mlsanityData to 14 Oct 2026

PAC-Bayesian Reconstruction Guarantees for Time Series Variational Autoencoders

arXiv 2609.05212 ↗First posted 4 Sep 2026stat.ML cs.LGcited by 1

Not on the map. Why?

Abstract (arXiv)

Forecasting time series accurately is critical for applications with complex data ranging from energy systems to healthcare and finance. Among current state of the art models, generative latent variable models are increasingly implemented; yet principled generalisation guarantees for modern latent variable models remain limited. In particular, while Variational AutoEncoders are widely used for sequential data, their theoretical analysis is largely restricted to i.i.d. settings. In this work, we develop a PAC-Bayesian framework for latent variables models applied to time series. Building on reconstruction-based bounds, we extend PAC-Bayesian guarantees to Markovian latent structures, capturing temporal dependencies through a sequential generative process. These guarantees do not grow with the length of the trajectory. Our bounds depend on assumptions which are common in the literature; we provide an example framework where they would be verified to show that they are not as restrictive as they may seem.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 48 of its 63 references that are in the map. First posted on arXiv 4 Sep 2026.

Builds on

48 of 63 references.

References · 48

  1. 2017Neural Discrete Representation Learningcited by 4,712
  2. 2013Auto-Encoding Variational Bayescited by 4,053
  3. 2022A Time Series is Worth 64 Words: Long-term Forecasting with Transformerscited by 2,039
  4. 2016Variational Inference: A Review for Statisticianscited by 1,472
  5. 2014On the Properties of Neural Machine Translation: Encoder-Decoder Approachescited by 1,304
  6. 2019An Introduction to Variational Autoencoderscited by 1,000
  7. 2024Chronos: Learning the Language of Time Seriescited by 759
  8. 2017Computing Nonvacuous Generalization Bounds for Deep (Stochastic) Neural Networks with Many More Parameters than Training Datacited by 533
  9. 2019Variational Autoencoders and Nonlinear ICA: A Unifying Frameworkcited by 503
  10. 2016Tutorial on Variational Autoencoderscited by 409
  11. 2025TabICL: A Tabular Foundation Model for In-Context Learning on Large Datacited by 289
  12. 2021User-friendly introduction to PAC-Bayes boundscited by 239
36 more
  1. 2021TimeVAE: A Variational Auto-Encoder for Multivariate Time Series Generationcited by 142
  2. 2019A Primer on PAC-Bayesian Learningcited by 132
  3. 2016Composing graphical models with neural networks for structured representations and fast inferencecited by 113
  4. 2020Tighter risk certificates for neural networkscited by 112
  5. 2017Entropy-SGD optimizes the prior of a PAC-Bayes bound: Generalization properties of Entropy-SGD and data-dependent priorscited by 81
  6. 2020PAC-Bayes Analysis Beyond the Usual Boundscited by 63
  7. 2023Generalization Bounds: Perspectives from Information Theory and PAC-Bayescited by 61
  8. 2016Simpler PAC-Bayesian Bounds for Hostile Datacited by 46
  9. 2023Causal Recurrent Variational Autoencoder for Medical Time Series Generationcited by 45
  10. 2020PAC-Bayes unleashed: generalisation bounds with unbounded lossescited by 43
  11. 2021Anomaly Detection of Time Series with Smoothness-Inducing Sequential Variational Auto-Encodercited by 39
  12. 2019Dichotomize and Generalize: PAC-Bayesian Binary Activated Deep Neural Networkscited by 38
  13. 2020Differentiable PAC-Bayes Objectives with Partially Aggregated Neural Networkscited by 29
  14. 2022PAC-Bayes Generalisation Bounds for Heavy-Tailed Losses through Supermartingalescited by 24
  15. 2022Non-Vacuous Generalisation Bounds for Shallow Neural Networkscited by 22
  16. 2022On PAC-Bayesian reconstruction guarantees for VAEscited by 20
  17. 2021On Margins and Derandomisation in PAC-Bayescited by 19
  18. 2021Online Variational Filtering and Parameter Learningcited by 19
  19. 2018Variational Message Passing with Structured Inference Networkscited by 17
  20. 2020Towards a Theoretical Understanding of the Robustness of Variational Autoencoderscited by 17
  21. 2022Decoupling Local and Global Representations of Time Seriescited by 17
  22. 2022Diffusion bridges vector quantized Variational AutoEncoderscited by 16
  23. 2023Statistical Guarantees for Variational Autoencoders using PAC-Bayesian Theorycited by 16
  24. 2025Lightweight Online Adaption for Time Series Foundation Model Forecastscited by 16
  25. 2020VEST: Automatic Feature Engineering for Forecastingcited by 14
  26. 2021Certifiably Robust Variational Autoencoderscited by 14
  27. 2023PAC-Bayesian Generalization Bounds for Adversarial Generative Modelscited by 13
  28. 2023A Note on the Convergence of Denoising Diffusion Probabilistic Modelscited by 9
  29. 2023Hybrid Variational Autoencoder for Time Series Forecastingcited by 8
  30. 2024Theoretical Convergence Guarantees for Variational Autoencoderscited by 4
  31. 2025Empirical PAC-Bayes Bounds for Markov Chainscited by 3
  32. 2026Generalizing Score-based generative models for Heavy-tailed Distributionscited by 3
  33. 2023Variational latent discrete representation for time series modellingcited by 2
  34. 2023Variational excess risk bound for general state space modelscited by 2
  35. 2026Cross-Domain Offshore Wind Power Forecasting: Transfer Learning Through Meteorological Clusterscited by 1
  36. 2026Efficient Online Variational Estimation via Monte Carlo Samplingcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Latent Inference-Time Guidance of Time Series Foundation Modelscited by 0