mlsanityData to 13 Oct 2026

There is No "apple" in Timeseries: Rethinking TSFM through the Lens of Invariance

arXiv 2510.20119 ↗First posted 23 Oct 2025cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Timeseries foundation models (TSFMs) have multiplied, yet lightweight supervised baselines and even classical models often match them. We argue this gap stems from the naive importation of NLP or CV pipelines. In language and vision, large web-scale corpora densely capture human concepts i.e. there are countless images and text of apples. In contrast, timeseries data is built to complement the image and text modalities. There are no timeseries dataset that contains the concept apple. As a result, the scrape-everything-online paradigm fails for TS. We posit that progress demands a shift from opportunistic aggregation to principled design: constructing datasets that systematically span the space of invariance that preserve temporal semantics. To this end, we suggest that the ontology of timeseries invariances should be built based on first principles. Only by ensuring representational completeness through invariance coverage can TSFMs achieve the aligned structure necessary for generalisation, reasoning, and truly emergent behaviour.

Key facts. No citing papers in the map yet (data through 13 Oct 2026). Builds on 12 of its 14 references that are in the map. First posted on arXiv 23 Oct 2025.

Builds on

12 of 14 references.

References · 12

  1. 2020Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecastingcited by 2,157
  2. 2021Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gaugescited by 1,147
  3. 2024Chronos: Learning the Language of Time Seriescited by 737
  4. 2023A decoder-only foundation model for time-series forecastingcited by 674
  5. 2024Unified Training of Universal Time Series Forecasting Transformerscited by 602
  6. 2024MOMENT: A Family of Open Time-series Foundation Modelscited by 468
  7. 2021Monash Time Series Forecasting Archivecited by 261
  8. 2024Timer: Generative Pre-trained Transformers Are Large Time Series Modelscited by 205
  9. 2025This Time is Different: An Observability Perspective on Time Series Foundation Modelscited by 64
  10. 2024Enhancing Foundation Models for Time Series Forecasting via Wavelet-based Tokenizationcited by 26
  11. 2024Specialized Foundation Models Struggle to Beat Supervised Baselinescited by 24
  12. 2025TimeHF: Billion-Scale Time Series Models Guided by Human Feedbackcited by 10

Built on by

Not cited yet.