Research map / Probabilistic and scientific modeling / Causal discovery
Latent-variable causal estimation
Learn structural causal models and deconfounded estimates using generative and latent-variable models.
117 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · an approach in Causal discovery, keyword cluster “counterfactual · causal inference · causal representation”
State History Tried 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.
29 papers in 2023–24 and 44 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 12.5% to 13.2%.
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
- 2017 · Entropic Causal Inference (AAAI 2017 · 40 citations)
- 2018 · Learning and Testing Causal Models with Interventions (NeurIPS 2018 · 22 citations)
- 2019 · Neural Network Attributions: A Causal Perspective (ICML 2019 · 32 citations)
- 2020 · A Causal View on Robustness of Neural Networks (NeurIPS 2020 · 41 citations)
- 2021 · Recovering Latent Causal Factor for Generalization to Distributional Shifts (NeurIPS 2021 · 19 citations)
- 2022 · VACA: Designing Variational Graph Autoencoders for Causal Queries (AAAI 2022 · 11 citations)
- 2023 · GaitGCI: Generative Counterfactual Intervention for Gait Recognition (CVPR 2023 · 70 citations)
- 2024 · Drift-Resilient TabPFN: In-Context Learning Temporal Distribution Shifts on Tabular Data (NeurIPS 2024 · 12 citations)
- 2025 · Image Quality Assessment: Investigating Causal Perceptual Effects with Abductive Counterfactual Inference (CVPR 2025 · 25 citations)
- 2026 · CausalProfiler: Generating Synthetic Benchmarks for Rigorous and Transparent Evaluation of Causal Machine Learning (ICML 2026 · 0 citations)
Earliest papers here
- Entropic Causal Inference (AAAI 2017 · 40 citations)
- GP CaKe: Effective brain connectivity with causal kernels (NeurIPS 2017 · 4 citations)
- Learning and Testing Causal Models with Interventions (NeurIPS 2018 · 22 citations)
What has been tried
76 papers here report 635 measured changes (a component added, removed or swapped, compared with a baseline): helped 65%, no clear effect 12%, hurt 18%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| replace original pretrained weights with causal-effect weights | LIVE · PLCC | 0.901 -> 0.929 | Helped | Image Quality Assessment: Investigating Causal Perceptual Effects with Abductive Counterfactual Inference (CVPR 2025 · 25 citations) |
| Add Time2Vec preprocessing | Synthetic + real-world datasets, OOD · accuracy | 0.742 -> 0.744 | No clear effect | Drift-Resilient TabPFN: In-Context Learning Temporal Distribution Shifts on Tabular Data (NeurIPS 2024 · 12 citations) |
| Add bidirectional causality between HRs and HOIs | VOLLEYBALL · accuracy | 93.1 -> 96.1 | Helped | Bi-Causal: Group Activity Recognition via Bidirectional Causality (CVPR 2024 · 10 citations) |
| Increase Gaussian-noise mean range to μ~U(0,4) | Synthetic regulatory relationships, n=1500 · F1 | 69.6 ± 1.9 -> 72.6 ± 1.8 | Helped | Gene Regulatory Network Inference in the Presence of Selection Bias and Latent Confounders (NeurIPS 2025 · 1 citation) |
| add outgoing edges among input features | Synthetic · RMSE | 0.08±0.01 -> 0.06±0.01 | Helped | Towards Learning and Explaining Indirect Causal Effects in Neural Networks (AAAI 2024 · 1 citation) |
| Replace velocity matching with data denoising objective | observational and interventional datasets · MMD | No clear effect | Learning Structural Causal Models from Ordering: Identifiable Flow Models (AAAI 2025 · 0 citations) | |
| Use the co-occurrence topic generator instead of Gibbs topics | Text datasets; top 5 words per topic · UMass coherence | 2.53-4.20 times better mean coherence | Helped | E-LDA: Toward Interpretable LDA Topic Models with Strong Guarantees in Logarithmic Parallel Time (ICML 2025 · 0 citations) |
| increase m/d from 1 to approximately 2 | Mixture C · R² | 0.55 ± 0.07 -> 0.90 ± 0.10 | Helped | Diverse Influence Component Analysis: A Geometric Approach to Nonlinear Mixture Identifiability (NeurIPS 2025 · 0 citations) |
| add latent factors for customers and time | Synthetic churn, N=100, T=5, k=2 · normalized tensor mean squared error | 0.53 (0.02) -> 0.42 (0.07) | Helped | Causal Customer Churn Analysis with Low-rank Tensor Block Hazard Model (ICML 2024 · 0 citations) |
| Assume anticausal generation when the covariance between X2 and Y is unknown | Out-of-variable generalization with unknown predictor-target covariance · ability to include X2 in the predictor | Helped | Causal vs. Anticausal merging of predictors (NeurIPS 2024 · 0 citations) | |
| Use estimated covariance through the iterative framework instead of known covariance | Synthetic environments · MSE | No clear effect | Balancing Interference and Correlation in Spatial Experimental Designs: A Causal Graph Cut Approach (ICML 2025) | |
| add image-content confounder handling to CIS | CID2013, LIVE-C, and KonIQ-10K zero-shot · SROCC | 0.642 -> 0.651; 0.652 -> 0.684; 0.628 -> 0.658 | Helped | Causal-IQA: Towards the Generalization of Image Quality Assessment Based on Causal Inference (ICML 2024) |
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 73 papers; from OpenAlex and the papers' affiliations).
- Carnegie Mellon University · 5 papers
- ETHZ - ETH Zurich · 4 papers
- Max Planck Institute for Intelligent Systems · 3 papers
- National University of Singapore · 3 papers
- Columbia University · 3 papers
- Beijing Big Data Centre · 2 papers
- Beihang University · 2 papers
- Deakin University · 2 papers
- Fudan University · 2 papers
- Zhejiang 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
- Causal effect identification (107 papers)
- Causal representation identifiability (122 papers)
- Counterfactual identification and bounds (92 papers)
- Causal approaches to fairness (46 papers)
Elsewhere in the topic
- Constraint-based structure learning with latents (90 papers)
- Identifiability with hidden confounders (85 papers)
- Interventional causal discovery (68 papers)
- Differentiable and score-based DAG learning (52 papers)
- Causal discovery from time series (51 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).