Image Classification on CIFAR-10 40% asymmetric noise (test)
95.6Final AccuracyDPS+
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DPS+Model=PreAct ResNet-182025.12 | 95.6 | — | |
| CSOTModel=PreAct ResNet-182025.12 | 95.5 | — | |
| DULCModel=PreAct ResNet-182025.12 | 95.2 | — | |
| OT-FilterModel=PreAct ResNet-182025.12 | 95.1 | — | |
| Nested+Co-teachingArchitecture=PreAct ResNet-18, Pre-Cleaning=DivideMix2022.06 | 94.2 | — | |
| Dropout+Co-teachingArchitecture=PreAct ResNet-18, Pre-Cleaning=DivideMix2022.06 | 93.8 | — | |
| DivideMixArchitecture=PreAct ResNet-182022.06 | 93.4 | — | |
| Nested+Co-teachingArchitecture=PreAct ResNet-18, Pre-Cleaning=false2022.06 | 93 | — | |
| Dropout+Co-teachingArchitecture=PreAct ResNet-18, Pre-Cleaning=false2022.06 | 92.9 | — | |
| MLNTArchitecture=PreAct ResNet-18, Pre-Cleaning=DivideMix2022.06 | 92.5 | — | |
| DivideMix2020.02 | 92.1 | 93.4 | |
| DPSModel=ResNet-182025.12 | 91.65 | — | |
| TEModel=ResNet-182025.12 | 90.81 | — | |
| MLNTArchitecture=PreAct ResNet-182022.06 | 89.2 | — | |
| JOArchitecture=PreAct ResNet-182022.06 | 88.9 | — | |
| Meta-Learning2020.02 | 88.6 | 89.2 | |
| Iterative-CVArchitecture=PreAct ResNet-182022.06 | 88.6 | — | |
| PENCILArchitecture=PreAct ResNet-182022.06 | 88.5 | — | |
| Joint-Optim2020.02 | 88.4 | 88.9 | |
| P-correction2020.02 | 88.1 | 88.5 | |
| Iterative-CV2020.02 | 88 | 88.6 | |
| M-correctionArchitecture=PreAct ResNet-182022.06 | 87.4 | — | |
| F-correctionArchitecture=PreAct ResNet-182022.06 | 87.2 | — | |
| M-correction2020.02 | 86.3 | 87.4 | |
| FOCINoise type=Asymmetric, Noise rate=40%2020.08 | 86.2 | — | |
| Cross-EntropyArchitecture=PreAct ResNet-182022.06 | 85 | — | |
| SELFIENoise type=Asymmetric, Noise rate=40%2020.08 | 84.9 | — | |
| MixUpModel=ResNet-182025.12 | 84.49 | — | |
| CoteachingNoise type=Asymmetric, Noise rate=40%2020.08 | 83.7 | — | |
| F-correction2020.02 | 83.1 | 87.2 | |
| LSModel=ResNet-182025.12 | 81.96 | — | |
| SLModel=ResNet-182025.12 | 80.57 | — | |
| BaselineModel=ResNet-182025.12 | 80.3 | — | |
| Active BiasNoise type=Asymmetric, Noise rate=40%2020.08 | 78 | — | |
| JoCoRBackbone=7-layer CNN2020.03 | 76.36 | — | |
| Co-teachingBackbone=7-layer CNN2020.03 | 73.78 | — | |
| Cross-Entropy2020.02 | 72.3 | 85 | |
| DefaultNoise type=Asymmetric, Noise rate=40%2020.08 | 71.8 | — | |
| F-correctionBackbone=7-layer CNN2020.03 | 70.6 | — | |
| StandardBackbone=7-layer CNN2020.03 | 69.43 | — | |
| Co-teaching+Backbone=7-layer CNN2020.03 | 68.84 | — | |
| DecouplingBackbone=7-layer CNN2020.03 | 68.72 | — |