Research map / Probabilistic and scientific modeling / Treatment effects
General Control Functions for Causal Effect Estimation from IVs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| replace CFN's additive-outcome assumption with general control function (GCFN) | CFN · Multiplicative outcome & additive treatment | RMSE: 0.58 ± 0.01 -> 0.09 ± 0.06 | Helped |
| GCFN vs DeepIV with multiplicative treatment process | DeepIV · Multiplicative treatment & additive outcome | RMSE: 0.58 ± 0.01 -> 0.13 ± 0.08 | Helped |
| GCFN vs 2SLS on multiplicative outcome simulation | 2SLS · Multiplicative outcome & additive treatment | RMSE: 0.55 ± 0.58 -> 0.09 ± 0.06 | Helped |
| add confounder observed on a subset (semi-supervised GCFN) vs supervised regression baseline | Supervised regression baseline · Partial confounder observation | RMSE of causal effect | Helped |
| GCFN vs DeepIV on high-dimensional covariates, 10,000 samples | DeepIV · DeepIV simulation, 10,000 samples | effect MSE range: 0.30 -0.50 -> 0.30 -0.42 | Helped |
| GCFN vs DeepIV, 20,000 samples | DeepIV · DeepIV simulation, 20,000 samples | effect MSE: 0.32 -> 0.305 | Helped |
| vary mutual-information coefficient kappa (0.1, 0.2, 0.3) (ablation) | GCFN · DeepIV simulation | effect MSE | No clear effect |
| GCFN vs DeepGMM on high-dimensional IV | DeepGMM · MNIST IV | effect MSE: 0.07 ± 0.02 -> 0.077 ± 0.022 | No clear effect |
| GCFN vs DeepIV on high-dimensional IV | DeepIV · MNIST IV | effect MSE: 0.11 ± 0.00 -> 0.077 ± 0.022 | Helped |
| vary kappa in {0.2, 0.4} vs 0.3 (ablation) | GCFN · MNIST IV | effect MSE | No clear effect |
| increase outcome network width beyond 10 hidden units (ablation) | GCFN · Slave export on trust | overfitting | Hurt |
| GCFN vs published linear IV estimate | Nunn and Wantchekon linear IV · Slave export on trust | estimate of beta: -0.27 ± 0.10 -> -0.21 ± 0.04 | Mixed |
Closest papers by meaning
- Doubly robust identification of treatment effects from multiple environments (ICLR 2025 · similarity 0.85)
- Generalized Bayes for Causal Inference (NeurIPS 2026 · similarity 0.84)
- Do Finetti: On Causal Effects for Exchangeable Data (NeurIPS 2024 · similarity 0.84)
- Unified Covariate Adjustment for Causal Inference (NeurIPS 2024 · similarity 0.84)
- An Analysis of Causal Effect Estimation using Outcome Invariant Data Augmentation (NeurIPS 2025 · similarity 0.84)
- Causal Effect Identification in lvLiNGAM from Higher-Order Cumulants (ICML 2025 · similarity 0.84)
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.