Backdoor Detection on Run-To-Go-Human (RTGH)
0.904AUROCPolicyGuard
Evaluation Results
| Method | Links | |
|---|---|---|
| PolicyGuardStrategy=Hard-coded2026.06 | 0.904 | |
| PolicyGuardStrategy=Original2026.06 | 0.895 | |
| SHINEStrategy=Hard-coded2026.06 | 0.796 | |
| SHINEStrategy=Original2026.06 | 0.789 | |
| PolicyCleanseStrategy=Original2026.06 | 0.759 | |
| PolicyCleanseStrategy=Hard-coded2026.06 | 0.502 |