Uplift Modeling on eicu
0.54AUROCHybrid+Pers.
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Hybrid+Pers.Training Mode=Hybrid FL+SL, Component=personalization adapter2026.02 | 0.54 | -0.29 | -0.29 | 9.92 | -0.71 | 3.81 | 4 | 0.5 | |
| CentralizedTraining Mode=Pooled2026.02 | 0.51 | -0.03 | -0.29 | 9.92 | -0.81 | 0 | 5 | — | |
| SplitTraining Mode=Split Learning2026.02 | 0.51 | -0.13 | -0.29 | 9.92 | -0.72 | 3.61 | 4 | 0.49 | |
| HybridTraining Mode=Hybrid FL+SL2026.02 | 0.49 | -0.1 | -0.29 | 9.92 | -0.06 | 3.81 | 4 | 0.48 | |
| FedAvgTraining Mode=Federated Learning2026.02 | 0.48 | -0.11 | -0.29 | 9.92 | -0.37 | 2.04 | 8 | — | |
| Hybrid+Def.Training Mode=Hybrid FL+SL, Defense=clipping + Gaussian noise2026.02 | 0.48 | -0.21 | -0.29 | 17.06 | -0.38 | 3.81 | 4 | 0.42 |