Classification on AD
87.4AccuracyInterpreTabNet
Evaluation results
| Method | Links | |
|---|---|---|
| InterpreTabNetInterpretability=Partially interpretable, Training Protocol=One split2025.12 | 87.4 | |
| TabRInterpretability=Blackbox, Training Protocol=One split2025.12 | 87 | |
| FT-TransformerInterpretability=Blackbox, Training Protocol=One split2025.12 | 85.9 | |
| T2G-FORMERInterpretability=Blackbox, Training Protocol=One split2025.12 | 85.9 | |
| TabMInterpretability=Blackbox, Training Protocol=One split2025.12 | 85.7 | |
| NODEInterpretability=Blackbox, Training Protocol=One split2025.12 | 85.7 | |
| XNNTabInterpretability=Fully interpretable, Training Protocol=5-fold-cross-validation2025.12 | 85 | |
| TabNetInterpretability=Partially interpretable, Training Protocol=One split2025.12 | 84.8 | |
| ReSSData augmentation=true, Reasoning path structure=x → z → y2026.04 | 76.07 | |
| ReSSData augmentation=false, Reasoning path structure=x → z → y2026.04 | 75.8 | |
| TabPFN2026.04 | 74.92 | |
| Direct RLReasoning path structure=x → z → y2026.04 | 74.75 | |
| TabNet2026.04 | 74.7 | |
| XGBoost2026.04 | 74.18 | |
| Direct SFTReasoning path structure=x → y2026.04 | 73.56 | |
| DRC + SFTReasoning path structure=x → z → y2026.04 | 73.34 | |
| Decision Tree2026.04 | 73.12 |