Image Classification (Spurious Correlation Robustness) on CelebA
89.3Worst Group Accuracy (WGA)Group DRO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Group DROuses training group labels=true, uses validation group labels=true, inference-time interpretability=false2026.06 | 89.3 | 92.8 | |
| CnCuses training group labels=false, uses validation group labels=true, inference-time interpretability=false2026.06 | 88.8 | 89.9 | |
| DFRuses training group labels=false, uses validation group labels=true, inference-time interpretability=false2026.06 | 86.9 | 91.1 | |
| NCTuses training group labels=false, uses validation group labels=false, inference-time interpretability=true2026.06 | 86.1 | 88.4 | |
| ExMapuses training group labels=false, uses validation group labels=false, inference-time interpretability=false2026.06 | 84.4 | 91.8 | |
| EIILuses training group labels=false, uses validation group labels=true, inference-time interpretability=false2026.06 | 81.3 | 89.5 | |
| JTTuses training group labels=false, uses validation group labels=true, inference-time interpretability=false2026.06 | 81.1 | 88 | |
| GEORGEuses training group labels=false, uses validation group labels=false, inference-time interpretability=false2026.06 | 53.7 | 94.6 | |
| ERMuses training group labels=false, uses validation group labels=false, inference-time interpretability=false2026.06 | 40.3 | 95.7 |