mlsanityData to 14 Oct 2026

Latent Inference-Time Guidance of Time Series Foundation Models

arXiv 2609.38058 ↗First posted 29 Sep 2026stat.ML cs.LG stat.MEcited by 0

Not on the map. Why?

Abstract (arXiv)

Time Series Foundation Models (TSFMs) currently provide state-of-the-art results in forecasting tasks. They are available out-of-the-box and rely on in-context learning to make their predictions, which makes the quality of their performance highly sensitive to the user-selected lookback, covariates, horizon and training data distributions. In practise, the quality of the forecasts are variable but complementary, which highlights the need for a principled ensembling approach, rather than selecting the best context. This paper introduces Latent Inference-Time Guidance for TSFMs, which adaptively combines a pool of TSFM forecasts through a time-dependent latent space with independent components. The framework comes equipped with identifiability and reconstruction guarantees, whilst maintaining the off-the-shelf aspect of foundation models. We provide experiments on datasets at various frequencies and from multiple domains: these show that the approach is competitive with traditional ensembling approaches.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 46 of its 56 references that are in the map. First posted on arXiv 29 Sep 2026.

Builds on

46 of 56 references.

References · 46

  1. 2017Attention Is All You Needcited by 60,361
  2. 2016Layer Normalizationcited by 5,669
  3. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,073
  4. 2022A Time Series is Worth 64 Words: Long-term Forecasting with Transformerscited by 2,039
  5. 2019An Introduction to Variational Autoencoderscited by 1,000
  6. 2024Chronos: Learning the Language of Time Seriescited by 759
  7. 2023A decoder-only foundation model for time-series forecastingcited by 692
  8. 2022TabPFN: A Transformer That Solves Small Tabular Classification Problems in a Secondcited by 674
  9. 2015Cyclical Learning Rates for Training Neural Networkscited by 629
  10. 2024Unified Training of Universal Time Series Forecasting Transformerscited by 614
  11. 2021Tabular Data: Deep Learning is Not All You Needcited by 535
  12. 2019Variational Autoencoders and Nonlinear ICA: A Unifying Frameworkcited by 503
34 more
  1. 2025TabICL: A Tabular Foundation Model for In-Context Learning on Large Datacited by 289
  2. 2024Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Expertscited by 287
  3. 2017VAE with a VampPriorcited by 248
  4. 2023When Do Neural Nets Outperform Boosted Trees on Tabular Data?cited by 223
  5. 2024GIFT-Eval: A Benchmark For General Time Series Forecasting Model Evaluationcited by 212
  6. 2024Timer: Generative Pre-trained Transformers Are Large Time Series Modelscited by 208
  7. 2025Chronos-2: From Univariate to Universal Forecastingcited by 194
  8. 2026TabICLv2: A better, faster, scalable, and open tabular foundation modelcited by 144
  9. 2022Weakly supervised causal representation learningcited by 140
  10. 2025TabPFN-2.5: Advancing the State of the Art in Tabular Foundation Modelscited by 132
  11. 2025TiRex: Zero-Shot Forecasting Across Long and Short Horizons with Enhanced In-Context Learningcited by 120
  12. 2014A Second-order Bound with Excess Lossescited by 95
  13. 2024Moirai-MoE: Empowering Time Series Foundation Models with Sparse Mixture of Expertscited by 95
  14. 2024Timer-XL: Long-Context Transformers for Unified Time Series Forecastingcited by 87
  15. 2025Moirai 2.0: When Less Is More for Time Series Forecastingcited by 69
  16. 2021Disentangling Identifiable Features from Noisy Data with Structured Nonlinear ICAcited by 64
  17. 2025fev-bench: A Realistic Benchmark for Time Series Forecastingcited by 64
  18. 2025No Other Representation Component Is Needed: Diffusion Transformers Can Provide Representation Guidance by Themselvescited by 56
  19. 2020Variational Autoencoder with Learned Latent Structurecited by 41
  20. 2014Optimal learning with Bernstein Online Aggregationcited by 35
  21. 2024In-context Time Series Predictorcited by 23
  22. 2026Brittlebench: Quantifying LLM robustness via prompt sensitivitycited by 21
  23. 2022On PAC-Bayesian reconstruction guarantees for VAEscited by 20
  24. 2023Statistical Guarantees for Variational Autoencoders using PAC-Bayesian Theorycited by 16
  25. 2025Lightweight Online Adaption for Time Series Foundation Model Forecastscited by 16
  26. 2024Learning biologically relevant features in a pathology foundation model using sparse autoencoderscited by 14
  27. 2024Benchmarking Time Series Foundation Models for Short-Term Household Electricity Load Forecastingcited by 12
  28. 2023Adaptive Probabilistic Forecasting of Electricity (Net-)Loadcited by 9
  29. 2025Conversational Time Series Foundation Models: Towards Explainable and Effective Forecastingcited by 8
  30. 2024Not all Views are Created Equal: Analyzing Viewpoint Instabilities in Vision Foundation Modelscited by 5
  31. 2026Visual Disentangled Diffusion Autoencoders: Scalable Counterfactual Generation for Foundation Modelscited by 4
  32. 2026Investigating simple target-covariate relationships for Chronos-2 and TabPFN-TScited by 2
  33. 2024Importance sampling for online variational learningcited by 1
  34. 2026PAC-Bayesian Reconstruction Guarantees for Time Series Variational Autoencoderscited by 1

Built on by

Not cited yet.