TORNADO is a latent probabilistic forecasting framework for partial differential equations that models next-step prediction as stochastic transport between VAE encoder posteriors. It learns a state-dependent stochastic differential equation, allowing predictive uncertainty to adapt to local latent dynamics rather than relying on fixed global stochasticity. The model is trained with analytic conditional endpoint targets and an energy distance regularizer, achieving better uncertainty calibration than several generative baselines while maintaining comparable predictive accuracy.
By Jorge Mifsut-Benet, Armand Kassa\"i Koupa\"i, Ramon Daniel Regueiro-Espino, Nicolas Baskiotis, Patrick Gallinari
arXiv:2602.02832v4 Announce Type: replace
Abstract: Forward forecasting and data assimilation are the two important aspects in physical simulation: one propagates the state forward, the other recover...
By Rares Grozavescu, Etienne Meunier, Pengyu Zhang, Mark Girolami
The paper introduces the Diffusion-Driven State Space Model (DDSSM), which replaces the Gaussian transition distribution in deep state space models with a diffusion model. This approach addresses the trade‑off between accurate forecasting and faithful capture of latent dynamics by enabling joint training of an autoencoder and a diffusion model on sequential data. Experiments show that DDSSM outperforms a state‑of‑the‑art deep SSM in fitting and forecasting a simulated time series with multimodal transitions.
By Jack Ruder, Michael Wojnowicz
The paper introduces prequential posteriors, a Bayesian approach that uses a predictive‑sequential loss function to update deep generative forecasting models (DGFMs) when new data arrive. By adopting a consistency notion suitable for model misspecification, the authors prove that both the loss minimizer and the posterior concentrate on parameters with optimal predictive performance. Scalable inference is achieved with parallelisable waste‑free sequential Monte Carlo samplers that employ preconditioned gradient kernels, and the method is validated on synthetic and real meteorological time‑series data.
By Shreya Sinha-Roy, Richard G. Everitt, Christian P. Robert, Ritabrata Dutta
The paper presents PAC‑Bayesian reconstruction guarantees for Variational Autoencoders applied to time‑series data. It extends existing bounds, which were limited to i.i.d. settings, to Markovian latent structures, allowing temporal dependencies to be captured without the bounds growing with trajectory length. The authors also provide an example framework showing that the required assumptions are not overly restrictive.
By Chlo\'e Hashimoto-Cullen, Ghislain Agoua, Benjamin Guedj, Sylvain Le Corff
arXiv:2507. 23615v2 Announce Type: replace-cross Abstract: Data augmentation is becoming increasingly important across various areas of time series analysis, including forecasting, classification, and anomaly detection.
By Luis Roque, Vitor Cerqueira, Carlos Soares, Luis Torgo
Neural PDE solvers provide efficient surrogates for time-dependent physical systems, but autoregressive prediction over long horizons remains challenging because local errors can induce distribution s...
The paper introduces a variational framework called VAMO that incorporates latent Markov dynamics for neural PDE solvers, aiming to improve long‑horizon predictions by mitigating error accumulation. By representing physical states as latent distributions and evolving them through probabilistic transitions, the method aligns learned dynamics with a spectral geometry induced by structured Gaussian perturbations. Experiments on fluid‑dynamics benchmarks show that VAMO reduces error growth and enhances rollout stability compared to deterministic and noise‑injection baselines.
By Junyi Liao, Johann Guilleminot, Vahid Tarokh
The paper introduces K$^2$SVD, a method that learns the leading singular functions of the Koopman operator by optimizing a Hilbert-Schmidt objective, producing a low‑rank, interpretable Koopman representation with a compact latent space. In this space, temporal evolution is modeled with a linear Gaussian state‑space model and inference is performed via Kalman filtering to reduce noise accumulation in multi‑step predictions. Experiments demonstrate that K$^2$SVD outperforms state‑of‑the‑art methods on multiple datasets, achieving faster prediction speeds and lower computational cost.
By Ruiquan Li, Yuheng Bu
arXiv:2607. 20521v1 Announce Type: new Abstract: The state of a dynamic system evolves over time, switching among several latent modes that govern its observable behavior.
By Lei Cao, Sihang Feng, Jixin Yan, Tao Sun, Naichen Shi
The paper introduces DV-Flow, a flow‑matching framework that treats deterministic prediction as a parallel velocity component instead of embedding it within the generative process. This design stabilizes end‑to‑end training and unifies several existing methods as special cases, enabling direct comparison. Experiments on six real‑world datasets show that DV‑Flow consistently outperforms strong baselines in probabilistic time‑series forecasting.
arXiv:2608. 20025v1 Announce Type: new Abstract: Probabilistic forecasting models are widely used for time series forecasting in domains such as energy systems, finance, medicine, and transportation.
By Alexander Marusov, Dmitry Anikin, Petr Sokerin, Vitaliy Pozdnyakov, Ilya Kuleshov, Alexey Zaytsev