Image Classification on MNIST-scale non-ideal downsampling (test)
98.8AccuracyOurs
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Oursdownsampling=non-ideal, module=Fourier layers2023.11 | 98.8 | 97.6 | 0.05 | |
| DISCOdownsampling=non-ideal2023.11 | 98.3 | 45 | 0.63 | |
| DSSdownsampling=non-ideal2023.11 | 97.72 | 57.16 | — | |
| Fourier CNNdownsampling=non-ideal2023.11 | 97.45 | 17.16 | 0.29 | |
| SESNdownsampling=non-ideal2023.11 | 97.1 | 66.66 | — | |
| SI-CovNetdownsampling=non-ideal2023.11 | 96.94 | 44.53 | — | |
| SS-CNNdownsampling=non-ideal2023.11 | 96.7 | 31.44 | — | |
| CNNdownsampling=non-ideal2023.11 | 96.42 | 10.33 | — | |
| Per Res. CNNdownsampling=non-ideal2023.11 | 94.5 | 7.42 | — |