mlsanityData to 14 Oct 2026

A Comparative Study on How Data Normalization Affects Zero-Shot Generalization in Time Series Foundation Models

arXiv 2512.02833 ↗First posted 2 Dec 2025cs.LG cs.AIcited by 1

Not on the map. Why?

Abstract (arXiv)

We investigate input normalization methods for Time-Series Foundation Models (TSFMs). While normalization is well-studied in dataset-specific time-series models, it remains overlooked in TSFMs where generalization is critical. Time-series data, unlike text or images, exhibits significant scale variation across domains and channels, coupled with non-stationarity, can undermine TSFM performance regardless of architectural complexity. Through systematic evaluation across four architecturally diverse TSFMs, we empirically establish REVIN as the most efficient approach, reducing zero-shot MASE by 89% relative to an un-normalized baseline and by 44% versus other normalization methods, while matching the best in-domain accuracy (0.84 MASE) without any dataset-level preprocessing -- yielding the highest accuracy-efficiency trade-off. Yet its effect utilization depends on architectural design choices and optimization objective, particularly with respect to training loss scale sensitivity and model type (probabilistic, point-forecast, or LLM-based models).

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 21 of its 31 references that are in the map. First posted on arXiv 2 Dec 2025.

Builds on

21 of 31 references.

References · 21

  1. 2020Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecastingcited by 2,169
  2. 2017DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networkscited by 1,013
  3. 2024Chronos: Learning the Language of Time Seriescited by 759
  4. 2023Time-LLM: Time Series Forecasting by Reprogramming Large Language Modelscited by 700
  5. 2023A decoder-only foundation model for time-series forecastingcited by 692
  6. 2024Unified Training of Universal Time Series Forecasting Transformerscited by 614
  7. 2024MOMENT: A Family of Open Time-series Foundation Modelscited by 482
  8. 2021SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interactioncited by 338
  9. 2024Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Expertscited by 287
  10. 2024Tiny Time Mixers (TTMs): Fast Pre-trained Models for Enhanced Zero/Few-Shot Forecasting of Multivariate Time Seriescited by 127
  11. 2023ForecastPFN: Synthetically-Trained Zero-Shot Forecastingcited by 104
  12. 2024ChatTime: A Unified Multimodal Time Series Foundation Model Bridging Numerical and Textual Datacited by 100
9 more
  1. 2024Moirai-MoE: Empowering Time Series Foundation Models with Sparse Mixture of Expertscited by 95
  2. 2025This Time is Different: An Observability Perspective on Time Series Foundation Modelscited by 68
  3. 2024Frequency Adaptive Normalization For Non-stationary Time Series Forecastingcited by 47
  4. 2024Towards Neural Scaling Laws for Time Series Foundation Modelscited by 33
  5. 2025LightGTS: A Lightweight General Time Series Forecasting Modelcited by 29
  6. 2020The Effectiveness of Discretization in Forecasting: An Empirical Study on Neural Time Series Modelscited by 18
  7. 2025Out-of-Distribution Generalization in Time Series: A Surveycited by 16
  8. 2024Nonstationary Time Series Forecasting via Unknown Distribution Adaptationcited by 2
  9. 2024VSFormer: Value and Shape-Aware Transformer with Prior-Enhanced Self-Attention for Multivariate Time Series Classificationcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Does Normalization Choice Matter for Causal Large Time-Series Models?cited by 0