Research map / Probabilistic and scientific modeling / Treatment effects
Meta-learners and partial identification
Estimates conditional treatment effects with proximal learning and bounds under partial identification.
78 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2019–2026) · an approach in Treatment effects, keyword cluster “treatment effects · heterogeneous treatment · average treatment”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
28 papers in 2023–24 and 35 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 21.4% to 21.3%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2019 · Machine Learning Estimation of Heterogeneous Treatment Effects with Instruments (NeurIPS 2019 · 32 citations)
- 2020 · Minimax Optimal Nonparametric Estimation of Heterogeneous Treatment Effects (NeurIPS 2020 · 6 citations)
- 2021 · Treatment Effect Estimation with Disentangled Latent Factors (AAAI 2021 · 62 citations)
- 2022 · A Tree-based Model Averaging Approach for Personalized Treatment Effect Estimation from Heterogeneous Data Sources (ICML 2022 · 9 citations)
- 2023 · Optimal Transport for Treatment Effect Estimation (NeurIPS 2023 · 7 citations)
- 2024 · Exploiting Geometry for Treatment Effect Estimation via Optimal Transport (AAAI 2024 · 3 citations)
- 2025 · Conformal Inference of Individual Treatment Effects Using Conditional Density Estimates (AAAI 2025 · 0 citations)
- 2026 · Conditional Distributional Treatment Effects: Doubly Robust Estimation and Testing (ICML 2026 · 0 citations)
Earliest papers here
- Machine Learning Estimation of Heterogeneous Treatment Effects with Instruments (NeurIPS 2019 · 32 citations)
- Debiased Bayesian inference for average treatment effects (NeurIPS 2019 · 5 citations)
- Minimax Optimal Nonparametric Estimation of Heterogeneous Treatment Effects (NeurIPS 2020 · 6 citations)
What has been tried
54 papers here report 458 measured changes (a component added, removed or swapped, compared with a baseline): helped 68%, no clear effect 9%, hurt 17%, mixed 6%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add inter-group geometry to the method using only intra-group geometry | Toy example · MAE | .355 -> .347 | Helped | Exploiting Geometry for Treatment Effect Estimation via Optimal Transport (AAAI 2024 · 3 citations) |
| Use anisotropic covariance informed by prior knowledge that X1 is the treatment effect modifier instead of isotropic covariance | Synthetic prior-knowledge experiment, n=100 · MSE | 0.240± 0.0745 -> 0.185± 0.188 | Helped | Uncertainty Quantification in Heterogeneous Treatment Effect Estimation with Gaussian-Process-Based Partially Linear Model (AAAI 2024 · 2 citations) |
| Add higher-order treatment and outcome-variance features | Non-LinearInMeans · TTE estimation error | Helped | Higher-Order Causal Message Passing for Experimentation with Complex Interference (NeurIPS 2024 · 1 citation) | |
| Residualize the outcome using a function of covariates instead of using unresidualized IV-DL | Simulation Setting 1 · MSE | 55.5(3.8) -> 40.5(2.9) | Helped | A Non-parametric Direct Learning Approach to Heterogeneous Treatment Effect Estimation under Unmeasured Confounding (NeurIPS 2024 · 1 citation) |
| Vary the number of discretizations from k=2 to k=3 | Dataset 1 · width | 1.62 ± 0.06 -> 0.83 ± 0.16 | Helped | Learning Representations of Instruments for Partial Identification of Treatment Effects (ICML 2025 · 0 citations) |
| Remove redundant backdoor variable B2 from {G1, G2, B2} | Synthetic Fig [ref], n=250 · variance | 0.0050 -> 0.0029 | Helped | Optimal Adjustment Sets for Nonparametric Estimation of Weighted Controlled Direct Effect (NeurIPS 2025 · 0 citations) |
| increase the number of policy re-allocations | IV, Conditional IV, and TripAdvisor · MSE | Helped | Adaptive Instrument Design for Indirect Experiments (ICLR 2024 · 0 citations) | |
| Use the T-Learner rather than the S-Learner to construct the score | ACIC 2016, LaLonde CPS, LaLonde PSID, and TWINS · Normalized PEHE | Helped | Empirical Analysis of Model Selection for Heterogeneous Causal Effect Estimation (ICLR 2024 · 0 citations) | |
| Use W2² rather than CRPS as the AU-learner target risk | HC-MNIST, upper bound · rCRPSout | 0.347 -> 0.422 | Hurt | Quantifying Aleatoric Uncertainty of the Treatment Effect: A Novel Orthogonal Learner (NeurIPS 2024 · 0 citations) |
| make the CCDF contrasting function constant as well as the CQC | Simulated experiment with uniform responses · mean absolute error | No clear effect | Conditional Outcome Equivalence: A Quantile Alternative to CATE (NeurIPS 2024 · 0 citations) | |
| Use fifth-order ACE rather than lower-order ACE estimators | Synthetic demand estimation · total RMSE | Helped | It’s Hard to Be Normal: The Impact of Noise on Structure-agnostic Estimation (NeurIPS 2025) | |
| Use a marginal rather than joint transformer when control covariates are independent | Uniform-transformer simulation, rho=0 · RMSE | 0.49 -> 0.47; 0.56 -> 0.51 | Mixed | Treatment Effects Estimation By Uniform Transformer (ICLR 2024) |
What reviewers will expect
Datasets and baselines named in the 19 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- IHDP · 3 of 19 papers
- simulated data · 2 of 19 papers
Baselines
- Causal Forest · 2 of 19 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 63 papers; from OpenAlex and the papers' affiliations).
- Ludwig-Maximilians-Universität München · 6 papers
- LMU Munich · 6 papers
- Stanford University · 6 papers
- Peking University · 3 papers
- Carnegie Mellon University · 3 papers
- Guangdong University of Technology · 2 papers
- Ant Group · 2 papers
- Zhejiang University · 2 papers
- North Carolina State University · 2 papers
- Princeton University · 2 papers
Use it
Download BibTeX of 37 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Benchmarking treatment effect models (77 papers)
- Distributional treatment effects (51 papers)
Elsewhere in the topic
- Proximal and deep causal effect estimation (105 papers)
- Latent-variable causal inference (91 papers)
- Instrumental variable estimation (19 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).