Medical Image Classification on RNFL-OCT (test)
0.418EOdds (Age)FairBatch
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| FairBatchBackbone=ResNet-34, Validation-set tuning=true, Youden’s J statistic=true2026.05 | 0.418 | 0.32 | 0.253 | 0.33 | 0.864 | 0.911 | 0.856 | 0.877 | -0.028 | |
| BaselineBackbone=ResNet-34, Validation-set tuning=true, Youden’s J statistic=true2026.05 | 0.402 | 0.398 | 0.261 | 0.354 | 0.841 | 0.869 | 0.912 | 0.874 | — | |
| AdvBackbone=ResNet-34, Validation-set tuning=true, Youden’s J statistic=true2026.05 | 0.37 | 0.304 | 0.195 | 0.29 | 0.861 | 0.905 | 0.926 | 0.897 | -0.008 | |
| GDROBackbone=ResNet-34, Validation-set tuning=true, Youden’s J statistic=true2026.05 | 0.345 | 0.352 | 0.271 | 0.323 | 0.877 | 0.93 | 0.915 | 0.907 | -0.011 | |
| JTTBackbone=ResNet-34, Validation-set tuning=true, Youden’s J statistic=true2026.05 | 0.314 | 0.335 | 0.171 | 0.273 | 0.855 | 0.894 | 0.899 | 0.883 | -0.061 | |
| Worst-Group Equalized Odds RegularizationBackbone=ResNet-34, Validation-set tuning=true, Youden’s J statistic=true2026.05 | 0.269 | 0.249 | 0.161 | 0.226 | 0.914 | 0.939 | 0.926 | 0.926 | -0.006 | |
| FaMIBackbone=ResNet-34, Validation-set tuning=true, Youden’s J statistic=true2026.05 | 0.258 | 0.325 | 0.173 | 0.252 | 0.834 | 0.871 | 0.916 | 0.874 | -0.061 |