mlsanityData to 14 Oct 2026

Towards Continuous-time Causal Foundation Models

arXiv 2605.28880 ↗First posted 26 May 2026cs.LG physics.data-an stat.MEcited by 4

Not on the map. Why?

Abstract (arXiv)

Extending discrete-time causal Prior-data Fitted Networks for time series to continuous time invites writing the mechanism as a stochastic differential equation (SDE) -- but if the SDE is integrated once per observation gap, the trajectory law depends on when it is observed, and the prior remains a discrete-time Markov model in SDE clothing. We propose a precise continuity criterion -- trajectory-law invariance to the observation schedule -- together with a three-tier taxonomy (discrete; naive observation-grid integration; fine-grid integration with decoupled observation) and a construction realising the top tier on a random DAG with OU or small-MLP nonlinear drifts, irregular observation schedules, and hard / soft / time-varying interventions. A 2 × 2 encoder × integrator ablation, run independently on a linear and a nonlinear prior, finds fine-grid integration beats naive on 8/8 cells (sign-consistency p < 1/256) with the gap growing as the eval grid refines; the encoder axis is null with fine integration but time-aware-leading with naive. We release the prior and a preliminary zero-shot protocol on pharmacokinetic and physical-system data.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 20 of its 29 references that are in the map. First posted on arXiv 26 May 2026.

Builds on

20 of 29 references.

References · 20

  1. 2018Neural Ordinary Differential Equationscited by 3,779
  2. 2022TabPFN: A Transformer That Solves Small Tabular Classification Problems in a Secondcited by 674
  3. 2021CSDI: Conditional Score-based Diffusion Models for Probabilistic Time Series Imputationcited by 582
  4. 2016Recurrent Neural Networks for Multivariate Time Series with Missing Valuescited by 554
  5. 2021Transformers Can Do Bayesian Inferencecited by 360
  6. 2021Multi-Time Attention Networks for Irregularly Sampled Time Seriescited by 238
  7. 2019Neural Stochastic Differential Equations: Deep Latent Gaussian Models in the Diffusion Limitcited by 176
  8. 2021Neural SDEs as Infinite-Dimensional GANscited by 161
  9. 2023ForecastPFN: Synthetically-Trained Zero-Shot Forecastingcited by 104
  10. 2023Statistical Foundations of Prior-Data Fitted Networkscited by 65
  11. 2025Do-PFN: In-Context Learning for Causal Effect Estimationcited by 51
  12. 2025Foundation Models for Causal Inference via Prior-Data Fitted Networkscited by 32
8 more
  1. 2025TimePFN: Effective Multivariate Time Series Forecasting with Synthetic Datacited by 29
  2. 2023Causal Modeling with Stationary Diffusionscited by 19
  3. 2024Dynamic Structural Causal Modelscited by 14
  4. 2024RealTCD: Temporal Causal Discovery from Interventional Data with Large Language Modelcited by 13
  5. 2024The Causal Chambers: Real Physical Systems as a Testbed for AI Methodologycited by 12
  6. 2025Causal Time Series Generation via Diffusion Modelscited by 8
  7. 2026Interventional Time Series Priors for Causal Foundation Modelscited by 7
  8. 2024CAnDOIT: Causal Discovery with Observational and Interventional Data from Time-Seriescited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 4.

4 papers.

  1. 2026DoTime: A Synthetic Benchmark Generator for Interventional and Counterfactual Time Seriescited by 2
  2. 2026Causal Foundation Modelscited by 1
  3. 2026Temporal Causal Prior-Data Fitted Networks for Panel Data with Learned Reliability Signalscited by 0
  4. 2026Market Impact Estimation via Causal Foundation Modelscited by 0