Medical Image Classification on ChestMNIST (Including Resource Metrics)
94.76AccuracyMLP-Mixer (Back-Propagation)
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MLP-Mixer (Back-Propagation)Arch=D5-W256, Training Protocol=BP2025.01 | 94.76 | 72.23 | 0.357 | — | — | — | |
| MLP-Mixer (Mono-Forward)Arch=D5-W256, Training Protocol=MF2025.01 | 94.76 | 71.03 | 0.145 | 0 | -1.2 | 0.41 | |
| MLP-Mixer (Back-Propagation)Arch=D5-W512, Training Protocol=BP2025.01 | 94.76 | 72.72 | 1.146 | — | — | — | |
| MLP-Mixer (Back-Propagation)Arch=D8-W256, Training Protocol=BP2025.01 | 94.76 | 72.11 | 0.545 | — | — | — | |
| MLP-Mixer (Back-Propagation)Arch=D8-W512, Training Protocol=BP2025.01 | 94.76 | 72.32 | 1.797 | — | — | — | |
| MLP-Mixer (Mono-Forward)Arch=D5-W512, Training Protocol=MF2025.01 | 94.75 | 71.79 | 0.476 | -0.01 | -0.93 | 0.42 | |
| MLP-Mixer (Mono-Forward)Arch=D8-W256, Training Protocol=MF2025.01 | 94.75 | 72.29 | 0.17 | 0 | 0.18 | 0.31 | |
| MLP-Mixer (Mono-Forward)Arch=D8-W512, Training Protocol=MF2025.01 | 94.75 | 71.93 | 0.571 | -0.02 | -0.4 | 0.32 | |
| MLP-Mixer (Back-Propagation)Arch=D12-W256, Training Protocol=BP2025.01 | 94.75 | 72.77 | 0.796 | — | — | — | |
| MLP-Mixer (Mono-Forward)Arch=D12-W256, Training Protocol=MF2025.01 | 94.75 | 71 | 0.202 | 0 | -1.77 | 0.25 | |
| MLP-Mixer (Back-Propagation)Arch=D12-W512, Training Protocol=BP2025.01 | 94.75 | 70.74 | 2.665 | — | — | — | |
| MLP-Mixer (Mono-Forward)Arch=D12-W512, Training Protocol=MF2025.01 | 94.75 | 71.31 | 0.698 | 0 | 0.57 | 0.26 |