Research map / Time series / Causal and generative time series
Irregular series and imputation
Handle missing values and irregular sampling with probabilistic, generative and continuous-time models.
59 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Causal and generative time series, keyword cluster “differential · continuous · equations”
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.
14 papers in 2023–24 and 26 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 9.0% to 11.4%.
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
- 2016 · A scalable end-to-end Gaussian process adapter for irregularly sampled time series classification (NeurIPS 2016 · 48 citations)
- 2018 · BRITS: Bidirectional Recurrent Imputation for Time Series (NeurIPS 2018 · 372 citations)
- 2019 · Interpolation-Prediction Networks for Irregularly Sampled Time Series (ICLR 2019 · 30 citations)
- 2020 · Set Functions for Time Series (ICML 2020 · 23 citations)
- 2021 · Generative Semi-supervised Learning for Multivariate Time Series Imputation (AAAI 2021 · 195 citations)
- 2022 · Graph-Guided Network for Irregularly Sampled Multivariate Time Series (ICLR 2022 · 24 citations)
- 2023 · Estimating Treatment Effects from Irregular Time Series Observations with Hidden Confounders (AAAI 2023 · 12 citations)
- 2024 · GraFITi: Graphs for Forecasting Irregularly Sampled Time Series (AAAI 2024 · 27 citations)
- 2025 · Self-attention-based Diffusion Model for Time-series Imputation in Partial Blackout Scenarios (AAAI 2025 · 9 citations)
- 2026 · Causal Discovery for Irregularly Time Series with Consistency Guarantees (ICML 2026 · 0 citations)
Earliest papers here
- A scalable end-to-end Gaussian process adapter for irregularly sampled time series classification (NeurIPS 2016 · 48 citations)
- BRITS: Bidirectional Recurrent Imputation for Time Series (NeurIPS 2018 · 372 citations)
- Multivariate Time Series Imputation with Generative Adversarial Networks (NeurIPS 2018 · 291 citations)
What has been tried
35 papers here report 335 measured changes (a component added, removed or swapped, compared with a baseline): helped 74%, no clear effect 6%, hurt 16%, 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 |
|---|---|---|---|---|
| Add FDE | Partial-blackout imputation across AgAID, Air Quality, Electricity, and NACSE · MSE | 1.43e-03 ± 9.00e-04 -> 2.93e-04 ± 1.04e-04 (AgAID, 3 missing features); 3.74e-02 ± 2.04e-02 -> 4.88e-03 ± | Helped | Self-attention-based Diffusion Model for Time-series Imputation in Partial Blackout Scenarios (AAAI 2025 · 9 citations) |
| Add paired metadata conditioning | Air Quality and Electricity · FTSD | Mixed | Time Weaver: A Conditional Time Series Generation Model (ICML 2024 · 3 citations) | |
| Add SITA | Physionet’12 · njNLL | -0.470±0.017 -> -0.647±0.078 | Helped | Probabilistic Forecasting of Irregularly Sampled Time Series with Missing Values via Conditional Normalizing Flows (AAAI 2025 · 2 citations) |
| Remove the two-stage curriculum and train all model parameters jointly in one stage | kdd-cup and solar-10min forecasting · validation NLL | Hurt | TACTiS-2: Better, Faster, Simpler Attentional Copulas for Multivariate Time Series (ICLR 2024 · 2 citations) | |
| add learned channel-wise weights instead of fixing all weights to one | Traffic-GuangZhou, 50% observed · RMSE | 11.032 -> 3.820 | Helped | BayOTIDE: Bayesian Online Multivariate Time Series Imputation with Functional Decomposition (ICML 2024 · 2 citations) |
| Replace full gain estimation with Seg-4 accelerated gain estimation | GEF · MSE | 0.1653(0.0008) -> 0.1666(0.0007) | Hurt | Task-oriented Time Series Imputation Evaluation via Generalized Representers (NeurIPS 2024 · 2 citations) |
| Add spectral consistency loss | PhysioNet, 10% missing · MAE | 0.213 -> 0.211 | Helped | LSCD: Lomb--Scargle Conditioned Diffusion for Time series Imputation (ICML 2025 · 0 citations) |
| Use GRU flow instead of ResNet flow | Sink, Triangle, Sawtooth, and Square (5-node synthetic) · MSE | 3.95 ± 0.32 -> 6.83 ± 0.23; 2.32 ± 0.11 -> 5.41 ± 0.23; 3.84 ± 0.06 -> 5.11 ± 0.13; 8.24 ± 0.64 -> 9.14 ± 0.61 | Mixed | Graph Neural Flows for Unveiling Systemic Interactions Among Irregularly Sampled Time Series (NeurIPS 2024 · 0 citations) |
| add the linear predictive prior (nonzero alpha) | regular Sines, Stocks, Energy, and MuJoCo · discriminative score | 0.006 -> 0.005; 0.023 -> 0.009; 0.155 -> 0.143; 0.087 -> 0.076 | Helped | Generative Modeling of Regular and Irregular Time Series Data via Koopman VAEs (ICLR 2024) |
What reviewers will expect
Datasets and baselines named in the 16 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- USHCN · 3 of 16 papers
- PhysioNet · 3 of 16 papers
- MIMIC-III · 2 of 16 papers
- Electricity · 2 of 16 papers
- Human Activity · 2 of 16 papers
- Beijing air quality · 2 of 16 papers
- ETTh · 2 of 16 papers
- Energy · 2 of 16 papers
Baselines
- BRITS · 5 of 16 papers
- CSDI · 4 of 16 papers
- SAITS · 4 of 16 papers
- CRU · 3 of 16 papers
- GP-VAE · 3 of 16 papers
- Latent-ODE · 3 of 16 papers
- US-GAN · 2 of 16 papers
- TimesNet · 2 of 16 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 40 papers; from OpenAlex and the papers' affiliations).
- Carnegie Mellon University · 3 papers
- University of Hildesheim · 2 papers
- Institute of Mathematics · 2 papers
- Alibaba Group · 2 papers
Use it
Download BibTeX of 38 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
- Neural differential equations (52 papers)
- Diffusion models for time series (55 papers)
Elsewhere in the topic
- Probabilistic and conformal forecasting (121 papers)
- State-space and latent dynamics forecasting (94 papers)
- Robust forecasting and evaluation (91 papers)
- Dynamical system reconstruction (90 papers)
- Generative temporal point processes (53 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).