Image Classification on MNIST-scale ideal downsampling, missing scales 1.0 (test)
99.14AccuracyDISCO
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DISCOtraining_augmentation=random scaling2023.11 | 99.14 | 53.71 | 0.35 | |
| SESNtraining_augmentation=random scaling2023.11 | 98.92 | 83.39 | — | |
| Scale-Equivariant Fourier Networkstraining_augmentation=none2023.11 | 98.88 | 93.66 | 0 | |
| DSStraining_augmentation=random scaling2023.11 | 98.84 | 81.05 | — | |
| SI-ConvNettraining_augmentation=random scaling2023.11 | 98.78 | 66.21 | — | |
| SS-CNNtraining_augmentation=random scaling2023.11 | 98.7 | 35.93 | — | |
| CNNtraining_augmentation=random scaling2023.11 | 98.42 | 76.17 | — | |
| Fourier CNNtraining_augmentation=random scaling2023.11 | 98.2 | 12.5 | 0.23 | |
| Per Res. CNNtraining_augmentation=random scaling2023.11 | 97.63 | 35.94 | — |