Image Classification on Tiny-ImageNet 202 (test)
87.52Top-1 AccuracyMANO-tiny
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MANO-tinyParams.=28M, Complexity=O(N), Evaluation Protocol=Linear Probing2025.07 | 87.52 | — | |
| DeiT-smallParams.=22M, Complexity=O(N^2), Evaluation Protocol=Linear Probing2025.07 | 81.34 | — | |
| SwinV2-TParams.=28M, Complexity=O(N), Evaluation Protocol=Linear Probing2025.07 | 80.53 | — | |
| ViT-baseParams.=86M, Complexity=O(N^2), Evaluation Protocol=Linear Probing2025.07 | 73.07 | — | |
| Manifold MixupBackbone=PreActResNet18, Mixing Strategy=Manifold Mixup, Alpha (α)=0.22018.06 | 58.7 | 73.59 | |
| Manifold MixupBackbone=PreActResNet18, Mixing Strategy=Manifold Mixup, Alpha (α)=0.52018.06 | 57.24 | 73.48 | |
| Manifold MixupBackbone=PreActResNet18, Mixing Strategy=Manifold Mixup, Alpha (α)=12018.06 | 56.83 | 73.75 | |
| Input MixupBackbone=PreActResNet18, Mixing Strategy=Input Mixup, Alpha (α)=0.22018.06 | 56.47 | 71.74 | |
| No MixupBackbone=PreActResNet18, Mixing Strategy=None2018.06 | 55.52 | 71.04 | |
| Input MixupBackbone=PreActResNet18, Mixing Strategy=Input Mixup, Alpha (α)=0.52018.06 | 55.49 | 71.62 | |
| Input MixupBackbone=PreActResNet18, Mixing Strategy=Input Mixup, Alpha (α)=12018.06 | 52.65 | 70.7 | |
| Manifold MixupBackbone=PreActResNet18, Mixing Strategy=Manifold Mixup, Alpha (α)=22018.06 | 48.14 | 71.69 | |
| Input MixupBackbone=PreActResNet18, Mixing Strategy=Input Mixup, Alpha (α)=22018.06 | 44.18 | 68.26 |