List your company
ResearchFormerly WizWand

Image Classification on CIFAR-10 Long-Tailed 1.0 (test)

6.77Top-1 ErrorClass-balanced fine-tuning
5.652413.196220.7428.2838Mar 24, 2020
Updated 5mo ago

Evaluation results

MethodLinks
Class-balanced fine-tuningImbalance factor=1, Backbone=ResNet-32, Source=Reported in [47]2020.036.77—
Focal lossImbalance factor=1, Backbone=ResNet-322020.036.97—
Class-balanced focal LossImbalance factor=1, Backbone=ResNet-322020.036.97—
Class-balanced fine-tuningImbalance factor=1, Backbone=ResNet-322020.037.08—
Cross-entropy trainingImbalance factor=1, Backbone=ResNet-32, Source=Reported in [47]2020.037.11—
Class-balanced cross-entropy lossImbalance factor=1, Backbone=ResNet-32, Source=Reported in [47]2020.037.11—
Meta-weight netImbalance factor=1, Backbone=ResNet-322020.037.19—
Ours with focal LossImbalance factor=1, Backbone=ResNet-322020.037.19—
Ours with cross-entropy lossImbalance factor=1, Backbone=ResNet-322020.037.21—
Cross-entropy trainingImbalance factor=1, Backbone=ResNet-322020.037.53—
Class-balanced cross-entropy lossImbalance factor=1, Backbone=ResNet-322020.037.53—
Ours with LDAM lossImbalance factor=1, Backbone=ResNet-322020.0310.29—
L2RWImbalance factor=1, Backbone=ResNet-32, Source=Reported in [47]2020.0310.75—
Ours with cross-entropy lossImbalance factor=10, Backbone=ResNet-322020.0311.15—
LDAM lossImbalance factor=1, Backbone=ResNet-32, Note=results reported in paper2020.0311.37—
L2RWImbalance factor=1, Backbone=ResNet-322020.0311.6—
Ours with focal LossImbalance factor=10, Backbone=ResNet-322020.0311.63—
LDAM-DRWImbalance factor=10, Backbone=ResNet-32, Note=results reported in paper2020.0311.84—
Meta-weight netImbalance factor=10, Backbone=ResNet-322020.0312.45—
Class-balanced focal LossImbalance factor=10, Backbone=ResNet-322020.0312.52—
Ours with LDAM lossImbalance factor=10, Backbone=ResNet-322020.0312.6—
LDAM lossImbalance factor=10, Backbone=ResNet-32, Note=results reported in paper2020.0313.04—
Class-balanced cross-entropy lossImbalance factor=10, Backbone=ResNet-322020.0313.23—
Focal lossImbalance factor=10, Backbone=ResNet-322020.0313.34—
Ours with cross-entropy lossImbalance factor=20, Backbone=ResNet-322020.0313.54—
Class-balanced fine-tuningImbalance factor=10, Backbone=ResNet-32, Source=Reported in [47]2020.0313.58—
Cross-entropy trainingImbalance factor=10, Backbone=ResNet-322020.0313.61—
Class-balanced fine-tuningImbalance factor=20, Backbone=ResNet-32, Source=Reported in [47]2020.0313.73—
Ours with focal LossImbalance factor=20, Backbone=ResNet-322020.0313.9—
L2RWImbalance factor=10, Backbone=ResNet-32, Source=Reported in [47]2020.0314.81—
Meta-weight netImbalance factor=20, Backbone=ResNet-322020.0315.55—
Ours with LDAM lossImbalance factor=20, Backbone=ResNet-322020.0315.63—
Class-balanced cross-entropy lossImbalance factor=20, Backbone=ResNet-322020.0315.64—
Class-balanced focal LossImbalance factor=20, Backbone=ResNet-322020.0316.22—
Class-balanced fine-tuningImbalance factor=20, Backbone=ResNet-322020.0316.78—
Class-balanced fine-tuningImbalance factor=10, Backbone=ResNet-322020.0316.83—
L2RWImbalance factor=20, Backbone=ResNet-32, Source=Reported in [47]2020.0316.9—
Ours with focal LossImbalance factor=50, Backbone=ResNet-322020.0317.12—
Focal lossImbalance factor=20, Backbone=ResNet-322020.0317.24—
Cross-entropy trainingImbalance factor=20, Backbone=ResNet-322020.0317.77—
Ours with LDAM lossImbalance factor=50, Backbone=ResNet-322020.0317.77—
L2RWImbalance factor=10, Backbone=ResNet-322020.0317.88—
L2RWImbalance factor=20, Backbone=ResNet-322020.0318.65—
Ours with cross-entropy lossImbalance factor=50, Backbone=ResNet-322020.0319.49—
Ours with LDAM lossImbalance factor=100, Backbone=ResNet-322020.0320—
Class-balanced focal LossImbalance factor=50, Backbone=ResNet-322020.0320.78—
Meta-weight netImbalance factor=50, Backbone=ResNet-322020.0320.9—
L2RWImbalance factor=50, Backbone=ResNet-32, Source=Reported in [47]2020.0321.07—
Ours with focal LossImbalance factor=100, Backbone=ResNet-322020.0321.1—
Class-balanced cross-entropy lossImbalance factor=50, Backbone=ResNet-322020.0321.95—
Class-balanced fine-tuningImbalance factor=50, Backbone=ResNet-322020.0322.56—
Class-balanced fine-tuningImbalance factor=50, Backbone=ResNet-32, Source=Reported in [47]2020.0322.58—
Ours with LDAM lossImbalance factor=200, Backbone=ResNet-322020.0322.77—
LDAM-DRWImbalance factor=100, Backbone=ResNet-32, Note=results reported in paper2020.0322.97—
Focal lossImbalance factor=50, Backbone=ResNet-322020.0323.29—
L2RWImbalance factor=50, Backbone=ResNet-322020.0323.55—
Ours with cross-entropy lossImbalance factor=100, Backbone=ResNet-322020.0323.59—
Cross-entropy trainingImbalance factor=50, Backbone=ResNet-322020.0325.19—
Class-balanced focal LossImbalance factor=100, Backbone=ResNet-322020.0325.43—
Ours with focal LossImbalance factor=200, Backbone=ResNet-322020.0325.57—
L2RWImbalance factor=100, Backbone=ResNet-32, Source=Reported in [47]2020.0325.84—
Meta-weight netImbalance factor=100, Backbone=ResNet-322020.0326.43—
LDAM lossImbalance factor=100, Backbone=ResNet-32, Note=results reported in paper2020.0326.65—
Class-balanced cross-entropy lossImbalance factor=100, Backbone=ResNet-322020.0327.63—
L2RWImbalance factor=100, Backbone=ResNet-322020.0327.77—
Class-balanced fine-tuningImbalance factor=100, Backbone=ResNet-322020.0328.66—
Class-balanced fine-tuningImbalance factor=100, Backbone=ResNet-32, Source=Reported in [47]2020.0328.67—
Ours with cross-entropy lossImbalance factor=200, Backbone=ResNet-322020.0329.34—
Focal lossImbalance factor=100, Backbone=ResNet-322020.0329.62—
Cross-entropy trainingImbalance factor=100, Backbone=ResNet-322020.0329.64—
Class-balanced cross-entropy lossImbalance factor=200, Backbone=ResNet-322020.0331.11—
Class-balanced focal LossImbalance factor=200, Backbone=ResNet-322020.0331.85—
Meta-weight netImbalance factor=200, Backbone=ResNet-322020.0332.8—
L2RWImbalance factor=200, Backbone=ResNet-32, Source=Reported in [47]2020.0333.49—
L2RWImbalance factor=200, Backbone=ResNet-322020.0333.75—
Class-balanced fine-tuningImbalance factor=200, Backbone=ResNet-322020.0333.76—
Class-balanced fine-tuningImbalance factor=200, Backbone=ResNet-32, Source=Reported in [47]2020.0333.92—
Cross-entropy trainingImbalance factor=200, Backbone=ResNet-322020.0334.32—
Focal lossImbalance factor=200, Backbone=ResNet-322020.0334.71—
BaseModelBackbone=ResNet-32, Imbalance factor=2002019.02—65.68
BaseModelBackbone=ResNet-32, Imbalance factor=1002019.02—70.36
BaseModelBackbone=ResNet-32, Imbalance factor=502019.02—74.81
BaseModelBackbone=ResNet-32, Imbalance factor=202019.02—82.23
BaseModelBackbone=ResNet-32, Imbalance factor=102019.02—86.39
BaseModelBackbone=ResNet-32, Imbalance factor=12019.02—92.89
BiKD2026.05—93.85
BKD2026.05—92.58
CCKD2026.05—92.96
Class-BalancedBackbone=ResNet-32, Imbalance factor=2002019.02—68.89
Class-BalancedBackbone=ResNet-32, Imbalance factor=1002019.02—74.57
Class-BalancedBackbone=ResNet-32, Imbalance factor=502019.02—79.27
Class-BalancedBackbone=ResNet-32, Imbalance factor=202019.02—84.36
Class-BalancedBackbone=ResNet-32, Imbalance factor=102019.02—87.49
Class-BalancedBackbone=ResNet-32, Imbalance factor=12019.02—92.89
DiVE2026.05—91.55
Fine-tuningBackbone=ResNet-32, Imbalance factor=2002019.02—66.08
Fine-tuningBackbone=ResNet-32, Imbalance factor=1002019.02—71.33
Fine-tuningBackbone=ResNet-32, Imbalance factor=502019.02—77.42
Fine-tuningBackbone=ResNet-32, Imbalance factor=202019.02—83.37
Fine-tuningBackbone=ResNet-32, Imbalance factor=102019.02—86.42