Research map / Probabilistic and scientific modeling / Causal discovery
Recovering Causal Structures from Low-Order Conditional Independencies
Marcel Wienöbst, Maciej Liskiewicz
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 |
|---|---|---|---|
| Use LOCI to remove edges between incompatible nodes from the 0-1 graph | Low-order CI causal structure estimation · Random DAGs, n=100 and expected degree d=3 | number of edges: 413{.}68 (0-1 graph) -> 266{.}92 (LOCI); 149{.}42 (true DAG) | Improved |
| Use LOCI to remove edges between incompatible nodes from the 0-1 graph | Low-order CI causal structure estimation · Random DAGs, n=100 and expected degree d=4 | number of edges: 1{,}061{.}39 (0-1 graph) -> 598{.}11 (LOCI); 199{.}34 (true DAG) | Improved |
| Use LOCI to remove edges between incompatible nodes from the 0-1 graph | Low-order CI causal structure estimation · Random DAGs, n=100 and expected degree d=2 | number of edges: 125{.}87 (0-1 graph) -> 113{.}37 (LOCI); 99{.}40 (true DAG) | Improved |
| Use LOCI to remove edges between incompatible nodes from the 0-1 graph | Low-order CI causal structure estimation · Random DAGs, n=60 and expected degree d=3 | number of edges: 226{.}03 (0-1 graph) -> 160{.}43 (LOCI); 88{.}55 (true DAG) | Improved |
| Use LOCI's orientations from low-order CIs | LOCI representation G · Random DAGs, n=100 and expected degree d=5 | shared v-structures per node: 3{.}872 (true DAG); 3{.}504 (in both) | Improved |
| Represent the model with the full LOCI graph, including its extra edges and orientations | LOCI representation G · Random DAGs, n=100 and expected degree d=5 | v-structures per node: 161{.}390 (LOCI); 3{.}872 (true DAG) | Hurt |
| Rely only on observed zero- and first-order CIs to remove corresponding edges | 0-1 graph · Random DAGs, n=100 and expected degree d=3 | number of edges: 413{.}68 (0-1 graph); 266{.}92 (LOCI); 149{.}42 (true DAG) | Hurt |
| Require prior knowledge of the underlying DAG's topological order | Campos and Huete method · Low-order CI causal structure learning | availability of a causal representation | Hurt |
Closest papers by meaning
- Testing Causal Models with Hidden Variables in Polynomial Delay via Conditional Independencies (AAAI 2025 · similarity 0.89)
- Sparse Additive Model Pruning for Order-Based Causal Structure Learning (AAAI 2026 · similarity 0.89)
- Causal Discovery with Fewer Conditional Independence Tests (ICML 2024 · similarity 0.89)
- Scores for Learning Discrete Causal Graphs with Unobserved Confounders (AAAI 2024 · similarity 0.88)
- Coarsening Linear Non-Gaussian Causal Models with Cycles (NeurIPS 2026 · similarity 0.88)
- Learning Structural Causal Models from Ordering: Identifiable Flow Models (AAAI 2025 · similarity 0.88)
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.
