Research map / Probabilistic and scientific modeling / Causal discovery
dcFCI: Robust Causal Discovery Under Latent Confounding, Unfaithfulness, and Mixed Data
Adèle H. Ribeiro, Dominik Heider
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase number of kept compatible PAGs from k=1 to k=2 (ablation) | dcFCI · Gaussian, N=10,000 | True PAG recovery rate (%): 74.7 -> 80.3 | Improved |
| increase k from 1 to 2 (ablation) | dcFCI · Gaussian, N=50,000 | True PAG recovery rate (%): 98.3 -> 98.67 | No clear effect |
| use data-PAG compatibility score instead of FCI orientation | dcFCI vs FCI · Mixed data, N=1,000 | True PAG recovery rate (%): 5.67 -> 15.67 | Improved |
| use dcFCI instead of FCI | dcFCI vs FCI · Mixed data, N=5,000 | True PAG recovery rate (%): 39.67 -> 55 | Improved |
| use dcFCI instead of cFCI for validity of output PAGs | dcFCI vs cFCI · Gaussian, N=1,000 | Valid PAGs (%): 46.7 -> 100.0 | Improved |
| use dcFCI instead of BCCD | dcFCI vs BCCD · Gaussian, N=50,000 | SHD (mean): 6.7 -> 0.05 | Improved |
| use dcFCI instead of FCI | dcFCI vs FCI · Gaussian, N=10,000 | SHD (mean): 1.7 -> 1 | Improved |
| use dcFCI instead of MAGSL | dcFCI vs MAGSL · Gaussian, N=5,000 | SHD (mean): 3.4 -> 3.5 | No clear effect |
| use dcFCI instead of FCI | dcFCI vs FCI · Mixed data, N=1,000 | SHD (mean): 5.3 -> 4.4 | Improved |
| use dcFCI instead of DCD | dcFCI vs DCD · Gaussian, N=1,000 | SHD (mean): 8 -> 5.6 | Improved |
| increase number of variables from five to six | dcFCI · Diabetes Health Indicators real data | PAG reliability | Hurt |
Closest papers by meaning
- Large-Scale Differentiable Causal Discovery of Factor Graphs (NeurIPS 2022 · similarity 0.91)
- Local Identifying Causal Relations in the Presence of Latent Variables (ICML 2025 · similarity 0.90)
- Differentiable Cyclic Causal Discovery Under Unmeasured Confounders (NeurIPS 2025 · similarity 0.90)
- Identifying Structural Biases from Causal Mechanism Shifts (NeurIPS 2026 · similarity 0.89)
- Differentiable Causal Discovery for Latent Hierarchical Causal Models (ICLR 2025 · similarity 0.89)
- CausalTab: Pretraining Across Causal Environments for Tabular Causal Discovery (NeurIPS 2026 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
