Failure Detection on Close Box In-Distribution (train test)
1TPRTemporal Non-Distr. Min.
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Temporal Non-Distr. Min.Category=Diffusion2026.03 | 1 | 0.97 | 5 | |
| Temporal Diffusion Recon.Category=Diffusion2026.03 | 1 | 0.96 | 8.47 | |
| DDPM LossCategory=Diffusion2026.03 | 1 | 0.9 | 8.27 | |
| Temporal DDPM LossCategory=Diffusion2026.03 | 1 | 0.95 | 7.53 | |
| CLIP PretrainedCategory=Embed.2026.03 | 1 | 0.95 | 15.73 | |
| ResNet PretrainedCategory=Embed.2026.03 | 1 | 0.95 | 17.87 | |
| STAC For. KLCategory=STAC2026.03 | 1 | 0.9 | 6.6 | |
| STAC Rev. KLCategory=STAC2026.03 | 1 | 0.95 | 7.6 | |
| STAC MMD*Category=STAC2026.03 | 1 | 0.94 | 7.2 | |
| GPT-4o Image QACategory=VLM2026.03 | 1 | 0 | 23.2 | |
| GPT-4o Video QA*Category=VLM2026.03 | 1 | 0.89 | 21.2 | |
| Sentinel (STAC MMD* + GPT-4o Video QA*)Configuration=Parallel Ensemble2026.03 | 1 | 0.86 | 5.47 | |
| Diffusion Recon.Category=Diffusion2026.03 | 0.33 | 0.95 | 13.6 | |
| Diffusion Output VarianceCategory=Diffusion2026.03 | 0.33 | 0.94 | 14 | |
| Policy EncoderCategory=Embed.2026.03 | 0.25 | 0.98 | 16.27 |