Research map / Probabilistic and scientific modeling / Causal discovery
Causal discovery from time series
Recover causal structure from temporal data under non-stationarity, latent confounding and Granger-style assumptions.
51 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · 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.
13 papers in 2023–24 and 25 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 5.6% to 7.5%.
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 · Uncertainty Assessment and False Discovery Rate Control in High-Dimensional Granger Causal Inference (ICML 2017 · 7 citations)
- 2019 · Causal Discovery and Forecasting in Nonstationary Environments with State-Space Models (ICML 2019 · 22 citations)
- 2020 · High-recall causal discovery for autocorrelated time series with latent confounders (NeurIPS 2020 · 38 citations)
- 2021 · Causal Hidden Markov Model for Time Series Disease Forecasting (CVPR 2021 · 36 citations)
- 2022 · IDYNO: Learning Nonparametric DAGs from Interventional Dynamic Data (ICML 2022)
- 2023 · CUTS: Neural Causal Discovery from Irregular Time-Series Data (ICLR 2023 · 9 citations)
- 2024 · CUTS+: High-Dimensional Causal Discovery from Irregular Time-Series (AAAI 2024 · 37 citations)
- 2025 · SPACETIME: Causal Discovery from Non-Stationary Time Series (AAAI 2025 · 6 citations)
- 2026 · TimeSAE: Causal Sparse Decoding for Faithful Explanations of Black-Box Time Series Models (ICML 2026 · 0 citations)
Earliest papers here
- Uncertainty Assessment and False Discovery Rate Control in High-Dimensional Granger Causal Inference (ICML 2017 · 7 citations)
- Discovering Potential Correlations via Hypercontractivity (NeurIPS 2017 · 4 citations)
- Causal Discovery and Forecasting in Nonstationary Environments with State-Space Models (ICML 2019 · 22 citations)
What has been tried
35 papers here report 293 measured changes (a component added, removed or swapped, compared with a baseline): helped 67%, no clear effect 8%, hurt 20%, mixed 4%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| replace CUTS prediction network with MPGNN (CUTS with C2FD to CUTS+) | VAR, RM p=0.3 · AUROC | 0.9594 -> 0.9907 | Helped | CUTS+: High-Dimensional Causal Discovery from Irregular Time-Series (AAAI 2024 · 37 citations) |
| remove the pre-training stage | Sock-shop · root cause analysis performance | Hurt | Root Cause Analysis in Microservice Using Neural Granger Causal Discovery (AAAI 2024 · 33 citations) | |
| add the information aggregation path | VAR-30 and Joint-VAR-20 · causal discovery performance | Helped | Towards the Disappearing Truth: Fine-Grained Joint Causal Influences Learning with Hidden Variable-Driven Causal Hypergraphs in Time Series (AAAI 2024 · 8 citations) | |
| Use temporal permutation instead of noise injection or zero-masking | Lorenz#1 · AUROC | .981±.056 / .974±.082 -> .999±.002 | Helped | UnCLe: Towards Scalable Dynamic Causal Discovery in Non-linear Temporal Systems (NeurIPS 2025 · 1 citation) |
| Use an xLSTM forecaster instead of a standard LSTM | Lorenz-96 F=40 · balanced accuracy | 93.0±0.3 -> 96.6±0.3 | Helped | Exploring Neural Granger Causality with xLSTMs: Unveiling Temporal Dependencies in Complex Data (NeurIPS 2025 · 1 citation) |
| replace normalizing-flow noise with additive Gaussian noise | AQI · discriminative score | 0.054 { ± 0.025} -> 0.488 { ± 0.001} | Hurt | CausalTime: Realistically Generated Time-series for Benchmarking of Causal Discovery (ICLR 2024 · 1 citation) |
| Remove the cosine similarity penalty (ρ = 0) | Synthetic System 6-2-2; 6-4-2; 12-11-2; 12-11-5; D4IC HSNR · mean optimal F1 | Mixed | Generating Hypotheses of Dynamic Causal Graphs in Neuroscience: Leveraging Generative Factor Models of Observed Time Series (ICML 2025 · 0 citations) | |
| Underspecify the number of mixture components | Nonlinear synthetic, D=10, K*=10 · clustering accuracy | Hurt | Discovering Mixtures of Structural Causal Models from Time Series Data (ICML 2024 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 15 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Lorenz96 · 2 of 15 papers
Baselines
- PCMCI · 5 of 15 papers
- DYNOTEARS · 5 of 15 papers
- VARLiNGAM · 4 of 15 papers
- R-PCMCI · 3 of 15 papers
- cMLP · 3 of 15 papers
- VAR · 3 of 15 papers
- CD-NOD · 2 of 15 papers
- TDRL · 2 of 15 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 38 papers; from OpenAlex and the papers' affiliations).
- Chinese PLA General Hospital · 3 papers
- Tsinghua University · 3 papers
- Zhejiang University · 3 papers
- Peking University · 3 papers
- University of Maryland, Baltimore County · 2 papers
- CISPA Helmholtz Center for Information Security · 2 papers
- University of California, San Diego · 2 papers
Use it
Download BibTeX of 34 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)
- Interventional causal discovery (68 papers)
- Differentiable and score-based DAG learning (52 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).