Image Classification on CIFAR-100 instance-dependent noise (IDN) (test)
68.82Acc (η=0.2)HOC Global
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HOC GlobalBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 68.82 | 62.29 | 52.96 | |
| HOC LocalBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 67.47 | 61.2 | 49.84 | |
| CEε+MAE2025.08 | 67.44 | 60.8 | 46.53 | |
| NCE+AGCE2025.08 | 65.33 | 58.59 | 43.42 | |
| NCE+RCE2025.08 | 64.13 | 57.15 | 43.22 | |
| GCE2025.08 | 61.95 | 56.99 | 44.19 | |
| Peer LossBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 61.13 | 48.01 | 33 | |
| LDR-KL2025.08 | 59.19 | 43.74 | 26.1 | |
| T-RevisionBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 58 | 40.01 | 40.88 | |
| ForwardBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 57.69 | 42.62 | 27.35 | |
| LDMIBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 57.36 | 43.06 | 26.13 | |
| CE (Standard)Backbone=ResNet-50, Pre-trained=ImageNet2021.02 | 57.26 | 41.33 | 25.08 | |
| LqBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 56.92 | 40.17 | 25.58 | |
| SCE2025.08 | 55.58 | 39.71 | 25.63 | |
| CE2025.08 | 54.46 | 40.81 | 25.57 | |
| JoCoRBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 44.55 | 23.92 | 13.05 | |
| Co-teachingBackbone=ResNet-50, Pre-trained=ImageNet2021.02 | 43.37 | 23.2 | 12.43 | |
| Co-teaching+Backbone=ResNet-50, Pre-trained=ImageNet2021.02 | 41.62 | 24.73 | 12.25 |