Causal Discovery on Synthetic Scale-Free (SF) networks d=10
6.56SHDDCD-PFN
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DCD-PFNdimension (d)=10, zero-shot=true, number of samples (n)=8002026.06 | 6.56 | 94.9 | 28.9 | 43.1 | |
| AVICIdimension (d)=10, zero-shot=true, number of samples (n)=8002026.06 | 7.44 | 69.5 | 17.3 | 26.4 | |
| ICALiNGAMdimension (d)=10, number of samples (n)=8002026.06 | 7.56 | 80 | 16 | 25.7 | |
| DirectLiNGAMdimension (d)=10, number of samples (n)=8002026.06 | 7.56 | 76 | 16 | 25.5 | |
| GOLEMdimension (d)=10, number of samples (n)=8002026.06 | 7.72 | 80 | 14.2 | 23.4 | |
| PC (E-KCIT)dimension (d)=10, number of samples (n)=8002026.06 | 7.96 | 21.2 | 14.2 | 16.7 | |
| Notearsdimension (d)=10, number of samples (n)=8002026.06 | 8.24 | 60 | 8.4 | 14.5 | |
| GESdimension (d)=10, number of samples (n)=8002026.06 | 8.64 | 23 | 17.3 | 19.6 | |
| PC (KCIT)dimension (d)=10, number of samples (n)=8002026.06 | 8.68 | 15.2 | 12 | 13.3 | |
| GraNDAGdimension (d)=10, number of samples (n)=8002026.06 | 13.32 | 32.3 | 15.1 | 17.4 |