Classification with Deferral on Covertype (test)
93.4System AccuracyDecoupled
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DecoupledNumber of model experts (J)=5, Evaluation protocol=mean±std over 4 seeds2026.04 | 93.4 | 0.066 | 94.1 | 71.2 | |
| Classifier onlyNumber of model experts (J)=5, Evaluation protocol=mean±std over 4 seeds2026.04 | 92.9 | 0.071 | 92.9 | 100 | |
| PiCCENumber of model experts (J)=5, Evaluation protocol=mean±std over 4 seeds2026.04 | 90.8 | 0.092 | 91.1 | 44.9 | |
| OvANumber of model experts (J)=5, Evaluation protocol=mean±std over 4 seeds2026.04 | 90.7 | 0.093 | 90.6 | 43.7 | |
| A-SMNumber of model experts (J)=5, Evaluation protocol=mean±std over 4 seeds2026.04 | 89.9 | 0.101 | 89.2 | 43.9 | |
| Add. CENumber of model experts (J)=5, Evaluation protocol=mean±std over 4 seeds2026.04 | 89.9 | 0.101 | 90.7 | 9.8 | |
| Mao25Number of model experts (J)=5, Evaluation protocol=mean±std over 4 seeds2026.04 | 87.7 | 0.123 | 35.1 | 0 |