Image Classification on CIFAR-10 Asymmetric Noise (test)
95.5Accuracy (40% Noise)CoDiM-Self
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CoDiM-Selfepoch_selection=Best2021.11 | 95.5 | — | — | — | — | — | |
| CoDiM-Supepoch_selection=Best2021.11 | 95.2 | — | — | — | — | — | |
| DMLP-DivideMixValidation=true2023.02 | 95 | — | 95.2 | — | — | — | |
| DM-AugDescepoch_selection=Best2021.11 | 94.6 | — | — | — | — | — | |
| CoDiM-bareepoch_selection=Best2021.11 | 94.4 | — | — | — | — | — | |
| CoDiM-CSSLepoch_selection=Best2021.11 | 94.2 | — | — | — | — | — | |
| DMLP-NaiveValidation=true2023.02 | 93.9 | — | 94.6 | — | — | — | |
| C2Depoch_selection=Best2021.11 | 93.6 | — | — | — | — | — | |
| DivideMixepoch_selection=Best2021.11 | 93.4 | — | — | — | — | — | |
| DivideMixValidation=false2023.02 | 93.4 | — | 93.4 | — | — | — | |
| C2D-DivideMixValidation=false2023.02 | 93.4 | — | 93.8 | — | — | — | |
| Sel-CL+Validation=false2023.02 | 93.4 | — | 95.2 | — | — | — | |
| REEDepoch_selection=Best2021.11 | 92.4 | — | — | — | — | — | |
| RRLValidation=false2023.02 | 92.4 | — | — | — | — | — | |
| REEDValidation=false2023.02 | 92.3 | — | 95 | — | — | — | |
| Joint Optimization FrameworkEpoch selection=last, Number of trials=52018.03 | 91.7 | 93.2 | 92.8 | 92.4 | 84.7 | — | |
| Joint-OptimValidation=false2023.02 | 91.7 | — | 92.8 | — | — | — | |
| MSLCValidation=true2023.02 | 91.6 | — | 94.4 | — | — | — | |
| Joint Optimization FrameworkEpoch selection=best, Number of trials=52018.03 | 91.5 | 93.2 | 92.7 | 92.4 | 84.6 | — | |
| PENCILValidation=false2023.02 | 91.2 | — | 92.4 | — | — | — | |
| ForwardEpoch selection=best, Number of trials=52018.03 | 90.3 | 92.4 | 91.4 | 91 | 83.8 | — | |
| Zhang, et al.Validation=true2023.02 | 90.2 | — | 92.7 | — | — | — | |
| CNN-CRFEpoch selection=best, Number of trials=52018.03 | 89.5 | 92 | 91.5 | 90.7 | 84 | — | |
| Meta-LearningValidation=true2023.02 | 88.6 | — | — | — | — | — | |
| Iterative-CVValidation=false2023.02 | 88 | — | — | — | — | — | |
| M-correctionepoch_selection=Best2021.11 | 87.4 | — | — | — | — | — | |
| Cross Entropy LossEpoch selection=best, Number of trials=52018.03 | 87.1 | 91.8 | 90.8 | 90 | 77.3 | — | |
| ForwardEpoch selection=last, Number of trials=52018.03 | 86.4 | 91.7 | 89.7 | 88 | 80.9 | — | |
| M-correctionValidation=false2023.02 | 86.3 | — | — | — | — | — | |
| Cross-Entropyepoch_selection=Best2021.11 | 85 | — | — | — | — | — | |
| CNN-CRFEpoch selection=last, Number of trials=52018.03 | 79.7 | 90.3 | 86.6 | 83.6 | 76.4 | — | |
| GCEValidation=false2023.02 | 78.1 | — | 87.3 | — | — | — | |
| Cross Entropy LossEpoch selection=last, Number of trials=52018.03 | 75.7 | 89.8 | 85.4 | 81 | 70.5 | — | |
| APLBackbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 88.75 | |
| APLBackbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 87.41 | |
| AUMBackbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 92.17 | |
| AUMBackbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 90.63 | |
| BootstrapBackbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 87.69 | |
| BootstrapBackbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 85.51 | |
| Co-teachingBackbone=ResNet-32, Number of runs=52019.05 | — | — | — | — | — | 84.55 | |
| Cross EntropyBackbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 88.59 | |
| Cross EntropyBackbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 86.14 | |
| D2LBackbone=ResNet-32, Number of runs=52019.05 | — | — | — | — | — | 85.57 | |
| DecouplingBackbone=ResNet-32, Number of runs=52019.05 | — | — | — | — | — | 75.27 | |
| ELRBackbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 93.48 | |
| ELRBackbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 92.21 | |
| F-correctionBackbone=ResNet-32, Number of runs=52019.05 | — | — | — | — | — | 83.55 | |
| FLCBackbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 89.42 | |
| FLCBackbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 88.25 | |
| INCVBackbone=ResNet-32, Number of runs=52019.05 | — | — | — | — | — | 86.04 | |
| Label Smoothing (LS)Backbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 90.61 | |
| Label Smoothing (LS)Backbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 88.04 | |
| MentorNetBackbone=ResNet-32, Number of runs=52019.05 | — | — | — | — | — | 77.33 | |
| Negative/Not Label Smoothing (NLS)Backbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 90.02 | |
| Negative/Not Label Smoothing (NLS)Backbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 88.42 | |
| Peer LossBackbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 91.38 | |
| Peer LossBackbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 89.65 | |
| SCEBackbone=ResNet-34, Noise Rate (epsilon)=0.22021.06 | — | — | — | — | — | 88.07 | |
| SCEBackbone=ResNet-34, Noise Rate (epsilon)=0.32021.06 | — | — | — | — | — | 85.93 |