AI Paper Atlas

Research map / Probabilistic and scientific modeling / Treatment effects

Stochastic Neural Networks for Causal Inference with Missing Confounders

Yaxin Fang, Faming Liang

ICLR 2026 · Probabilistic and scientific modeling › Treatment effects · approach: Proximal and deep causal effect estimation · 2 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
use CI-StoNet instead of DragonNetDragonNet · Simulated proxy data (10 datasets)out-of-sample PEHE: 0.4305 -> 0.3731Improved
use CI-StoNet instead of CEVAECEVAE · Simulated proxy data (10 datasets)out-of-sample PEHE: 0.6246 -> 0.3731Improved
use CI-StoNet instead of X-learner-BARTX-learner-BART · ACIC 2019 (10 datasets)out-of-sample ATE MAE: 0.0817 -> 0.0709Improved
use CI-StoNet instead of CMDECMDE · ACIC 2019 (10 datasets)out-of-sample ATE MAE: 0.0877 -> 0.0709Improved
use CI-StoNet instead of DSE when confounder gestat10 is missingDSE · Twins (gestat10 missing)absolute error of ATE: 0.0211 -> 0.0135Improved
use CI-StoNet instead of X-learner-RF when gestat10 observedX-learner-RF · Twins (with gestat10)absolute error of ATE: 0.0056 -> 0.0099Hurt
use CI-StoNet instead of X-learner-RF when gestat10 observed (RMSE)X-learner-RF · Twins (with gestat10)RMSE of ATE: 0.0252 -> 0.0133Improved
remove confounder gestat10 from Twins (ablation)CI-StoNet · Twinsabsolute error of ATE: 0.0099 -> 0.0135Hurt
misspecify DAG as outcome-depending proxy (ablation)CI-StoNet with basic proxy structure · DAG misspecification simulationout-of-sample ATE MAE: 0.0702 -> 0.0976Hurt
misspecify DAG as treatment-depending proxy (ablation)CI-StoNet with basic proxy structure · DAG misspecification simulationout-of-sample ATE MAE: 0.0702 -> 0.1239Hurt
misspecify DAG as outcome-depending proxy, in-sample (ablation)CI-StoNet with basic proxy structure · DAG misspecification simulationin-sample ATE MAE: 0.0682 -> 0.0705No clear effect

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