Research map / Graph learning / Graph algorithms and clustering
Direct Estimation of Differences in Causal Graphs
Yuhao Wang, Chandler Squires, Anastasiya Belyaeva, Caroline Uhler
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 |
|---|---|---|---|
| Estimate the difference graph directly instead of subtracting separately estimated DAGs | DCI versus separate estimation with PC or GES · Synthetic low-dimensional, p=10, s=3 | exact recovery rate (D-DAG) | Improved |
| Use PC rather than GES for the separate-estimation baseline | Separate DAG estimation · Synthetic low-dimensional, p=10, s=3 | exact recovery rate (D-DAG) | Improved |
| Initialize DCI with the complete graph rather than the constraint-based initialization (ablation) | DCI · Synthetic low-dimensional, p=10, s=3 | exact recovery rate (D-DAG) | No clear effect |
| Initialize DCI with KLIEP rather than the constraint-based initialization (ablation) | DCI · Synthetic high-dimensional, p=100, s=10, n=300, 5% change | ROC performance (D-DAG) | No clear effect |
| Initialize DCI with KLIEP rather than separate DAG estimation | DCI-K versus PC and GES · Synthetic high-dimensional, p=100, s=10, n=300, 5% change | ROC performance (D-DAG) | Improved |
| Increase the number of nodes with changed noise variances | DCI · Synthetic low-dimensional, p=10, s=3 | number of edges with determined orientation | Hurt |
| Change noise variances across the two models | DCI · Synthetic low-dimensional, p=10, s=3 | accuracy | No clear effect |
Closest papers by meaning
- Estimating Causal Structure Using Conditional DAG Models (JMLR 2016 · similarity 0.91)
- Counting and Sampling from Markov Equivalent DAGs Using Clique Trees (AAAI 2019 · similarity 0.90)
- Interventional Causal Discovery in a Mixture of DAGs (NeurIPS 2024 · similarity 0.89)
- Scalable and Flexible Causal Discovery with an Efficient Test for Adjacency (ICML 2024 · similarity 0.88)
- Coarsening Linear Non-Gaussian Causal Models with Cycles (NeurIPS 2026 · similarity 0.87)
- Differentiable Cyclic Causal Discovery Under Unmeasured Confounders (NeurIPS 2025 · similarity 0.87)
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.
