pith. sign in

arxiv: 2407.13278 · v3 · submitted 2024-07-18 · 💻 cs.LG

Deep Time Series Models: A Comprehensive Survey and Benchmark

Yuxuan Wang , Haixu Wu , Jiaxiang Dong , Yong Liu , Chen Wang , Mingsheng Long , Jianmin Wang This is my paper

Pith reviewed 2026-05-23 23:02 UTC · model grok-4.3

classification 💻 cs.LG
keywords deep time series modelsbenchmarksurveytime series analysisTSLibforecastinganomaly detectionmodel evaluation
0
0 comments X

The pith

A benchmark of 41 deep time series models across 30 datasets shows each structure fits only particular analysis tasks.

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

This paper surveys deep time series models by reviewing their basic modules and overall architectures. It builds and releases TSLib, a library that implements 41 models, covers 30 datasets from varied domains, and supports five standard analysis tasks. Evaluations of sixteen popular models plus six foundation models produce the finding that models with certain structures succeed only on distinct tasks. A reader would care because time series data drives decisions in forecasting, anomaly detection, and classification across many fields, and mismatched models waste effort or reduce accuracy.

Core claim

Through TSLib the authors establish that deep time series models with specific structures are apt only at distinct analytical tasks, rather than performing uniformly across forecasting, imputation, anomaly detection, classification, and related problems.

What carries the argument

TSLib, the library that standardizes implementations of 41 deep time series models for comparison on 30 datasets and 5 tasks.

If this is right

  • Model selection for a given task should prioritize structures shown to match that task in the benchmark.
  • Design efforts can target the module combinations that align with particular task demands.
  • Adoption decisions gain clearer guidance from task-specific rankings rather than overall averages.
  • Both small-scale and large foundation models can be compared under the same standardized conditions.
  • The library itself supplies a reusable platform for adding new models and datasets.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Task-specific model choice may reduce the need for ever-larger general models in practice.
  • The structure-task matches could weaken if future datasets include heavier noise or irregular sampling not present in the current 30.
  • Hybrid architectures that combine modules from different structures might cover multiple tasks more efficiently than single-structure models.
  • The benchmark results invite direct tests on domain-specific data such as medical signals or financial ticks to check transfer of the observed patterns.

Load-bearing premise

The chosen 30 datasets and 5 tasks represent real-world time series problems, and the 41 model implementations accurately reproduce the originals.

What would settle it

A re-run of the same evaluations on a new collection of datasets or with corrected model code that removes the observed task-specific performance patterns would falsify the central empirical claim.

Figures

Figures reproduced from arXiv: 2407.13278 by Chen Wang, Haixu Wu, Jianmin Wang, Jiaxiang Dong, Mingsheng Long, Yong Liu, Yuxuan Wang.

Figure 1
Figure 1. Figure 1: Schematic illustration of different time series analysis tasks. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Matrix factorization for multiple time series. b [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: An overview of representative time series models in chronological order. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of a basic MLP-based model in forecasting task, which captures the future-past dependencies with learnable MLP parameters. attempted to utilize MLPs to model both temporal and variate dependencies. TSMixer [96] contains interleaving time-mixing and feature-mixing MLPs to extract information from different perspectives. To better model the global dependencies in time series data, FreTS [90] inv… view at source ↗
Figure 5
Figure 5. Figure 5: Illustration of RNN-based model in the forecasting task. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of existing CNN-based time series models from the [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Illustration of modeling multivariate time series using GNN. The core goal of GNN architecture is to model the un￾derlying topological relations in multivariate data, therefore existing GNN-based works can be roughly divided into two categories based on whether graph structure is part of the input into the model. DCRNN [138] models the spatial dependency of traffic as a diffusion process on a directed grap… view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of different types of tokenization used for [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Architecture and experiment pipeline of Time Series Library. Left: Unified training and evaluation process. Right: Overall Architecture. Data Source Our TSLib provides extensive support for a wide range of diverse and multi-type datasets in a variety of formats, including ”.csv”, ”.npz”, ”.txt”, etc. As shown in [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Comparison of model performance in Time Series Library. Full results are averaged from a diverse set of datasets supported by TSLib [PITH_FULL_IMAGE:figures/full_fig_p013_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Overall performance of models with different deep architectures. [PITH_FULL_IMAGE:figures/full_fig_p013_11.png] view at source ↗
read the original abstract

Time series, characterized by a sequence of data points organized in a discrete-time order, are ubiquitous in real-world scenarios. Unlike other data modalities, time series present unique challenges in learning and modeling due to their intricate and dynamic nature, including the entanglement of nonlinear patterns and time-variant trends. Recent years have witnessed remarkable breakthroughs in time series analysis, with techniques shifting from traditional statistical methods to contemporary deep learning models. In this paper, we delve into the design of deep time series models across various analysis tasks and review the existing literature from two perspectives: basic modules and model architectures. Further, we develop and release Time Series Library (TSLib) as a fair benchmark of deep time series models for diverse analysis tasks. TSLib implements 41 prominent models, including both small- and large-scale time series models, covers 30 datasets from different domains, and supports 5 prevalent analysis tasks. Based on TSLib, we evaluate 16 popular deep time series models and 6 advanced time series foundation models. Empirical findings indicate that models with specific structures are apt only at distinct analytical tasks, providing insights for research and adoption of deep time series models. Code and datasets are available at https://github.com/thuml/Time-Series-Library.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript surveys deep time series models from the perspectives of basic modules and model architectures. It introduces and releases the Time Series Library (TSLib), which implements 41 prominent models (small- and large-scale), covers 30 datasets from multiple domains, and supports 5 analysis tasks. Using TSLib, the authors evaluate 16 popular deep time series models plus 6 foundation models and report the empirical finding that models with specific structures are apt only for distinct analytical tasks.

Significance. The open release of TSLib together with code and datasets constitutes a concrete contribution that can standardize future comparisons in the field. If the reported evaluations prove reliable, the task-specific performance patterns supply actionable guidance for model selection and adoption. The dual survey lens on modules versus architectures also organizes the literature in a potentially useful way.

major comments (2)
  1. [TSLib description and evaluation sections] The TSLib description states that the library 'implements 41 prominent models' and that the evaluations rest on these implementations, yet supplies no protocol for code review, architecture matching, training-procedure replication, or hyper-parameter selection that would confirm fidelity to the original papers. This verification step is load-bearing for the central empirical claim that specific structures suit only distinct tasks.
  2. [Dataset and task coverage] The manuscript asserts that the 30 datasets and 5 tasks are representative but provides no explicit selection criteria, domain-coverage argument, or statistical justification for why performance patterns observed on these resources generalize. This assumption directly supports the generalization of the task-specific aptness finding.
minor comments (2)
  1. The abstract summarizes the main finding without any quantitative performance numbers or statistical test results from the 22-model evaluation.
  2. A compact table listing the 41 models by category, scale, and original reference would improve readability of the benchmark contribution.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our survey and benchmark paper. We address each major comment point by point below, proposing revisions where appropriate to strengthen the manuscript.

read point-by-point responses

  1. Referee: The TSLib description states that the library 'implements 41 prominent models' and that the evaluations rest on these implementations, yet supplies no protocol for code review, architecture matching, training-procedure replication, or hyper-parameter selection that would confirm fidelity to the original papers. This verification step is load-bearing for the central empirical claim that specific structures suit only distinct tasks.

    Authors: We acknowledge that an explicit replication protocol in the paper would improve transparency and verifiability of the empirical results. The full TSLib implementation is already publicly released at https://github.com/thuml/Time-Series-Library, enabling direct code inspection. In the revised manuscript, we will add a dedicated subsection under the TSLib description that details: (i) the architecture-matching process against original papers, (ii) training-procedure replication steps, and (iii) hyper-parameter selection criteria used for the 16 models and 6 foundation models. This will directly address concerns about fidelity to the source works. revision: yes

  2. Referee: The manuscript asserts that the 30 datasets and 5 tasks are representative but provides no explicit selection criteria, domain-coverage argument, or statistical justification for why performance patterns observed on these resources generalize. This assumption directly supports the generalization of the task-specific aptness finding.

    Authors: We agree that more explicit justification is warranted to support generalization claims. The datasets were chosen as standard benchmarks widely adopted in prior time series literature, spanning domains including energy, transportation, weather, and finance, with diversity in length, frequency, and characteristics. The five tasks are the most common in the field. In the revision, we will expand the dataset and task coverage section (or add an appendix) with explicit selection criteria, references to prior surveys using overlapping collections, and a qualitative argument for domain coverage to better substantiate why the task-specific patterns are expected to generalize. revision: yes

Circularity Check

0 steps flagged

No circularity: survey + independent benchmark with new empirical runs on external data

full rationale

The paper performs a literature review of deep time series models and then releases TSLib, which implements 41 models, covers 30 external datasets across 5 tasks, and reports fresh evaluation results for 22 models. The central empirical claim (structure-specific aptitude for distinct tasks) is produced by these new runs rather than any derivation, fitted parameter, or self-citation chain. No equations, ansatzes, uniqueness theorems, or predictions appear that could reduce to the paper's own inputs by construction. The benchmark is self-contained against external datasets and prior model papers; fidelity concerns are experimental-validity issues, not circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the representativeness of the chosen datasets and tasks plus faithful model implementations; no free parameters are fitted as this is an evaluation study rather than a modeling derivation.

axioms (2)
  • domain assumption The five analysis tasks and thirty datasets adequately represent the range of real-world time series problems.
    This assumption underpins the generalizability of the task-specific performance findings.
  • domain assumption Deep time series models can be usefully decomposed into basic modules and overall architectures for review purposes.
    This structures the literature review but is a conventional organizational choice.

pith-pipeline@v0.9.0 · 5764 in / 1461 out tokens · 35899 ms · 2026-05-23T23:02:19.762725+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 21 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Olivia: Harmonizing Time Series Foundation Models with Power Spectral Density

    cs.LG 2026-05 unverdicted novelty 7.0

    Olivia harmonizes time series datasets via normalized power spectral density using a Harmonizer module and resonator-based HarmonicAttention, achieving state-of-the-art zero-shot, few-shot, and full-shot forecasting o...

  2. LeapTS: Rethinking Time Series Forecasting as Adaptive Multi-Horizon Scheduling

    cs.LG 2026-05 unverdicted novelty 7.0

    LeapTS reformulates forecasting as adaptive multi-horizon scheduling via hierarchical control and NCDEs, delivering at least 7.4% better performance and 2.6-5.3x faster inference than Transformer baselines while adapt...

  3. CastFlow: Learning Role-Specialized Agentic Workflows for Time Series Forecasting

    cs.LG 2026-04 unverdicted novelty 7.0

    CastFlow introduces a role-specialized agentic workflow with memory retrieval and multi-view toolkit for iterative ensemble time series forecasting, using two-stage SFT+RLVR training on a domain-specific LLM to outper...

  4. Adaptive Conformal Anomaly Detection with Time Series Foundation Models for Signal Monitoring

    cs.LG 2026-04 unverdicted novelty 7.0

    A model-agnostic adaptive conformal anomaly detection approach uses weighted quantile bounds learned from past foundation model predictions to deliver interpretable p-value scores with stable calibration under shifts ...

  5. GenTS: A Comprehensive Benchmark Library for Generative Time Series Models

    cs.LG 2026-05 unverdicted novelty 6.0

    GenTS is a modular benchmark library providing unified data pipelines, generative models, and evaluation metrics for time series synthesis, forecasting, and imputation, with open-source code and initial benchmarking e...

  6. Partial Effective Information Decomposition for Synergistic Causality

    stat.ML 2026-05 unverdicted novelty 6.0

    PEID decomposes the causal effect of multiple sources on a target under maximum-entropy interventions into unique and synergistic information, enabling hyperedge causal graphs and downward causation analysis.

  7. Timer-S1: A Billion-Scale Time Series Foundation Model with Serial Scaling

    cs.AI 2026-03 unverdicted novelty 6.0

    Timer-S1 is a released 8.3B-parameter MoE time series model that achieves state-of-the-art MASE and CRPS scores on GIFT-Eval using serial scaling and Serial-Token Prediction.

  8. DeMa: Dual-Path Delay-Aware Mamba for Efficient Multivariate Time Series Analysis

    cs.LG 2026-01 unverdicted novelty 6.0

    DeMa is a dual-path delay-aware Mamba architecture that decomposes MTS into intra-series temporal and inter-series variate paths to achieve SOTA performance with linear complexity on forecasting, imputation, anomaly d...

  9. Neural CDEs as Correctors for Learned Time Series Models

    cs.LG 2025-12 unverdicted novelty 6.0

    Neural CDEs serve as correctors that reduce error accumulation in multi-step forecasts from learned time-series models across synthetic, physics, and real-world data.

  10. Delta-XAI: A Unified Framework for Explaining Prediction Changes in Online Time Series Monitoring

    cs.LG 2025-11 unverdicted novelty 6.0

    Delta-XAI wraps existing XAI methods for online time series and introduces SWING to explain prediction changes while accounting for temporal dependencies.

  11. AlphaCast: A Human Wisdom-LLM Intelligence Co-Reasoning Framework for Interactive Time Series Forecasting

    cs.AI 2025-11 conditional novelty 6.0

    AlphaCast is a training-free LLM framework that performs interactive multi-stage reasoning for time series forecasting by integrating feature extraction, knowledge bases, case libraries, and contextual pools.

  12. DistDF: Time-Series Forecasting Needs Joint-Distribution Wasserstein Alignment

    cs.LG 2025-10 unverdicted novelty 6.0

    DistDF replaces biased MSE training in time-series forecasting with a tractable joint Wasserstein discrepancy that upper-bounds the desired conditional distributional alignment.

  13. PeakFocus: Bridging Peak Localization and Intensity Regression via a Unified Multi-Scale Framework for Electricity Load Forecasting

    cs.LG 2026-05 unverdicted novelty 5.0

    PeakFocus introduces a unified multi-scale framework with hybrid loss, peak locator, and location-aware decoder to jointly predict peak timing and intensity in electricity load forecasting.

  14. Quantifying the Pre-training Dividend: Generative versus Latent Self-Supervised Learning for Time Series Foundation Models

    cs.LG 2026-05 unverdicted novelty 5.0

    Self-supervised pre-training delivers large gains up to 375% on time series anomaly detection and classification but only marginal benefits for forecasting, driven by a precision-invariance trade-off in the learned re...

  15. TSNN: A Non-parametric and Interpretable Framework for Traffic Time Series Forecasting

    cs.LG 2026-05 unverdicted novelty 5.0

    TSNN matches time series entries to a training-derived memory bank to forecast traffic without any trainable parameters and achieves competitive accuracy on four real-world datasets.

  16. TimeRFT: Stimulating Generalizable Time Series Forecasting for TSFMs via Reinforcement Finetuning

    eess.SP 2026-04 unverdicted novelty 5.0

    TimeRFT applies reinforcement learning with multi-faceted step-wise rewards and informative sample selection to improve generalization and accuracy in TSFM adaptation beyond supervised fine-tuning.

  17. MambaSL: Exploring Single-Layer Mamba for Time Series Classification

    cs.LG 2026-04 unverdicted novelty 5.0

    A single-layer Mamba variant with targeted redesigns sets new state-of-the-art average performance on all 30 UEA time series classification datasets under a unified reproducible protocol.

  18. MSDformer: Multi-scale Discrete Transformer For Time Series Generation

    cs.LG 2025-05 unverdicted novelty 5.0

    MSDformer introduces a multi-scale discrete transformer that tokenizes time series at multiple scales and models them autoregressively in discrete space, claiming superior performance over prior DTM methods with rate-...

  19. FinTSB: A Comprehensive and Practical Benchmark for Financial Time Series Forecasting

    cs.CE 2025-02 unverdicted novelty 5.0

    FinTSB introduces a benchmark addressing diversity, standardization, and real-world applicability gaps in financial time series forecasting evaluations.

  20. CASE-NET: Deep Spatio-Temporal Representation Learning via Causal Attention and Channel Recalibration for Multivariate Time Series Classification

    cs.LG 2026-05 unverdicted novelty 4.0

    CASE-NET combines a causal temporal encoder with adaptive channel recalibration and reports new state-of-the-art accuracy on four of six evaluated multivariate time series tasks.

  21. Universal Time-Series Representation Learning: A Survey

    cs.LG 2024-01 unverdicted novelty 3.0

    A survey that proposes a taxonomy for universal time-series representation learning and reviews existing deep learning studies along with experimental setups.

Reference graph

Works this paper leans on

214 extracted references · 214 canonical work pages · cited by 21 Pith papers · 22 internal anchors

  1. [1]

    G. E. Box, G. M. Jenkins, G. C. Reinsel, and G. M. Ljung, Time series analysis: forecasting and control. John Wiley & Sons, 2015

    work page 2015

  2. [2]

    J. D. Hamilton, Time series analysis. Princeton university press, 2020

    work page 2020

  3. [3]

    Timesnet: Temporal 2d-variation modeling for general time series analysis,

    H. Wu, T. Hu, Y. Liu, H. Zhou, J. Wang, and M. Long, “Timesnet: Temporal 2d-variation modeling for general time series analysis,” in ICLR, 2023

    work page 2023

  4. [4]

    Trafficbert: Pre-trained model with large-scale data for long-range traffic flow forecasting,

    K. Jin, J. Wi, E. Lee, S. Kang, S. Kim, and Y. Kim, “Trafficbert: Pre-trained model with large-scale data for long-range traffic flow forecasting,” Expert Systems with Applications, vol. 186, p. 115738, 2021

    work page 2021

  5. [5]

    Anomaly transformer: Time series anomaly detection with association discrepancy,

    J. Xu, H. Wu, J. Wang, and M. Long, “Anomaly transformer: Time series anomaly detection with association discrepancy,” in ICLR, 2022

    work page 2022

  6. [6]

    Interpretable weather forecasting for worldwide stations with a unified deep model,

    H. Wu, H. Zhou, M. Long, and J. Wang, “Interpretable weather forecasting for worldwide stations with a unified deep model,” Nature Machine Intelligence, pp. 1–10, 2023

    work page 2023

  7. [7]

    L. H. Koopmans, The spectral analysis of time series . Elsevier, 1995

    work page 1995

  8. [8]

    Deep learning for time series classification: a review,

    H. Ismail Fawaz, G. Forestier, J. Weber, L. Idoumghar, and P .-A. Muller, “Deep learning for time series classification: a review,” Data mining and knowledge discovery , vol. 33, no. 4, pp. 917–963, 2019

    work page 2019

  9. [9]

    Anomaly detection in univariate time-series: A survey on the state-of-the-art,

    M. Braei and S. Wagner, “Anomaly detection in univariate time-series: A survey on the state-of-the-art,” arXiv preprint arXiv:2004.00433, 2020

    work page arXiv 2004

  10. [10]

    Deep learning for time series forecasting: a survey,

    J. F. Torres, D. Hadjout, A. Sebaa, F. Mart ´ınez-´Alvarez, and A. Troncoso, “Deep learning for time series forecasting: a survey,” Big Data, vol. 9, no. 1, pp. 3–21, 2021

    work page 2021

  11. [11]

    A re- view on outlier/anomaly detection in time series data,

    A. Bl´azquez-Garc´ıa, A. Conde, U. Mori, and J. A. Lozano, “A re- view on outlier/anomaly detection in time series data,” Computing Surveys, vol. 54, no. 3, pp. 1–33, 2021

    work page 2021

  12. [12]

    Transformers in time series: A survey,

    Q. Wen, T. Zhou, C. Zhang, W. Chen, Z. Ma, J. Yan, and L. Sun, “Transformers in time series: A survey,” arXiv preprint arXiv:2202.07125, 2022

    work page arXiv 2022

  13. [13]

    A survey on graph neural networks for time series: Forecasting, classification, imputation, and anomaly detection,

    M. Jin, H. Y. Koh, Q. Wen, D. Zambon, C. Alippi, G. I. Webb, I. King, and S. Pan, “A survey on graph neural networks for time series: Forecasting, classification, imputation, and anomaly detection,” arXiv preprint arXiv:2307.03759, 2023

    work page arXiv 2023

  14. [14]

    Exploring progress in multivariate time se- ries forecasting: Comprehensive benchmarking and heterogeneity analysis,

    Z. Shao, F. Wang, Y. Xu, W. Wei, C. Yu, Z. Zhang, D. Yao, G. Jin, X. Cao, G. Cong et al., “Exploring progress in multivariate time se- ries forecasting: Comprehensive benchmarking and heterogeneity analysis,” arXiv preprint arXiv:2310.06119, 2023

    work page arXiv 2023

  15. [15]

    Tfb: Towards comprehensive and fair benchmarking of time series forecasting methods,

    X. Qiu, J. Hu, L. Zhou, X. Wu, J. Du, B. Zhang, C. Guo, A. Zhou, C. S. Jensen, Z. Sheng et al., “Tfb: Towards comprehensive and fair benchmarking of time series forecasting methods,” arXiv preprint arXiv:2403.20150, 2024

    work page arXiv 2024

  16. [16]

    BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert: Pre- training of deep bidirectional transformers for language under- standing,” arXiv preprint arXiv:1810.04805, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  17. [17]

    GPT-4 Technical Report

    J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat et al., “Gpt-4 technical report,” arXiv preprint arXiv:2303.08774, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  18. [18]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in CVPR, 2016

    work page 2016

  19. [19]

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

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An image is worth 16x16 words: Transformers for image recognition at scale,” in ICLR, 2021

    work page 2021

  20. [20]

    Deep learning based recommender system: A survey and new perspectives,

    S. Zhang, L. Yao, A. Sun, and Y. Tay, “Deep learning based recommender system: A survey and new perspectives,”Computing surveys, vol. 52, no. 1, pp. 1–38, 2019

    work page 2019

  21. [21]

    Informer: Beyond efficient transformer for long sequence time- series forecasting,

    H. Zhou, S. Zhang, J. Peng, S. Zhang, J. Li, H. Xiong, and W. Zhang, “Informer: Beyond efficient transformer for long sequence time- series forecasting,” in AAAI, 2021

    work page 2021

  22. [22]

    Autoformer: Decomposition transformers with auto-correlation for long-term series forecast- ing,

    H. Wu, J. Xu, J. Wang, and M. Long, “Autoformer: Decomposition transformers with auto-correlation for long-term series forecast- ing,” in NeurIPS, 2021

    work page 2021

  23. [23]

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

    Y. Nie, N. H. Nguyen, P . Sinthong, and J. Kalagnanam, “A time series is worth 64 words: Long-term forecasting with transformers,” in ICLR, 2023

    work page 2023

  24. [24]

    N- BEATS: neural basis expansion analysis for interpretable time series forecasting,

    B. N. Oreshkin, D. Carpov, N. Chapados, and Y. Bengio, “N- BEATS: neural basis expansion analysis for interpretable time series forecasting,” in ICLR, 2020

    work page 2020

  25. [25]

    Some properties of time series data and their use in econometric model specification,

    C. W. Granger, “Some properties of time series data and their use in econometric model specification,” Journal of econometrics, vol. 16, no. 1, pp. 121–130, 1981

    work page 1981

  26. [26]

    Co-integration and error correc- tion: representation, estimation, and testing,

    R. F. Engle and C. W. Granger, “Co-integration and error correc- tion: representation, estimation, and testing,” Econometrica: journal of the Econometric Society, pp. 251–276, 1987

    work page 1987

  27. [27]

    Vector autoregressions,

    J. H. Stock and M. W. Watson, “Vector autoregressions,” Journal of Economic perspectives, vol. 15, no. 4, pp. 101–115, 2001

    work page 2001

  28. [28]

    Spectral temporal graph neural network for multivariate time-series forecasting,

    D. Cao, Y. Wang, J. Duan, C. Zhang, X. Zhu, C. Huang, Y. Tong, B. Xu, J. Bai, J. Tonget al., “Spectral temporal graph neural network for multivariate time-series forecasting,” in NeurIPS, 2020

    work page 2020

  29. [29]

    Crossformer: Transformer utilizing cross- dimension dependency for multivariate time series forecasting,

    Y. Zhang and J. Yan, “Crossformer: Transformer utilizing cross- dimension dependency for multivariate time series forecasting,” in ICLR, 2023

    work page 2023

  30. [30]

    iTransformer: Inverted Transformers Are Effective for Time Series Forecasting

    Y. Liu, T. Hu, H. Zhang, H. Wu, S. Wang, L. Ma, and M. Long, “itransformer: Inverted transformers are effective for time series forecasting,” arXiv preprint arXiv:2310.06625, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  31. [31]

    A. S. Weigend, Time series prediction: forecasting the future and understanding the past. Routledge, 2018

    work page 2018

  32. [32]

    Time series forecasting using a hybrid arima and neural network model,

    G. P . Zhang, “Time series forecasting using a hybrid arima and neural network model,” Neurocomputing, vol. 50, pp. 159–175, 2003. IEEE TRANSACTIONS ON PATTERN ANAL YSIS AND MACHINE INTELLIGENCE, VOL. XX, NO. X 17

    work page 2003

  33. [33]

    Time series data imputation: A survey on deep learning approaches,

    C. Fang and C. Wang, “Time series data imputation: A survey on deep learning approaches,” arXiv preprint arXiv:2011.11347, 2020

    work page arXiv 2011

  34. [34]

    Multivariate time series imputation with generative adversarial networks,

    Y. Luo, X. Cai, Y. Zhang, J. Xu et al., “Multivariate time series imputation with generative adversarial networks,” in NeurIPS, 2018

    work page 2018

  35. [35]

    imputets: time series missing value imputation in r

    S. Moritz and T. Bartz-Beielstein, “imputets: time series missing value imputation in r.” R Journal, vol. 9, no. 1, p. 207, 2017

    work page 2017

  36. [36]

    Convolutional neural networks for time series classification,

    B. Zhao, H. Lu, S. Chen, J. Liu, and D. Wu, “Convolutional neural networks for time series classification,” Journal of Systems Engineering and Electronics, vol. 28, no. 1, pp. 162–169, 2017

    work page 2017

  37. [37]

    Generic and scalable framework for automated time-series anomaly detection,

    N. Laptev, S. Amizadeh, and I. Flint, “Generic and scalable framework for automated time-series anomaly detection,” in SIGKDD, 2015

    work page 2015

  38. [38]

    Instance Normalization: The Missing Ingredient for Fast Stylization

    D. Ulyanov, A. Vedaldi, and V . Lempitsky, “Instance normaliza- tion: The missing ingredient for fast stylization,” arXiv preprint arXiv:1607.08022, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  39. [39]

    Deep adaptive input normalization for time series forecasting,

    N. Passalis, A. Tefas, J. Kanniainen, M. Gabbouj, and A. Iosifidis, “Deep adaptive input normalization for time series forecasting,” IEEE transactions on neural networks and learning systems , vol. 31, no. 9, pp. 3760–3765, 2019

    work page 2019

  40. [40]

    Reversible instance normalization for accurate time-series forecasting against distribution shift,

    T. Kim, J. Kim, Y. Tae, C. Park, J.-H. Choi, and J. Choo, “Reversible instance normalization for accurate time-series forecasting against distribution shift,” in ICLR, 2021

    work page 2021

  41. [41]

    Non-stationary trans- formers: Exploring the stationarity in time series forecasting,

    Y. Liu, H. Wu, J. Wang, and M. Long, “Non-stationary trans- formers: Exploring the stationarity in time series forecasting,” in NeurIPS, 2022

    work page 2022

  42. [42]

    Client: Cross-variable linear integrated enhanced transformer for multivariate long-term time series forecasting,

    J. Gao, W. Hu, and Y. Chen, “Client: Cross-variable linear integrated enhanced transformer for multivariate long-term time series forecasting,” arXiv preprint arXiv:2305.18838, 2023

    work page arXiv 2023

  43. [43]

    Dance of channel and sequence: An efficient attention-based approach for multivariate time series forecasting,

    H. Wang, Y. Mo, N. Yin, H. Dai, B. Li, S. Fan, and S. Mo, “Dance of channel and sequence: An efficient attention-based approach for multivariate time series forecasting,” arXiv preprint arXiv:2312.06220, 2023

    work page arXiv 2023

  44. [44]

    Adaptive normalization for non-stationary time series forecasting: A temporal slice perspective,

    Z. Liu, M. Cheng, Z. Li, Z. Huang, Q. Liu, Y. Xie, and E. Chen, “Adaptive normalization for non-stationary time series forecasting: A temporal slice perspective,” in NeurIPS, 2024

    work page 2024

  45. [45]

    Stl: A seasonal-trend decomposition,

    R. B. Cleveland, W. S. Cleveland, J. E. McRae, and I. Terpenning, “Stl: A seasonal-trend decomposition,” J. Off. Stat, vol. 6, no. 1, pp. 3–73, 1990

    work page 1990

  46. [46]

    Time-series. 2nd edn

    O. D. Anderson, “Time-series. 2nd edn.” 1976

    work page 1976

  47. [47]

    Stl: A seasonal-trend decomposition procedure based on loess,

    C. RB, “Stl: A seasonal-trend decomposition procedure based on loess,” J Off Stat, vol. 6, pp. 3–73, 1990

    work page 1990

  48. [48]

    E. B. Dagum and S. Bianconcini, Seasonal adjustment methods and real time trend-cycle estimation. Springer, 2016

    work page 2016

  49. [49]

    Robuststl: A robust seasonal-trend decomposition algorithm for long time series,

    Q. Wen, J. Gao, X. Song, L. Sun, H. Xu, and S. Zhu, “Robuststl: A robust seasonal-trend decomposition algorithm for long time series,” in AAAI, 2019

    work page 2019

  50. [50]

    Forecasting time series with complex seasonal patterns using exponential smoothing,

    A. M. De Livera, R. J. Hyndman, and R. D. Snyder, “Forecasting time series with complex seasonal patterns using exponential smoothing,” Journal of the American statistical association , vol. 106, no. 496, pp. 1513–1527, 2011

    work page 2011

  51. [51]

    Forecasting at scale,

    S. J. Taylor and B. Letham, “Forecasting at scale,” The American Statistician, vol. 72, no. 1, pp. 37–45, 2018

    work page 2018

  52. [52]

    Are transformers effective for time series forecasting?

    A. Zeng, M. Chen, L. Zhang, and Q. Xu, “Are transformers effective for time series forecasting?” in AAAI, 2023

    work page 2023

  53. [53]

    Inpar- former: evolutionary decomposition transformers with interactive parallel attention for long-term time series forecasting,

    H. Cao, Z. Huang, T. Yao, J. Wang, H. He, and Y. Wang, “Inpar- former: evolutionary decomposition transformers with interactive parallel attention for long-term time series forecasting,” in AAAI, 2023

    work page 2023

  54. [54]

    Preformer: predictive transformer with multi-scale segment-wise correlations for long-term time series forecasting,

    D. Du, B. Su, and Z. Wei, “Preformer: predictive transformer with multi-scale segment-wise correlations for long-term time series forecasting,” in ICASSP, 2023

    work page 2023

  55. [55]

    O., Pfister, T., Zheng, Y., Ye, W., and Liu, Y

    D. Cao, F. Jia, S. O. Arik, T. Pfister, Y. Zheng, W. Ye, and Y. Liu, “Tempo: Prompt-based generative pre-trained transformer for time series forecasting,” arXiv preprint arXiv:2310.04948, 2023

    work page arXiv 2023

  56. [56]

    N- beats: Neural basis expansion analysis for interpretable time series forecasting,

    B. N. Oreshkin, D. Carpov, N. Chapados, and Y. Bengio, “N- beats: Neural basis expansion analysis for interpretable time series forecasting,” arXiv preprint arXiv:1905.10437, 2019

    work page arXiv 1905

  57. [57]

    Nhits: Neural hierarchical interpo- lation for time series forecasting,

    C. Challu, K. G. Olivares, B. N. Oreshkin, F. G. Ramirez, M. M. Canseco, and A. Dubrawski, “Nhits: Neural hierarchical interpo- lation for time series forecasting,” in AAAI, 2023

    work page 2023

  58. [58]

    Depts: Deep expansion learning for periodic time series forecasting,

    W. Fan, S. Zheng, X. Yi, W. Cao, Y. Fu, J. Bian, and T.-Y. Liu, “Depts: Deep expansion learning for periodic time series forecasting,” in ICLR, 2022

    work page 2022

  59. [59]

    Dewp: Deep expansion learning for wind power forecasting,

    W. Fan, Y. Fu, S. Zheng, J. Bian, Y. Zhou, and H. Xiong, “Dewp: Deep expansion learning for wind power forecasting,” ACM Transactions on Knowledge Discovery from Data , vol. 18, no. 3, pp. 1–21, 2024

    work page 2024

  60. [60]

    Temporal collaborative filtering with bayesian probabilistic tensor factorization,

    L. Xiong, X. Chen, T.-K. Huang, J. Schneider, and J. G. Carbonell, “Temporal collaborative filtering with bayesian probabilistic tensor factorization,” in SIAM-SDM, 2010

    work page 2010

  61. [61]

    Temporal regularized matrix factorization for high-dimensional time series prediction,

    H.-F. Yu, N. Rao, and I. S. Dhillon, “Temporal regularized matrix factorization for high-dimensional time series prediction,” in NeurIPS, 2016

    work page 2016

  62. [62]

    Autoregressive tensor factorization for spatio-temporal predictions,

    K. Takeuchi, H. Kashima, and N. Ueda, “Autoregressive tensor factorization for spatio-temporal predictions,” in ICDM, 2017

    work page 2017

  63. [63]

    Nonsta- tionary temporal matrix factorization for multivariate time series forecasting,

    X. Chen, C. Zhang, X.-L. Zhao, N. Saunier, and L. Sun, “Nonsta- tionary temporal matrix factorization for multivariate time series forecasting,” arXiv preprint arXiv:2203.10651, 2022

    work page arXiv 2022

  64. [64]

    Bayesian temporal factorization for multi- dimensional time series prediction,

    X. Chen and L. Sun, “Bayesian temporal factorization for multi- dimensional time series prediction,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 9, pp. 4659–4673, 2021

    work page 2021

  65. [65]

    Think globally, act locally: A deep neural network approach to high-dimensional time series forecasting,

    R. Sen, H.-F. Yu, and I. S. Dhillon, “Think globally, act locally: A deep neural network approach to high-dimensional time series forecasting,” in NeurIPS, 2019

    work page 2019

  66. [66]

    Real-time spatiotemporal prediction and imputation of traffic status based on lstm and graph laplacian regularized matrix factorization,

    J.-M. Yang, Z.-R. Peng, and L. Lin, “Real-time spatiotemporal prediction and imputation of traffic status based on lstm and graph laplacian regularized matrix factorization,” Transportation Research Part C: Emerging Technologies, vol. 129, p. 103228, 2021

    work page 2021

  67. [67]

    Graph regularized nonnegative matrix factorization for data representation,

    D. Cai, X. He, J. Han, and T. S. Huang, “Graph regularized nonnegative matrix factorization for data representation,” IEEE transactions on pattern analysis and machine intelligence, vol. 33, no. 8, pp. 1548–1560, 2010

    work page 2010

  68. [68]

    I. N. Sneddon, Fourier transforms. Courier Corporation, 1995

    work page 1995

  69. [69]

    J. W. Goodman, Introduction to Fourier optics . Roberts and Company publishers, 2005

    work page 2005

  70. [70]

    Bloomfield,Fourier analysis of time series: an introduction

    P . Bloomfield,Fourier analysis of time series: an introduction . John Wiley & Sons, 2004

    work page 2004

  71. [71]

    Epileptic seizure classifi- cation of eeg time-series using rational discrete short-time fourier transform,

    K. Samiee, P . Kovacs, and M. Gabbouj, “Epileptic seizure classifi- cation of eeg time-series using rational discrete short-time fourier transform,” IEEE transactions on Biomedical Engineering , vol. 62, no. 2, pp. 541–552, 2014

    work page 2014

  72. [72]

    E. O. Brigham, The fast Fourier transform and its applications . Prentice-Hall, Inc., 1988

    work page 1988

  73. [73]

    Meyer, Wavelets and operators: volume 1

    Y. Meyer, Wavelets and operators: volume 1. Cambridge university press, 1992, no. 37

    work page 1992

  74. [74]

    Multilevel wavelet decompo- sition network for interpretable time series analysis,

    J. Wang, Z. Wang, J. Li, and J. Wu, “Multilevel wavelet decompo- sition network for interpretable time series analysis,” in SIGKDD, 2018

    work page 2018

  75. [75]

    Stfnets: Learning sensing signals from the time-frequency perspective with short-time fourier neural networks,

    S. Yao, A. Piao, W. Jiang, Y. Zhao, H. Shao, S. Liu, D. Liu, J. Li, T. Wang, S. Hu et al. , “Stfnets: Learning sensing signals from the time-frequency perspective with short-time fourier neural networks,” in The World Wide Web Conference, 2019, pp. 2192–2202

    work page 2019

  76. [76]

    Wavelet transform application for/in non-stationary time-series analysis: A review,

    M. Rhif, A. Ben Abbes, I. R. Farah, B. Mart ´ınez, and Y. Sang, “Wavelet transform application for/in non-stationary time-series analysis: A review,” Applied Sciences, vol. 9, no. 7, p. 1345, 2019

    work page 2019

  77. [77]

    T-wavenet: A tree-structured wavelet neural network for time series signal analysis,

    L. Minhao, A. Zeng, L. Qiuxia, R. Gao, M. Li, J. Qin, and Q. Xu, “T-wavenet: A tree-structured wavelet neural network for time series signal analysis,” in ICLR, 2021

    work page 2021

  78. [78]

    Unsupervised time-series representation learning with iterative bilinear temporal-spectral fusion,

    L. Yang and S. Hong, “Unsupervised time-series representation learning with iterative bilinear temporal-spectral fusion,” in ICML, 2022

    work page 2022

  79. [79]

    When: A wavelet-dtw hybrid attention network for heterogeneous time series analysis,

    J. Wang, C. Yang, X. Jiang, and J. Wu, “When: A wavelet-dtw hybrid attention network for heterogeneous time series analysis,” in SIGKDD, 2023

    work page 2023

  80. [80]

    Waveform: Graph enhanced wavelet learning for long sequence forecasting of multivariate time series,

    F. Yang, X. Li, M. Wang, H. Zang, W. Pang, and M. Wang, “Waveform: Graph enhanced wavelet learning for long sequence forecasting of multivariate time series,” in AAAI, 2023

    work page 2023

Showing first 80 references.