List your company
ResearchFormerly WizWand

Image Classification on SVHN (test) (Top-1 Error Rate)

2.03Top-1 Error RateDash (RA)
0.00613.66827.3340.992Sep 1, 2021
Updated 8mo ago

Evaluation results

MethodLinks
Dash (RA)Number of labels=1000, Backbone=Wide ResNet-28-2, Augmentation strategy=RandAugment2021.092.03
Dash (CTA)Number of labels=1000, Backbone=Wide ResNet-28-2, Augmentation strategy=CTAugment2021.092.14
Dash (RA)Number of labels=250, Backbone=Wide ResNet-28-2, Augmentation strategy=RandAugment2021.092.17
FixMatch (RA)Number of labels=1000, Backbone=Wide ResNet-28-2, Augmentation strategy=RandAugment2021.092.28
RYS (UDA)Number of labels=1000, Backbone=Wide ResNet-28-22021.092.32
RYS (FixMatch)Number of labels=1000, Backbone=Wide ResNet-28-22021.092.34
FixMatch (CTA)Number of labels=1000, Backbone=Wide ResNet-28-2, Augmentation strategy=CTAugment2021.092.36
Dash (CTA)Number of labels=250, Backbone=Wide ResNet-28-2, Augmentation strategy=CTAugment2021.092.38
RYS (UDA)Number of labels=250, Backbone=Wide ResNet-28-22021.092.45
UDANumber of labels=1000, Backbone=Wide ResNet-28-22021.092.46
FixMatch (RA)Number of labels=250, Backbone=Wide ResNet-28-2, Augmentation strategy=RandAugment2021.092.48
RYS (FixMatch)Number of labels=250, Backbone=Wide ResNet-28-22021.092.63
FixMatch (CTA)Number of labels=250, Backbone=Wide ResNet-28-2, Augmentation strategy=CTAugment2021.092.64
ReMixMatchNumber of labels=1000, Backbone=Wide ResNet-28-22021.092.65
ReMixMatchNumber of labels=250, Backbone=Wide ResNet-28-22021.092.92
Dash (RA)Number of labels=40, Backbone=Wide ResNet-28-2, Augmentation strategy=RandAugment2021.093.03
Dash (CTA)Number of labels=40, Backbone=Wide ResNet-28-2, Augmentation strategy=CTAugment2021.093.14
ReMixMatchNumber of labels=40, Backbone=Wide ResNet-28-22021.093.34
Mean TeacherNumber of labels=1000, Backbone=Wide ResNet-28-22021.093.42
MixMatchNumber of labels=1000, Backbone=Wide ResNet-28-22021.093.5
Mean TeacherNumber of labels=250, Backbone=Wide ResNet-28-22021.093.57
FixMatch (RA)Number of labels=40, Backbone=Wide ResNet-28-2, Augmentation strategy=RandAugment2021.093.96
MixMatchNumber of labels=250, Backbone=Wide ResNet-28-22021.093.98
UDANumber of labels=250, Backbone=Wide ResNet-28-22021.095.69
Pi-ModelNumber of labels=1000, Backbone=Wide ResNet-28-22021.097.54
FixMatch (CTA)Number of labels=40, Backbone=Wide ResNet-28-2, Augmentation strategy=CTAugment2021.097.65
Pseudo-LabelingNumber of labels=1000, Backbone=Wide ResNet-28-22021.099.94
Pi-ModelNumber of labels=250, Backbone=Wide ResNet-28-22021.0918.96
Pseudo-LabelingNumber of labels=250, Backbone=Wide ResNet-28-22021.0920.21
MixMatchNumber of labels=40, Backbone=Wide ResNet-28-22021.0942.55
UDANumber of labels=40, Backbone=Wide ResNet-28-22021.0952.63