Diffusion
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Fake Image Detection | Diffusion | Accuracy95.3 | 20 | |
| Recurrence Core Coverage Prediction | DIFFUSION | Coverage86.26 | 16 | |
| Relational Inference | Diffusion WS50 low-data | AUC98.21 | 7 | |
| Relational Inference | Diffusion ER50 low-data | AUC81.72 | 7 | |
| Relational Inference | Diffusion (DIFF) BA50 low-data | AUC91.98 | 7 | |
| Relational Inference | Diffusion (DIFF) BA-50 | AUC96.55 | 7 | |
| Relational Inference | Diffusion (DIFF) ER-50 | AUC99.36 | 7 | |
| Information Extraction | Diffusion 10-PDF subsample | F1 Score0.0871 | 6 | |
| Information Extraction | Diffusion | Precision27.51 | 6 | |
| AI-generated image detection | Diffusion 6-cls (test) | Dall-E Accuracy98.8 | 5 | |
| AI-generated image detection | Diffusion 6-cls LDM-100 (test) | Accuracy98.8 | 5 | |
| AI-generated image detection | Diffusion 6-cls (Glide-50-27) (test) | Accuracy97.2 | 5 | |
| AI-generated image detection | Diffusion 6-cls PNDM (test) | Accuracy99.3 | 5 | |
| AI-generated image detection | Diffusion-6-cls Guided (test) | Accuracy92.9 | 5 | |
| AI-generated image detection | Diffusion-6-cls Dall-E (test) | Accuracy98.8 | 5 | |
| Inverse reconstruction | Diffusion N=500 per family (test) | Relative L2 Error0.45 | 4 | |
| Forward prediction | Diffusion N=500 per family (test) | Relative L2 Error55 | 4 | |
| Scientific Information Extraction | Diffusion (full) | Precision33.51 | 4 | |
| Membership Inference Attack | 32nm Diffusion | Accuracy93 | 2 |
Showing 19 of 19 rows