AI Paper Atlas

Research map / Probabilistic and scientific modeling / Causal discovery

Causal Reasoning in the Presence of Latent Confounders via Neural ADMG Learning

Matthew Ashman, Chao Ma, Agrin Hilmkil, Joel Jennings, Cheng Zhang

ICLR 2023 · Probabilistic and scientific modeling › Causal discovery · approach: Differentiable and score-based DAG learning · 3 citations (OpenAlex)

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 changedWhereMeasuredEffect
Add the bow-free constraint (ablation)N-ADMG-G · Synthetic Fork-ColliderATE RMSE: 0.239 (0.067) -> 0.022 (0.003)Improved
Add the bow-free constraint (ablation)N-ADMG-G · Synthetic Fork-Colliderdirected G_D Fscore: 0.49 (0.02) -> 0.64 (0.06)Improved
Add the bow-free constraint (ablation)N-ADMG-G · Synthetic Fork-Colliderbidirected G_B Fscore: 0.99 (0.00) -> 0.93 (0.07)Hurt
Use N-ADMG rather than a method assuming no latent confoundersN-DAG-G · Synthetic Fork-Colliderbidirected G_B Fscore: 0.00 (0.00) -> 0.93 (0.07)Improved
Use a nonlinear model rather than linear ANM methodsCausal discovery and inference methods · Synthetic Fork-ColliderATE RMSEMixed
Use N-BF-ADMG with non-Gaussian exogenous assumptionsN-BF-ADMG · IHDPATE RMSEImproved
Use DoWhy plug-in estimators instead of the SCM inference procedureSCM models · IHDPATE RMSEHurt
Use linear assumptionsCausal inference methods · IHDPATE RMSEHurt
Use N-ADMG variants rather than the compared baselinesN-ADMG · Random Confounded ER synthetic datasetscausal discovery performanceImproved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy