Number of seeds=25, Labeled data points=5067-51522021.098.02
P+L (IMax Binning)Backbone=VGG-19, Number of seeds=25, Labeled data points=5067-5152, Calibration=IMax Binning2021.098.62
L+LBackbone=VGG-19, Number of seeds=25, Labeled data points=5067-51522021.099.01
P+L (Ensemble TS)Backbone=VGG-19, Number of seeds=25, Labeled data points=5067-5152, Calibration=Ensemble TS2021.099.32
VGG-19Backbone=VGG-192021.0912.58
Human2021.0914.07