Pith. sign in

REVIEW 3 major objections 5 minor 54 references

Time Series Representations for Classification Lie Hidden in Pretrained Vision Transformers

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that frozen image-pretrained vision transformers, fed images made by stacking time-series segments, classify time series better than specialized time-series foundation models without any fine-tuning.

desk verdict Frozen vision transformers are a genuinely strong feature extractor for time series; the paper's main empirical result holds, but the headline significance test is over-claimed and the theory is not load-bearing. read the letter →

arxiv 2506.08641 v2 pith:SUMDZIVL submitted 2025-06-10 cs.LG cs.AIcs.CV

Simon Roschmann , Quentin Bouniot , Vasilii Feofanov , Ievgen Redko , Zeynep Akata This is my paper
classification cs.LGcs.AIcs.CV
keywords timeseriesclassificationvisiontransformersfrozenfeatureextraction2Dpatchingrepresentationtransferfoundationmodelscomplementaryrepresentationsintrinsicdimension
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a time series can be treated as an image, and that frozen vision transformers pretrained on billions of natural images become state-of-the-art time-series classifiers. The proposed TiViT pipeline splits each univariate series into segments, stacks them into a 2D matrix, resizes it to a ViT's input resolution, and averages hidden-layer token representations before fitting a linear classifier. On the 128-dataset UCR benchmark this reaches 81.3% mean accuracy, surpassing the time-series foundation model Mantis at 80.1% (paired t-test p=0.03); on the 27 multivariate UEA datasets it matches Mantis, and concatenating TiViT features with Mantis features reaches 83.0% on UCR. The paper also argues theoretically that 2D patching increases the number of label-relevant tokens seen by a Transformer, which reduces sample complexity during training. If correct, this shows that large-scale vision pretraining transfers to a non-visual domain without any fine-tuning.

What carries the argument

The load-bearing mechanism is the time-series-to-image transformation followed by token averaging at an intermediate layer. A univariate series is normalized by robust scaling, cut into M overlapping patches of length P = $\sqrt$(T) with stride P/10, stacked into an M-by-P grayscale image, resized to the ViT's square resolution, and fed through the frozen ViT; the embedding is the mean of all K+1 tokens at a selected layer. The theoretical support is Proposition 1, which uses the label-relevant-token sample-complexity result of the paper's reference [31]: in a simplified T = $k^{2}$, p = k setting, 2D patching turns each label-relevant segment of length k into contributions to $\sqrt$(k) patches, so the fraction of label-relevant tokens is at least as large as in 1D patching and strictly larger when the number of label-relevant segments is not a multiple of $\sqrt$(k). The intermediate layers with highest intrinsic dimension are the ones that transfer best.

What would settle it

Compute, on several real UCR datasets, the fraction of tokens that are closer to the true class-mean pattern in 1D versus 2D patching, using the paper's own definition of label-relevant tokens. If 2D patching does not increase this fraction whenever the number of label-relevant segments is not a multiple of sqrt(k), the theoretical mechanism would be refuted even if the empirical accuracy numbers stand.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the hidden representations of a frozen image-pretrained ViT, extracted at roughly 40 to 70 percent of the layer depth, encode enough temporal structure to classify time series competitively with, and on UCR better than, foundation models trained specifically on time-series data. The best configuration uses a contrastively pretrained CLIP ViT-H/14 at layer 14, giving 81.3% mean accuracy across 128 UCR datasets and 72.0% across UEA. The same representations are complementary to time-series foundation models: concatenating TiViT with Mantis raises UCR accuracy to 83.0% and UEA to 73.7%, and the mutual-kNN alignment of the two representation spaces is low, indicating that the two models capture different signal.

Load-bearing premise

The argument's load-bearing premise is the assumption behind Proposition 1: a 2D patch containing at least sqrt(k) elements of the label-relevant pattern automatically counts as label-relevant, because its dot product with the two class means falls on the right side of the decision threshold. The proof does not show that this condition still holds after the real pipeline resizes the stacked segments with nearest interpolation and re-patches them with the frozen ViT's own patch size.

Editorial extensions

If this is right

  • Frozen vision models can replace or match task-specific time-series foundation models in linear-probe classification, at least on the standard UCR and UEA benchmarks.
  • Hidden-layer features matter: using the mean token representation from a mid-depth layer outperforms both the final layer and the CLS token for time-series classification.
  • TiViT's gains concentrate where data is scarce: on the ten UCR datasets with the fewest training samples it reaches 89.8% versus 86.6% for Mantis, and on the longest series 75.0% versus 70.5%.
  • Combining vision-derived and time-series-derived features is a cheap accuracy boost: simple concatenation beats every single model on both benchmarks and also improves zero-shot nearest-centroid classification from about 71% to 73.4%.
  • The finding that intermediate layers with high intrinsic dimension are the most effective gives a principled way to select which frozen layer to use for a new dataset.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The theory in Proposition 1 is proven for a simplified non-overlapping square-patch setting; carrying the label-relevance argument through nearest-interpolation resizing and the frozen ViT's own patch embedding would turn the paper's motivation into a theorem for the actual pipeline.
  • The complementarity result suggests a cheap strategy for a new time-series benchmark: concatenate a frozen ViT's mid-layer embedding with a contrastively trained time-series foundation model's embedding rather than fine-tuning either model.
  • If the intrinsic-dimension pattern holds across domains, layer selection for frozen encoders could be done by measuring intrinsic dimension on unlabeled target data instead of by validation accuracy.
  • The same image-conversion recipe could be tested on other non-visual signals, such as audio spectrograms or sensor arrays, to see whether the transfer benefit extends beyond time series.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes TiViT, a framework that converts univariate time series into grayscale images by segmenting the series into patches, stacking them into a matrix, resizing to a ViT-compatible square resolution, and feeding the result through a frozen pretrained Vision Transformer (CLIP, DINOv2, or SigLIP 2). A linear classifier is trained on the mean of the hidden token representations from a selected layer. The authors provide a theoretical proposition claiming that 2D patching increases the number of label-relevant tokens and reduces sample complexity relative to 1D patching. Empirically, they report 81.3% mean accuracy on the 128-dataset UCR benchmark versus 80.1% for the Mantis TSFM (paired t-test p=0.03), 72.0% versus 72.4% on UEA, and 83.0% when concatenating TiViT with Mantis features. They also analyze layer-wise intrinsic dimension and representation alignment between ViTs and TSFMs, and report ablations over patch size, overlap, backbone size, and classifier choice.

Significance. If the empirical claims survive proper statistical treatment, this is a significant result: it demonstrates that large-scale vision pretraining can transfer to time series classification without any fine-tuning, and it quantifies a useful complementarity with dedicated time series foundation models. The paper's strengths include the use of public UCR/UEA splits, results averaged over three seeds, per-dataset tables in the appendix, a consistent linear-probing protocol, no fine-tuning of the ViT backbone, and released code. The theoretical contribution is currently a conditional statement that is not directly connected to the actual TiViT pipeline, and the headline statistical claim is computed after model selection. With those issues addressed, the paper would be a solid empirical contribution to the reuse of vision representations for time series.

major comments (3)
  1. [4.3 (Benchmark), Tables 3-4] The headline comparison against Mantis (81.3 vs 80.1, p=0.03) is computed for TiViT-CLIP after the model and its hyperparameters were selected on the same benchmarks. Table 3 reports the best test accuracy among three pretraining paradigms (CLIP 81.3, DINOv2 80.0, SigLIP 2 80.6); the best layer per backbone is chosen by validation, and the pipeline also selects the patch overlap (Table 2b, highest validation accuracy at 0.9) and the token aggregation (Table A3) on validation data. A paired t-test that treats this configuration as fixed ignores this selection. Because the reported p-value is close to the 0.05 threshold, it may not survive any reasonable correction for the number of configurations explored. The authors should either pre-specify a fixed configuration before seeing test labels, report corrected p-values (e.g., Bonferroni or Westfall-Young permutation over the 128 datasets), or add a nested validation scheme that accounts for model selection. Without this, the claim that TiViT significantly outperforms TSFMs is not supported by the reported test.
  2. [3.1, Eq. (1); Section 3.2] Proposition 1 is not a theorem about the TiViT pipeline used in the experiments. It assumes T=k^2, p=k, non-overlapping sqrt(k) x sqrt(k) patches of a square matrix, and the conditional assumption that any 2D token containing at least sqrt(k) elements of the label-relevant pattern satisfies Eq. (1). The actual pipeline uses overlapping patches of size sqrt(T) with stride P/10, resizes the stacked matrix to the ViT resolution with nearest interpolation, applies the ViT's own patch embedding, and averages hidden-layer token representations; the proposition says nothing about these operations. Moreover, the cited sample-complexity bound from Li et al. concerns training a shallow ViT from scratch, while TiViT uses frozen pretrained features with linear probing. The proposition can remain as a motivating toy model, but the paper should not claim it explains the empirical gains of TiViT. At minimum, the authors should empirically test the label-relevance condition on real UCR data (e.g., compute the fraction of tokens satisfying Eq. (1) under 1D vs 2D patching for oracle pattern assignments) or clearly restrict the theoretical claim to the simplified setting.
  3. [4.3 (Figure 5), Section 4.4] The paper claims that intermediate layers with high intrinsic dimension are the most effective for time series classification, but this is supported only by visual inspection of the accuracy and intrinsic-dimension curves. No correlation coefficient, rank statistic, or hypothesis test is reported, and the intrinsic-dimension estimates depend on the subsampling ratio (mitigated only partially by the appendix stability check). In addition, the alignment analysis in Table 5 uses the mutual k-NN metric, but the relationship between alignment scores and the reported joint-accuracy gains is not quantified beyond the summary table. These analyses are secondary to the main benchmark claim, but they are presented as contributions and should be either substantiated with quantitative tests or softened.
minor comments (5)
  1. [4.3, Table 4] The statement that TiViT is 'statistically on par' with Mantis on UEA is asserted without reporting a test; please report the p-value or replace 'statistically' with a direct comparison.
  2. [4.5, Table 6] The claims that TiViT 'significantly outperforms' Mantis on the small-training and long-series subsets are made without p-values; add significance tests or soften the wording.
  3. [Appendix C.6, Table A6] The random forest results (TiViT 77.4 vs Mantis 77.5) show that the superiority of TiViT is classifier-dependent; this caveat should be mentioned in the main text, not only in the appendix.
  4. [Author block and references] There are several typos: 'quention.bouniot' in the author email block should be 'quentin.bouniot'; reference [25] spells 'Pattern Recognitio' instead of 'Recognition'; and 'LBFGS' in Section 4 should be 'L-BFGS'.
  5. [3.2] The matrix X' is used both for the stacked real matrix and for the rendered grayscale image in R^{M x P x 3}; please use distinct notation for the three-channel image.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: TiViT's performance claims are direct benchmark evaluations and its theoretical motivation is a conditional result on an external sample-complexity bound.

full rationale

The paper's central empirical claims are computed on the public UCR and UEA test splits with a held-out validation split used for layer selection, so the reported accuracies are not fitted values renamed as predictions. The theoretical motivation (Proposition 1) is an explicitly conditional statement: it assumes that any 2D token containing at least sqrt(k) elements of the label-relevant pattern satisfies the label-relevance inequality, and then counts tokens; this is a stated assumption rather than a concealed equivalence, and the sample-complexity reduction it invokes is imported from Li et al. [31], an external source. The only self-citation of note is the Mantis baseline [18], which shares two co-authors with this paper; however, Mantis is used as a directly evaluated comparison model, not as a citation-based proof of superiority, so the self-citation is not load-bearing. Concerns about multiple-comparison correction for layer selection or about the strength of Proposition 1's assumption are correctness or robustness issues, not circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The empirical value of TiViT depends on external pretrained ViTs and public benchmarks. The theoretical contribution adds a conditional assumption (Eq. 1) and a simplified setup that are not validated on real data. Hyperparameters (P, S, contrast, layer, aggregation) are tuned on the validation split.

free parameters (5)
  • Patch size P = √T (selected via validation on UCR; Table 2a)
    The image transformation stacks segments of length P; P=√T was chosen based on validation accuracy to avoid a 20x search over P.
  • Stride S (patch overlap) = S = P/10 (overlap 0.9)
    Selected from the overlap sweep in Table 2b, yielding the best validation accuracy (80.4 at overlap 0.9).
  • Contrast factor = 0.8
    Implementation detail for rendering grayscale images; no sensitivity analysis is reported.
  • Best hidden layer per ViT backbone = CLIP ViT-H: 14; DINOv2 ViT-L: 15; SigLIP 2 SoViT-400m: 10
    Selected by highest validation accuracy averaged over UCR; used for all subsequent experiments and for UEA.
  • Layer aggregation = Mean over K+1 tokens
    Chosen because it outperforms the CLS token in Table A3; this is a design choice validated on UCR.
assumptions (5)
  • standard math Li et al. sample-complexity theorem: a shallow ViT trained with SGD on a token majority-vote data model requires N = Ω(1/α∗²) samples under their initialization conditions.
    Invoked in Section 3.1 and Appendix A.1 as the bridge between label-relevant token fraction α and sample complexity; the theorem is cited, not re-proven.
  • domain assumption Data model: each time series token is a noisy version of one of M patterns {µ1,...,µM}, and labels are determined by a majority vote over tokens closest to µ1/µ2.
    Used to define label-relevant tokens and derive Prop. 1; real UCR time series may not follow this token-majority-vote model.
  • ad hoc to paper Proposition 1 assumption: 2x′·(µ1−µ2) ≤ ||µ1||²−||µ2||² whenever a 2D token contains at least √k elements of the label-relevant pattern µ2.
    Introduced in the statement of Prop. 1 (Section 3.1) to guarantee that every 2D patch touched by µ2 becomes label-relevant; no evidence is given that real time series satisfy it.
  • ad hoc to paper Simplified setup: T=k², p=k, k=g² so that √k is an integer; 1D tokens are contiguous segments of length k; 2D patches are non-overlapping √k×√k blocks of a k×k matrix.
    Stated as 'for ease of derivations' in Section 3.1; this does not match the TiViT pipeline's arbitrary T, overlapping P=√T segments, and resizing.
  • domain assumption Channel independence: multivariate time series are split into univariate channels, each processed separately by the ViT.
    Adopted from Nie et al. [39] (Section 3.2); ignores inter-channel correlations, which may limit UEA performance. The paper acknowledges this in Limitations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Time Series Representations for Classification Lie Hidden in Pretrained Vision Transformers." pith.science (2026). https://pith.science/paper/SUMDZIVL

@misc{pith2026250608641,
  author       = {Pith},
  title        = {Pith review of: Time Series Representations for Classification Lie Hidden in Pretrained Vision Transformers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SUMDZIVL}},
  note         = {Machine review of arXiv:2506.08641}
}
read the original abstract

Time series classification is a fundamental task in healthcare and industry, yet the development of time series foundation models (TSFMs) remains limited by the scarcity of publicly available time series datasets. In this work, we propose Time Vision Transformer (TiViT), a framework that converts time series into images to leverage the representational power of frozen Vision Transformers (ViTs) pretrained on large-scale image datasets. First, we theoretically motivate our approach by analyzing the 2D patching of ViTs for time series, showing that it can increase the number of label-relevant tokens and reduce the sample complexity. Second, we empirically demonstrate that TiViT achieves state-of-the-art performance on standard time series classification benchmarks by utilizing the hidden representations of large OpenCLIP models. We explore the structure of TiViT representations and find that intermediate layers with high intrinsic dimension are the most effective for time series classification. Finally, we assess the alignment between TiViT and TSFM representation spaces and identify a strong complementarity, with further performance gains achieved by combining their features. Our findings reveal a new direction for reusing vision representations in a non-visual domain. Code is available at https://github.com/ExplainableML/TiViT.

Figures

Figures reproduced from arXiv: 2506.08641 by the authors.

Figure 1
Figure 1. While TSFMs such as Mantis operate directly on the 1D time series signal, TiViT transforms [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Benefits of 2D patching for time series. We consider a binary classification problem with [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Illustration of TiViT on a time series sample from ECG200 [ [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Effect of patch size P on the time series-to-image transformation. To match the ViT input resolution, a small patch size (P = 1) requires horizontal stretching, while a large patch size (P = T 2 ) requires vertical stretching. Both scenarios result in redundant tokens.…
Figure 5
Figure 5. Figure 5: (a) Validation accuracy and (b) Intrinsic dimensionality using hidden representations at [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Evaluation on 155 datasets from the UEA and UCR archive. Model UCR UEA Moment 79.0 69.9 Mantis 80.1 72.4 TiViT (Ours) 81.3 72.0 TiViT + Moment (Ours) 82.5 72.6 TiViT + Mantis (Ours) 83.0 73.7 [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

54 extracted references · 41 canonical work pages

  1. [1]

    Andrzejak, Klaus Lehnertz, Florian Mormann, Christoph Rieke, Peter David, and Christian E

    Ralph G. Andrzejak, Klaus Lehnertz, Florian Mormann, Christoph Rieke, Peter David, and Christian E. Elger. Indications of nonlinear deterministic and finite-dimensional structures in time series of brain electrical activity: Dependence on recording region and brain state. Physical Review E, 64(6):061907, 2001

    work page 2001

  2. [2]

    Reyes-Ortiz

    Davide Anguita, Alessandro Ghio, Luca Oneto, Xavier Parra, and Jorge L. Reyes-Ortiz. A public domain dataset for human activity recognition using smartphones. In 21st European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, pages 437–442, 2013

    work page 2013

  3. [3]

    Maddix, Hao Wang, Michael W

    Abdul Fatir Ansari, Lorenzo Stella, Ali Caner Turkmen, Xiyuan Zhang, Pedro Mercado, Huibin Shen, Oleksandr Shchur, Syama Sundar Rangapuram, Sebastian Pineda Arango, Shubham Kapoor, Jasper Zschiegner, Danielle C. Maddix, Hao Wang, Michael W. Mahoney, Kari Torkkola, Andrew Gordon Wilson, Michael Bohlke-Schneider, and Bernie Wang. Chronos: Learning the langu...

    work page 2024

  4. [4]

    The uea multivariate time series classification archive, 2018

    Anthony Bagnall, Hoang Anh Dau, Jason Lines, Michael Flynn, James Large, Aaron Bostrom, Paul Southam, and Eamonn Keogh. The uea multivariate time series classification archive, 2018. arXiv preprint arXiv:1811.00075, 2018

    arXiv 2018

  5. [5]

    Mamba4Cast: Efficient zero-shot time series forecasting with state space models

    Sathya Kamesh Bhethanabhotla, Omar Swelam, Julien Siems, David Salinas, and Frank Hutter. Mamba4Cast: Efficient zero-shot time series forecasting with state space models. arXiv preprint arXiv:2410.09385, 2024

    arXiv 2024

  6. [6]

    Arik, Tomas Pfister, Yixiang Zheng, Wen Ye, and Yan Liu

    Defu Cao, Furong Jia, Sercan O. Arik, Tomas Pfister, Yixiang Zheng, Wen Ye, and Yan Liu. TEMPO: Prompt-based generative pre-trained transformer for time series forecasting. In The Twelfth International Conference on Learning Representations, 2024

    work page 2024

  7. [7]

    LLM4TS: Aligning pre-trained llms as data-efficient time-series forecasters

    Ching Chang, Wei-Yao Wang, Wen-Chih Peng, and Tien-Fu Chen. LLM4TS: Aligning pre-trained llms as data-efficient time-series forecasters. ACM Transactions on Intelligent Systems and Technology, 16(3), 2025. 10

    work page 2025

  8. [8]

    VisionTS: Vi- sual masked autoencoders are free-lunch zero-shot time series forecasters.arXiv preprint arXiv:2408.17253, 2024

    Mouxiang Chen, Lefei Shen, Zhuo Li, Xiaoyun Joy Wang, Jianling Sun, and Chenghao Liu. VisionTS: Vi- sual masked autoencoders are free-lunch zero-shot time series forecasters.arXiv preprint arXiv:2408.17253, 2024

    arXiv 2024

Show all 54 references
  1. [9]

    Reproducible scaling laws for contrastive language-image learning

    Mehdi Cherti, Romain Beaumont, Ross Wightman, Mitchell Wortsman, Gabriel Ilharco, Cade Gordon, Christoph Schuhmann, Ludwig Schmidt, and Jenia Jitsev. Reproducible scaling laws for contrastive language-image learning. In Proceedings of the IEEE/CVF Conference on Computer Vision...

    2023

  2. [10]

    Clifford, Chengyu Liu, Benjamin Moody, Li-wei H

    Gari D. Clifford, Chengyu Liu, Benjamin Moody, Li-wei H. Lehman, Ikaro Silva, Qiao Li, Alistair E. Johnson, and Roger G. Mark. AF classification from a short single lead ECG recording: The Phys- ioNet/computing in cardiology challenge 2017. In 2017 Computing in Cardiology (Cin...

    2017

  3. [11]

    A decoder-only foundation model for time-series forecasting

    Abhimanyu Das, Weihao Kong, Rajat Sen, and Yichen Zhou. A decoder-only foundation model for time-series forecasting. In Proceedings of the 41st International Conference on Machine Learning, pages 10148–10167, 2024

    2024

  4. [12]

    The ucr time series archive

    Hoang Anh Dau, Anthony Bagnall, Kaveh Kamgar, Chin-Chia Michael Yeh, Yan Zhu, Shaghayegh Gharghabi, Chotirat Ann Ratanamahatana, and Eamonn Keogh. The ucr time series archive. IEEE/CAA Journal of Automatica Sinica, 6(6):1293–1305, 2019

    2019

  5. [13]

    ImageNet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. ImageNet: A large-scale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition, pages 248–255, 2009

    2009

  6. [14]

    Heller, and Chris Donahue

    Satvik Dixit, Laurie M. Heller, and Chris Donahue. Vision language models are few-shot audio spectrogram classifiers. In NeurIPS Audio Imagination Workshop, 2024

    2024

  7. [15]

    From colors to classes: Emergence of concepts in vision transformers

    Teresa Dorszewski, Lenka Tˇetková, Robert Jenssen, Lars Kai Hansen, and Kristoffer Knutsen Wickstrøm. From colors to classes: Emergence of concepts in vision transformers. arXiv preprint arXiv:2503.24071, 2025

    2025 arXiv

  8. [16]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at...

    2021

  9. [17]

    Estimating the intrinsic dimension of datasets by a minimal neighborhood information

    Elena Facco, Maria d’Errico, Alex Rodriguez, and Alessandro Laio. Estimating the intrinsic dimension of datasets by a minimal neighborhood information. Scientific Reports, 7:12140, 2017

    2017

  10. [18]

    Mantis: Lightweight calibrated foundation model for user-friendly time series classification

    Vasilii Feofanov, Songkang Wen, Marius Alonso, Romain Ilbert, Hongbo Guo, Malik Tiomoko, Lujia Pan, Jianfeng Zhang, and Ievgen Redko. Mantis: Lightweight calibrated foundation model for user-friendly time series classification. arXiv preprint arXiv:2502.15637, 2025

    2025 arXiv

  11. [19]

    UniTS: A unified multi-task time series model

    Shanghua Gao, Teddy Koker, Owen Queen, Thomas Hartvigsen, Theodoros Tsiligkaridis, and Marinka Zitnik. UniTS: A unified multi-task time series model. In Advances in Neural Information Processing Systems, pages 140589–140631, 2024

    2024

  12. [20]

    DADApy: Distance-based analysis of data-manifolds in python

    Aldo Glielmo, Iuri Macocco, Diego Doimo, Matteo Carli, Claudio Zeni, Romina Wild, Maria d’Errico, Alex Rodriguez, and Alessandro Laio. DADApy: Distance-based analysis of data-manifolds in python. Patterns, 3(10):100589, 2022

    2022

  13. [21]

    Goldberger, Luis A

    Ary L. Goldberger, Luis A. N. Amaral, Leon Glass, Jeffrey M. Hausdorff, Plamen Ch. Ivanov, Roger G. Mark, Joseph E. Mietus, George B. Moody, Chung-Kang Peng, and H. Eugene Stanley. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex phys...

    2000

  14. [22]

    MOMENT: A family of open time-series foundation models

    Mononito Goswami, Konrad Szafer, Arjun Choudhry, Yifu Cai, Shuo Li, and Artur Dubrawski. MOMENT: A family of open time-series foundation models. In Proceedings of the 41st International Conference on Machine Learning, pages 16115–16152, 2024

    2024

  15. [23]

    Large language models are zero shot time series forecasters

    Nate Gruver, Marc Finzi, Shikai Qiu, and Andrew Gordon Wilson. Large language models are zero shot time series forecasters. In Advances in Neural Information Processing Systems, 2023

    2023

  16. [24]

    Momentum contrast for unsupervised visual representation learning

    Kaiming He, Haoqi Fan, Yuxin Wu, Saining Xie, and Ross Girshick. Momentum contrast for unsupervised visual representation learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9729–9738, 2020. 11

    2020

  17. [25]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, and Ross Girshick. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognitio, pages 16000–16009, 2022

    2022

  18. [26]

    Position: The platonic representation hypothesis

    Minyoung Huh, Brian Cheung, Tongzhou Wang, and Phillip Isola. Position: The platonic representation hypothesis. In Proceedings of the 41st International Conference on Machine Learning, pages 20617–20642, 2024

    2024

  19. [27]

    OpenCLIP, 2021

    Gabriel Ilharco, Mitchell Wortsman, Ross Wightman, Cade Gordon, Nicholas Carlini, Rohan Taori, Achal Dave, Vaishaal Shankar, Hongseok Namkoong, John Miller, Hannaneh Hajishirzi, Ali Farhadi, and Ludwig Schmidt. OpenCLIP, 2021. Version 0.1. URL: https://doi.org/10.5281/zenodo.5143773

    2021 doi

  20. [28]

    Time-LLM: Time series forecasting by reprogram- ming large language models

    Ming Jin, Shiyu Wang, Lintao Ma, Zhixuan Chu, James Y Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, and Qingsong Wen. Time-LLM: Time series forecasting by reprogram- ming large language models. In International Conference on Learning Representations, 2024

    2024

  21. [29]

    Zwinderman, Bert Tuk, Hilbert A

    Bastiaan Kemp, Aeilko H. Zwinderman, Bert Tuk, Hilbert A. C. Kamphuisen, and Josefien J. L. Oberye. Analysis of a sleep-dependent neuronal feedback loop: the slow-wave microcontinuity of the EEG. IEEE Transactions on Biomedical Engineering, 47(9):1185–1194, 2000

    2000

  22. [30]

    Kimotho, Detmar Zimmer, and Walter Sextro

    Christian Lessmeier, James K. Kimotho, Detmar Zimmer, and Walter Sextro. Condition monitoring of bearing damage in electromechanical drive systems by using motor current signals of electric motors: A benchmark data set for data-driven classification. PHM Society European Confe...

    2016

  23. [31]

    A theoretical understanding of shallow vision transformers: Learning, generalization, and sample complexity

    Hongkang Li, Meng Wang, Sijia Liu, and Pin-Yu Chen. A theoretical understanding of shallow vision transformers: Learning, generalization, and sample complexity. In International Conference on Learning Representations, 2023

    2023

  24. [32]

    Time series as images: Vision transformer for irregularly sampled time series

    Zekun Li, Shiyang Li, and Xifeng Yan. Time series as images: Vision transformer for irregularly sampled time series. In Advances in Neural Information Processing Systems, pages 49187–49204, 2023

    2023

  25. [33]

    NuTime: Numerically multi-scaled embedding for large-scale time-series pretraining

    Chenguo Lin, Xumeng Wen, Wei Cao, Congrui Huang, Jiang Bian, Stephen Lin, and Zhirong Wu. NuTime: Numerically multi-scaled embedding for large-scale time-series pretraining. arXiv preprint arXiv:2310.07402, 2023

    2023 arXiv

  26. [34]

    SparseTSF: Modeling long-term time series forecasting with 1k parameters

    Shengsheng Lin, Weiwei Lin, Wentai Wu, Haojun Chen, and Junjie Yang. SparseTSF: Modeling long-term time series forecasting with 1k parameters. InProceedings of the 41st International Conference on Machine Learning, pages 30211–30226, 2024

    2024

  27. [35]

    uWave: Accelerometer-based personalized gesture recognition and its applications

    Jiayang Liu, Lin Zhong, Jehan Wickramasuriya, and Venu Vasudevan. uWave: Accelerometer-based personalized gesture recognition and its applications. In 2009 IEEE International Conference on Pervasive Computing and Communications, 2009

    2009

  28. [36]

    Moirai-MoE: Empowering time series foundation models with sparse mixture of experts

    Xu Liu, Juncheng Liu, Gerald Woo, Taha Aksu, Yuxuan Liang, Roger Zimmermann, Chenghao Liu, Silvio Savarese, Caiming Xiong, and Doyen Sahoo. Moirai-MoE: Empowering time series foundation models with sparse mixture of experts. arXiv preprint arXiv:2410.10469, 2024

    arXiv

  29. [37]

    Timer: Generative pre-trained transformers are large time series models

    Yong Liu, Haoran Zhang, Chenyu Li, Xiangdong Huang, Jianmin Wang, and Mingsheng Long. Timer: Generative pre-trained transformers are large time series models. In Proceedings of the 41st International Conference on Machine Learning, pages 32369–32399, 2024

    2024

  30. [38]

    Harnessing vision models for time series analysis: A survey

    Jingchao Ni, Ziming Zhao, ChengAo Shen, Hanghang Tong, Dongjin Song, Wei Cheng, Dongsheng Luo, and Haifeng Chen. Harnessing vision models for time series analysis: A survey. arXiv preprint arXiv:2502.08869, 2025

    2025 arXiv

  31. [39]

    A time series is worth 64 words: Long-term forecasting with transformers

    Yuqi Nie, Nam H Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. A time series is worth 64 words: Long-term forecasting with transformers. In International Conference on Learning Representations, 2023

    2023

  32. [40]

    Generalized feature extraction for structural pattern recognition in time-series data

    Robert Thomas Olszewski. Generalized feature extraction for structural pattern recognition in time-series data. PhD thesis, Carnegie Mellon University, 2001

    2001

  33. [41]

    Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy V . V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel HAZIZA, Francisco Massa, Alaaeldin El-Nouby, Mido Assran, Nicolas Ballas, Wojciech Galuba, Russell Howes, Po-Yao Huang, Shang-Wen Li, Ishan Misra, Michael Rab...

    2024

  34. [42]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision. In Proceedings of th...

    2021

  35. [43]

    Lag-Llama: Towards foundation models for probabilistic time series forecasting

    Kashif Rasul, Arjun Ashok, Andrew Robert Williams, Hena Ghonia, Rishika Bhagwatkar, Arian Kho- rasani, Mohammad Javad Darvishi Bayazi, George Adamopoulos, Roland Riachi, Nadhir Hassen, Marin Biloš, Sahil Garg, Anderson Schneider, Nicolas Chapados, Alexandre Drouin, Valentina Z...

    arXiv

  36. [44]

    LAION-5B: An open large-scale dataset for training next generation image-text models

    Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade W Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Wortsman, Patrick Schramowski, Srivatsa R Kundurthy, Katherine Crowson, Ludwig Schmidt, Robert Kaczmarczyk, and Jenia Jitsev. L...

    2022

  37. [45]

    SigLIP 2: Multilingual vision-language encoders with improved semantic understanding, localization, and dense features

    Michael Tschannen, Alexey Gritsenko, Xiao Wang, Muhammad Ferjad Naeem, Ibrahim Alabdulmohsin, Nikhil Parthasarathy, Talfan Evans, Lucas Beyer, Ye Xia, Basil Mustafa, Olivier Hénaff, Jeremiah Harmsen, Andreas Steiner, and Xiaohua Zhai. SigLIP 2: Multilingual vision-language enc...

    2025 arXiv

  38. [46]

    The geometry of hidden representations of large transformer models

    Lucrezia Valeriani, Diego Doimo, Francesca Cuturello, Alessandro Laio, Alessio Ansuini, and Alberto Cazzaniga. The geometry of hidden representations of large transformer models. In Advances in Neural Information Processing Systems, pages 51234–51252, 2023

    2023

  39. [47]

    Attention is all you need.Advances in Neural Information Processing Systems, pages 5998–6008, 2017

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need.Advances in Neural Information Processing Systems, pages 5998–6008, 2017

    2017

  40. [48]

    ROSE: Register assisted general time series forecasting with decomposed frequency learning

    Yihang Wang, Yuying Qiu, Peng Chen, Kai Zhao, Yang Shu, Zhongwen Rao, Lujia Pan, Bin Yang, and Chenjuan Guo. ROSE: Register assisted general time series forecasting with decomposed frequency learning. arXiv preprint arXiv:2405.17478, 2024

    2024 arXiv

  41. [49]

    TimesNet: Tem- poral 2d-variation modeling for general time series analysis

    Haixu Wu, Tengge Hu, Yong Liu, Hang Zhou, Jianmin Wang, and Mingsheng Long. TimesNet: Tem- poral 2d-variation modeling for general time series analysis. In International Conference on Learning Representations, 2023

    2023

  42. [50]

    SonicVisionLM: Playing sound with vision language models

    Zhifeng Xie, Shengye Yu, Qile He, and Mengtian Li. SonicVisionLM: Playing sound with vision language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26866–26875, 2024

    2024

  43. [51]

    Hao Xue and Flora D. Salim. PromptCast: A new prompt-based learning paradigm for time series forecasting. IEEE Transactions on Knowledge and Data Engineering, 36(11):6851–6864, 2024

    2024

  44. [52]

    Sigmoid loss for language image pre-training

    Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, and Lucas Beyer. Sigmoid loss for language image pre-training. In 2023 IEEE/CVF International Conference on Computer Vision, pages 11941–11952, 2023

    2023

  45. [53]

    Davison, Hui Ren, Jing Huang, Chen Chen, Yuyin Zhou, Sunyang Fu, Wei Liu, Tianming Liu, Xiang Li, Yong Chen, Lifang He, James Zou, Quanzheng Li, Hongfang Liu, and Lichao Sun

    Kai Zhang, Rong Zhou, Eashan Adhikarla, Zhiling Yan, Yixin Liu, Jun Yu, Zhengliang Liu, Xun Chen, Brian D. Davison, Hui Ren, Jing Huang, Chen Chen, Yuyin Zhou, Sunyang Fu, Wei Liu, Tianming Liu, Xiang Li, Yong Chen, Lifang He, James Zou, Quanzheng Li, Hongfang Liu, and Lichao ...

    2024

  46. [54]

    One fits all: Power general time series analysis by pretrained lm

    Tian Zhou, Peisong Niu, Liang Sun, and Rong Jin. One fits all: Power general time series analysis by pretrained lm. In Advances in Neural Information Processing Systems, volume 36, pages 43322–43355, 2023. 13 Appendix In Section A, we summarize the theoretical analysis of Li e...

    2023

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.