Image Classification on CIFAR-10N (test) (Robustness Metrics)
98.8Accuracy (Random 1)FE + NI-ERM
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| FE + NI-ERMBackbone=DINOv2, Classifier=Logistic Regression, Data Augmentation=false2024.10 | 98.8 | — | 98.65 | 98.67 | 98.69 | 95.71 | — | |
| ProMix2024.10 | 97.39 | — | 97.55 | 97.52 | 97.65 | 96.34 | — | |
| ILL2024.10 | 96.06 | — | 95.98 | 96.1 | 96.4 | 93.55 | — | |
| ILLBackbone=ResNet-34, Number of independent runs=32023.05 | 96.06 | 96.21 | 95.98 | 96.1 | 96.4 | 93.55 | — | |
| PLS2024.10 | 95.86 | — | 95.96 | 96.1 | 96.09 | 93.78 | — | |
| SOP+Backbone=PreActResNet182022.02 | 95.28 | 96.38 | 95.31 | 95.39 | 95.61 | 93.24 | — | |
| SOPBackbone=ResNet-34, Number of independent runs=32023.05 | 95.28 | 96.38 | 95.31 | 95.39 | 95.61 | 93.24 | — | |
| DivideMix2024.10 | 95.16 | — | 95.23 | 95.21 | 95.01 | 92.56 | — | |
| DivideMixBackbone=ResNet-34, Number of independent runs=32023.05 | 95.16 | — | 95.23 | 95.21 | 95.01 | 92.56 | — | |
| CORES*Backbone=ResNet342022.02 | 94.45 | 94.16 | 94.88 | 94.74 | 95.25 | 91.66 | — | |
| CORESBackbone=ResNet-34, Number of independent runs=32023.05 | 94.45 | 94.16 | 94.88 | 94.74 | 95.25 | 91.66 | — | |
| ELR+Backbone=ResNet342022.02 | 94.43 | 95.39 | 94.2 | 94.34 | 94.83 | 91.09 | — | |
| ELRBackbone=ResNet-34, Number of independent runs=32023.05 | 94.43 | 95.39 | 94.2 | 94.34 | 94.83 | 91.09 | — | |
| GNL - p(X,Y)Backbone=ResNet-34, Classifier Type=single classifier2023.08 | 91.97 | — | 91.42 | 91.83 | 92.57 | 86.99 | — | |
| GNL - p(Y|X)Backbone=ResNet-34, Classifier Type=single classifier2023.08 | 91.04 | — | 91.19 | 91.11 | 92.41 | 85.67 | — | |
| CALBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 90.93 | — | 90.75 | 90.74 | 91.97 | 85.36 | — | |
| Co-teachingBackbone=ResNet342022.02 | 90.33 | 93.35 | 90.3 | 90.15 | 91.2 | 83.83 | — | |
| Co-teachingBackbone=ResNet-34, Number of independent runs=32023.05 | 90.33 | 93.35 | 90.3 | 90.15 | 91.2 | 83.83 | — | |
| Negative-LSBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 90.29 | — | 90.37 | 90.13 | 91.97 | 82.99 | — | |
| Positive-LSBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 89.8 | — | 89.35 | 89.82 | 91.57 | 82.76 | — | |
| F-DivBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 89.7 | — | 89.79 | 89.55 | 91.64 | 82.53 | — | |
| CORES^2Backbone=ResNet-34, Classifier Type=single classifier2023.08 | 89.66 | — | 89.91 | 89.79 | 91.23 | 83.6 | — | |
| Ours2024.05 | 89.42 | — | 89.31 | 89.8 | — | 84.35 | — | |
| Peer LossBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 89.06 | — | 88.76 | 88.57 | 90.75 | 82 | — | |
| T-RevisionBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 88.33 | — | 87.71 | 80.48 | 88.52 | 80.48 | — | |
| VolMinNetBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 88.3 | — | 88.27 | 88.19 | 89.7 | 80.53 | — | |
| Joint2024.05 | 88.2 | — | 87.54 | 87.67 | — | 84.29 | — | |
| SIGUA2024.05 | 87.67 | — | 89.01 | 88.4 | — | 80.65 | — | |
| ForwardBackbone=ResNet342022.02 | 86.88 | 93.02 | 86.14 | 87.04 | 88.24 | 79.79 | — | |
| Forward TBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 86.88 | — | 86.14 | 87.04 | 88.24 | 79.79 | — | |
| ForwardBackbone=ResNet-34, Number of independent runs=32023.05 | 86.88 | 93.02 | 86.14 | 87.04 | 88.24 | 79.79 | — | |
| CEBackbone=ResNet342022.02 | 85.02 | 92.92 | 86.46 | 85.16 | 87.77 | 77.69 | — | |
| CEBackbone=ResNet-34, Classifier Type=single classifier2023.08 | 85.02 | — | 86.46 | 85.16 | 87.77 | 77.69 | — | |
| CEBackbone=ResNet-34, Number of independent runs=32023.05 | 85.02 | 92.92 | 86.46 | 85.16 | 87.77 | 77.69 | — | |
| APL2024.05 | 84.4 | — | 84.45 | 84.35 | — | 78.16 | — | |
| CE2024.05 | 83.17 | — | 82.74 | 82.9 | — | 76.57 | — | |
| CELC2024.05 | 83.11 | — | 83.09 | 82.6 | — | 73.49 | — | |
| Identifiability2024.05 | 82.52 | — | 81.97 | 82.09 | — | 71.62 | — | |
| Co-teaching2024.05 | 82.28 | — | 82.45 | 82.09 | — | 79.62 | — | |
| T-Revision2024.05 | 80.99 | — | 78.99 | 78.8 | — | 78.37 | — | |
| Co-Dis2024.05 | 80.81 | — | 80.36 | 80.76 | — | 78.12 | — | |
| AUL2024.05 | 76.26 | — | 75.24 | 75.48 | — | 63.61 | — | |
| PCE2024.05 | 63.06 | — | 62.26 | 35.47 | — | 33.8 | — | |
| CAL2022.07 | — | — | — | — | — | — | 90.93 | |
| CE2022.07 | — | — | — | — | — | — | 85.02 | |
| Co-teaching+2022.07 | — | — | — | — | — | — | 89.7 | |
| CORES*2022.07 | — | — | — | — | — | — | 94.45 | |
| Divide-Mix2022.07 | — | — | — | — | — | — | 95.16 | |
| ELR+2022.07 | — | — | — | — | — | — | 94.43 | |
| F-Div2022.07 | — | — | — | — | — | — | 89.7 | |
| Forward T2022.07 | — | — | — | — | — | — | 86.88 | |
| Negative-LS2022.07 | — | — | — | — | — | — | 90.29 | |
| Peer Loss2022.07 | — | — | — | — | — | — | 89.06 | |
| PES (Semi)mode=Semi-supervised2022.07 | — | — | — | — | — | — | 95.06 | |
| Positive-LS2022.07 | — | — | — | — | — | — | 89.8 | |
| ProMix2022.07 | — | — | — | — | — | — | 97.39 | |
| VolMinNet2022.07 | — | — | — | — | — | — | 88.3 |