Classification on GE 50% corrupted features (50% noise)
60Accuracy (50% Noise)RF + FT
Evaluation Results
| Method | Links | |
|---|---|---|
| RF + FTFeature Selection=RF, Downstream Model=FT-Transformer2026.03 | 60 | |
| LassoFlexNetFeature Selection=End-to-end2026.03 | 59.9 | |
| AGL + FTFeature Selection=AGL, Downstream Model=FT-Transformer2026.03 | 59.1 | |
| LassoNet + FTFeature Selection=LassoNet, Downstream Model=FT-Transformer2026.03 | 57.8 | |
| Deep Lasso + FTFeature Selection=Deep Lasso, Downstream Model=FT-Transformer2026.03 | 57.7 | |
| No FS + FTFeature Selection=None, Downstream Model=FT-Transformer2026.03 | 57.6 | |
| Univariate + FTFeature Selection=Univariate, Downstream Model=FT-Transformer2026.03 | 57.4 | |
| AM + FTFeature Selection=AM, Downstream Model=FT-Transformer2026.03 | 57.2 | |
| 1L Lasso + FTFeature Selection=1L Lasso, Downstream Model=FT-Transformer2026.03 | 57.1 | |
| RF + MLPFeature Selection=RF, Downstream Model=MLP2026.03 | 56.8 | |
| XGBoost + FTFeature Selection=XGBoost, Downstream Model=FT-Transformer2026.03 | 56.4 | |
| LassoNet + MLPFeature Selection=LassoNet, Downstream Model=MLP2026.03 | 55.6 | |
| Lasso + FTFeature Selection=Lasso, Downstream Model=FT-Transformer2026.03 | 55.6 | |
| AGL + MLPFeature Selection=AGL, Downstream Model=MLP2026.03 | 53.8 | |
| AM + MLPFeature Selection=AM, Downstream Model=MLP2026.03 | 52.7 | |
| No FS + MLPFeature Selection=None, Downstream Model=MLP2026.03 | 52.5 | |
| Deep Lasso + MLPFeature Selection=Deep Lasso, Downstream Model=MLP2026.03 | 52.5 | |
| 1L Lasso + MLPFeature Selection=1L Lasso, Downstream Model=MLP2026.03 | 51.5 | |
| Univariate + MLPFeature Selection=Univariate, Downstream Model=MLP2026.03 | 51.4 | |
| Lasso + MLPFeature Selection=Lasso, Downstream Model=MLP2026.03 | 51.2 | |
| XGBoost + MLPFeature Selection=XGBoost, Downstream Model=MLP2026.03 | 51 |