Research map / Probabilistic and scientific modeling / Causal discovery
Front-door Adjustment Beyond Markov Equivalence with Limited Graph Knowledge
Abhin Shah, Karthikeyan Shanmugam, Murat Kocaoglu
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 |
|---|---|---|---|
| search z of size at most 5 instead of exhaustive search (ablation) | front-door-like adjustment search · Random SMCMs, 100 graphs | successes out of 100: 43 -> 43 | No clear effect |
| restricted search at p=15, d=4, q=0.0 (ablation) | front-door-like adjustment search · Random SMCMs p=15 | successes out of 100: 4 -> 2 | Hurt |
| use the proposed adjustment instead of IDP on the true PAG | effect identification · Random SMCMs | successes out of 100: 43 -> 0 | Improved |
| increase hidden-confounder parameter q from 0.0 to 1.0 | proposed method · Random SMCMs p=10, d=2 | successes out of 100: 43 -> 6 | Hurt |
| increase graph size from p=10 to p=15 at q=0.0, d=2 | proposed method · Random SMCMs | successes out of 100: 43 -> 27 | Hurt |
| use proposed adjustment rather than the baseline front-door formula with z=b at n=1000 and 10000 | ATE estimation · Synthetic SMCMs p=10, d=2, q=1.0 | ATE error | Improved |
| use the proposed adjustment instead of the baseline at n=100 | ATE estimation · Synthetic SMCMs | ATE error | No clear effect |
| use the standard front-door formula with z=b when b is not the full mediator set | standard front-door adjustment · synthetic SEM | ATE error: 0.3842 ± 0.0207 | Hurt |
| remove applicant's job from the assumed children b (ablation) | adjustment on German Credit · German Credit | ATE: 0.0105 -> -0.0046 | Mixed |
| include job in children set b (ablation) | adjustment on German Credit · German Credit | ATE_z vs ATE_s agreement: 0.0125 vs 0.0105 | Improved |
| apply adjustment to Adult dataset | proposed method · Adult | finding valid z | No clear effect |
Closest papers by meaning
- Linear-Time Algorithms for Front-Door Adjustment in Causal Graphs (AAAI 2024 · similarity 0.92)
- Causal Inference with Conditional Front-Door Adjustment and Identifiable Variational Autoencoder (ICLR 2024 · similarity 0.88)
- Unified Covariate Adjustment for Causal Inference (NeurIPS 2024 · similarity 0.88)
- Complete Graphical Criterion for Sequential Covariate Adjustment in Causal Inference (NeurIPS 2024 · similarity 0.87)
- Interventional Fairness on Partially Known Causal Graphs: A Constrained Optimization Approach (ICLR 2024 · similarity 0.87)
- Graph Distance Based on Cause-Effect Estimands with Latents (NeurIPS 2026 · 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.
