Research map / Probabilistic and scientific modeling / Causal discovery
Interventional causal discovery
Use interventional data and targeting strategies to learn causal graphs efficiently.
68 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in Causal discovery, keyword cluster “causal discovery · causal structure · graphs”
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.
21 papers in 2023–24 and 29 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 9.1% to 8.7%.
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
- 2018 · SELF: Structural Equational Likelihood Framework for Causal Discovery (AAAI 2018 · 27 citations)
- 2019 · Structural Causal Bandits with Non-Manipulable Variables (AAAI 2019 · 24 citations)
- 2020 · Causal Discovery in Physical Systems from Videos (NeurIPS 2020 · 43 citations)
- 2021 · Systematic Evaluation of Causal Discovery in Visual Model Based Reinforcement Learning (NeurIPS 2021 · 5 citations)
- 2022 · Large-Scale Differentiable Causal Discovery of Factor Graphs (NeurIPS 2022 · 9 citations)
- 2023 · Rhino: Deep Causal Temporal Relationship Learning with History-dependent Noise (ICLR 2023 · 4 citations)
- 2024 · Multi-Granularity Causal Structure Learning (AAAI 2024 · 5 citations)
- 2025 · Identifying biological perturbation targets through causal differential networks (ICML 2025 · 1 citation)
- 2026 · Horizontal and Vertical Federated Causal Structure Learning via Higher-order Cumulants (AAAI 2026 · 2 citations)
Earliest papers here
- SELF: Structural Equational Likelihood Framework for Causal Discovery (AAAI 2018 · 27 citations)
- Structural Causal Bandits with Non-Manipulable Variables (AAAI 2019 · 24 citations)
- Characterization and Learning of Causal Graphs with Latent Variables from Soft Interventions (NeurIPS 2019 · 23 citations)
What has been tried
48 papers here report 359 measured changes (a component added, removed or swapped, compared with a baseline): helped 63%, no clear effect 11%, hurt 17%, mixed 8%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add the proposed simplified acyclicity constraint to NOTEARS-MLP | Nonlinear models with additive Gaussian processes, d=50, 100, 200 · F1 score | Hurt | Multi-Granularity Causal Structure Learning (AAAI 2024 · 5 citations) | |
| add temporal causal discovery | ACL18, CMIN-US, CMIN-CN · ACC | 51.08 vs 63.42; 51.48 vs 54.64; 51.37 vs 56.19 | Helped | CausalStock: Deep End-to-end Causal Discovery for News-driven Multi-stock Movement Prediction (NeurIPS 2024 · 5 citations) |
| finetune CDN on Perturb-seq data rather than train on synthetic data only | Perturb-seq datasets, seen and unseen cell lines · rank | Helped | Identifying biological perturbation targets through causal differential networks (ICML 2025 · 1 citation) | |
| Use MHSA + MHCA attention among samples rather than masked-MHSA | MACE-TNP architecture experiments · NLPID | Helped | Estimating Interventional Distributions with Uncertain Causal Graphs through Meta-Learning (NeurIPS 2025 · 1 citation) | |
| Use a non-parametric CIT instead of a parametric CIT | Simulated graphs without cycles, no link assumptions · false positive rate | Helped | Causal discovery with endogenous context variables (NeurIPS 2024 · 1 citation) | |
| Treat Z1 as latent | Synthetic 18-node DAG · WCDE | 1.70 [1.66,1.75] | Hurt | Local Causal Discovery for Structural Evidence of Direct Discrimination (AAAI 2025 · 0 citations) |
| Remove all first-order effects for nodes with multiple parents, retaining only the highest-order interaction | Small DAG simulation (d<10), ablated first-order effects · SHD | No clear effect | Differentiable Structure Learning and Causal Discovery for General Binary Data (NeurIPS 2025 · 0 citations) | |
| Use OOCDM instead of OOFULL, which uses the full OOCG | Mouse o.o.d. · average instance log-likelihood | -1.5e9 ± 3.0e9 -> 11.2 ± 17.2 | Helped | Learning Causal Dynamics Models in Object-Oriented Environments (ICML 2024 · 0 citations) |
| lower the cutoff ratio from 0.30 to 0.01 | Protein Signaling · Jaccard similarity | 0.00 -> 0.07 | Helped | Learning Mixtures of Unknown Causal Interventions (NeurIPS 2024 · 0 citations) |
| Add multi-step optimization | Bivariate causal discovery, Setting-1 and Setting-2 · performance | Helped | Bayesian Active Learning for Bivariate Causal Discovery (ICML 2025) |
What reviewers will expect
Datasets and baselines named in the 21 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Sachs · 3 of 21 papers
- ANDES · 2 of 21 papers
Baselines
- PC · 7 of 21 papers
- NOTEARS · 4 of 21 papers
- DCDI · 4 of 21 papers
- DAGMA · 3 of 21 papers
- AVICI · 3 of 21 papers
- GIES · 3 of 21 papers
- GES · 2 of 21 papers
- LDECC · 2 of 21 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 50 papers; from OpenAlex and the papers' affiliations).
- Purdue University · 5 papers
- Columbia University · 4 papers
- University of Oxford · 3 papers
- Microsoft · 3 papers
- Mohamed bin Zayed University of Artificial Intelligence · 3 papers
- Peking University · 3 papers
- Massachusetts Institute of Technology · 3 papers
- Booth School of Business · 3 papers
- Shandong University · 2 papers
- Amazon · 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
- Constraint-based structure learning with latents (90 papers)
- Identifiability with hidden confounders (85 papers)
- Differentiable and score-based DAG learning (52 papers)
- Causal discovery from time series (51 papers)
Elsewhere in the topic
- Causal representation identifiability (122 papers)
- Latent-variable causal estimation (117 papers)
- Causal effect identification (107 papers)
- Counterfactual identification and bounds (92 papers)
- Causal approaches to fairness (46 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).