Causal Discovery on Asia synthetic averaged over n={500, 1000, 5000, 10000}
0.641RNBGRaSP
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GRaSP2026.06 | 0.641 | 0.497 | |
| LLM-Greedybackbone=GPT-5.52026.06 | 0.625 | 0.414 | |
| BOSS2026.06 | 0.531 | 0.645 | |
| FCI2026.06 | 0.422 | 0.588 | |
| PC2026.06 | 0.422 | 0.588 | |
| DAGBag2026.06 | 0.406 | 0.599 | |
| ICA-LiNGAM2026.06 | 0.391 | 0.628 | |
| Uniform CEAmode=uniform weights2026.06 | 0.281 | 0.732 | |
| GES2026.06 | 0.219 | 0.806 | |
| LLM-BFSbackbone=GPT-5.52026.06 | 0.188 | 0.834 | |
| CEAbackbone=GPT-5.52026.06 | 0.093 | 0.918 |