Pith. sign in

REVIEW 4 major objections 7 minor 47 references

PaMMA-Net: Plasmas magnetic measurement evolution based on data-driven incremental accumulative prediction

T0 review · 4 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read PaMMA-Net forecasts tokamak magnetic signals across full discharges, beating Transformer and LSTM baselines on EAST data.

desk verdict Strong reported numbers, but the 'known future inputs' include measured plasma state, so the evolution claim is partly oracle; the incremental prediction trick is solid. read the letter →

arxiv 2501.14003 v1 pith:3L54ZCVX submitted 2025-01-23 physics.plasm-ph cs.AI

Yunfei Ling , Zijie Liu , Jun Du , Yao Huang , Yuehang Wang , Bingjia Xiao , Xin Fang This is my paper · ORCID
classification physics.plasm-phcs.AI
keywords deeplearningtokamakmagneticmeasurementsincrementalpredictiontimeseriesforecastingdataaugmentationEASTequilibriumreconstruction
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

The paper tries to establish that a deep network, PaMMA-Net, can evolve the magnetic measurements of a tokamak discharge—the magnetic surface (BP) and flux loop (FL) signals—over long horizons, up to an entire shot, using only a window of recent observations plus control signals that are known in advance. This matters because fast and reliable plasma evolution models are needed for tokamak control and for studying plasma behavior, while physics-based evolution is often too slow or fragile. The two key ideas are incremental accumulative prediction (predicting the time-difference of each signal and accumulating it back rather than predicting the absolute value directly) and a physically consistent data augmentation that mixes high-frequency components in the spectrogram domain. On EAST data the model reaches a mean absolute error of $2.661\times10^{-3}$, similarity of 95.86%, relative error of 99.34%, and correlation of 99.66%, outperforming Transformer, LSTM, Flowformer, and non-stationary transformer baselines with fewer parameters.

What carries the argument

The load-bearing object is the incremental accumulative prediction head: the network is trained to predict $\Delta y_{t+1:t+N}$ (with an auxiliary supervised loss on the increments) and a non-parametric accumulation layer converts these into $\hat y_{t+n} = y_t + \sum_{\tau=1}^n \Delta \hat y_{t+\tau}$. This converts a regression problem on slowly varying, non-normal absolute signals into one on near-normally distributed, small-range increments, which the authors show accelerates early training and lowers final validation error. The rest of the architecture supports that head: two embedding branches (state fusion and PCS command) that concatenate the raw input with an MLP tokenization and add positional encoding, cascaded decoder layers with causal attention so that inference can be autoregressive, and variable-separated output projections that give BP and FL their own MLP heads. The data augmentation completes the design by performing CutMix on the high-frequency part (above 80 Hz) of short-time Fourier transform spectrograms of the magnetic signals, enriching the training set without altering the overall trend.

What would settle it

Take a discharge in which a spontaneous event with no preceding command (e.g., an unmitigated disruption or a sawtooth crash not driven by the given PF voltages) occurs after the observed window. If the model's predicted magnetic measurements continue along the pre-event trend while the real BP/FL signals show the event, the claim that observed measurements plus known commands suffice for evolution is refuted.

Watch

Extended reading notes

Core claim

The central claim is that predicting the increment $\Delta y_t = y_t - y_{t-1}$ of each magnetic measurement channel, then reconstructing the absolute signal by $\hat y_{t+n} = y_t + \sum_{\tau=1}^n \Delta \hat y_{t+\tau}$, makes long-sequence magnetic measurement evolution both more accurate and easier to train. The paper argues that the increments are near-normally distributed and have far smaller dynamic range than the measurements themselves, so the network converges faster and captures finer-grained plasma behavior. PaMMA-Net combines this incremental head with a Transformer decoder that uses causal masking, two embedding branches (state fusion and PCS commands), and parameter-separated output projections for BP and FL modalities, plus an STFT-domain CutMix augmentation that preserves the low-frequency trend while diversifying high-frequency content. The resulting model reports the best evolution metrics on EAST data among the tested architectures, generalizes across shots with different configurations, and, when coupled with equilibrium reconstruction, produces plasma boundaries that track the measured ones.

Load-bearing premise

The model assumes that the future control signals (PCS commands, including PF coil voltages) are known in advance for the whole prediction horizon and that these, together with the observed magnetic measurements, fully determine the future magnetic measurements; if unmeasured plasma dynamics such as internal reconnection or turbulence significantly affect the signals, the predictions degrade.

Editorial extensions

If this is right

  • If the claim holds, a single fast network can replace or supplement physics-based evolution for offline controller design and discharge scenario studies, producing whole-shot magnetic measurement evolution in milliseconds rather than minutes.
  • The incremental accumulative prediction trick appears to be architecture-agnostic: the paper shows it improves not only PaMMA-Net but also Transformer, LSTM, Flowformer, and non-stationary transformer baselines on this task, suggesting it transfers to other slow-varying signal prediction problems.
  • The coupling with equilibrium reconstruction means the predicted magnetic measurements can be converted into macroscopic quantities such as the last closed-flux surface, yielding predicted plasma shape evolution from the same model.
  • Because the model uses only 18 known future input channels (mostly PF coil voltages and a few macroscopic signals), it may generalize to unseen discharges more readily than models requiring more complete knowledge of the machine state.

Reading between the lines

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

  • The known-future-input assumption positions PaMMA-Net as a simulation surrogate rather than a standalone forecaster: its accuracy is contingent on the control commands being prescribed, which fits offline scenario exploration and controller verification, but limits use for predicting the outcome of events not commanded (e.g., spontaneous disruptions).
  • The claim that absolute magnetic measurement distributions are 'non-normal' while increments are 'normal' is heuristic; a testable extension would be to quantify the distributional shift (e.g., via a normality test or entropy) per channel and correlate it with the observed gain from the incremental design.
  • The augmentation mixes only high-frequency content above 80 Hz; since the paper shows most signal energy is in the fundamental frequency, an extension would be to mix low-frequency components across shots with similar operating regimes to test whether the model can generalize to unobserved equilibrium shapes.
  • A testable extension is to apply the same incremental accumulative architecture to other tokamak devices or to synthetic ITER-like scenarios to see whether the reported EAST gains persist across different machine geometries and control systems.
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

4 major / 7 minor

Summary. PaMMA-Net proposes a Transformer-based model for predicting tokamak magnetic measurements (magnetic surface signals BP and flux signals FL) on EAST discharges. The model uses an incremental accumulative prediction head, a state-fusion embedding that concatenates observed signals with MLP embeddings, and a spectrogram CutMix data augmentation that mixes high-frequency components across shots. The paper reports quantitative results on a large EAST dataset (7671 training shots, 830 test shots) and compares against Transformer, LSTM, Flowformer, and Non-stationary transformer baselines, with and without the incremental design. Ablations examine data augmentation, incremental design, positional embedding, and normalization. A qualitative autoregressive case study and an equilibrium-reconstruction-derived plasma boundary example are also presented.

Significance. If the task were causal predictive evolution from pre-programmed control commands, the model could be a fast surrogate for magnetic measurements and could enable offline controller interaction, which would be a useful contribution to plasma control. The incremental accumulative prediction idea is simple and shows consistent improvements across all baselines, and the large-scale EAST data test set is a strength. However, the current task formulation includes future measured plasma state variables as known inputs, which makes the reported quantitative results an evaluation of a decoding/synthesis task rather than a genuine evolution prediction. The paper also lacks comparisons with plasma-specific prediction models, despite claiming superiority over existing studies. With reframing and additional experiments, the approach could still be valuable, but as presented the central claims are overstated.

major comments (4)
  1. [Sec. 3.1.1, Table 1, Eq. (1)] The 'known future inputs' Z include measured or reconstructed plasma state variables (Ip, Vloop, WMHD, βp, li) in addition to coil voltage commands. Because these macroscopic variables are themselves consequences of the plasma evolution, supplying them as known future inputs converts the task from causal prediction into a decoding/synthesis problem: the model maps already-known future plasma behavior into magnetic signals. For the claimed use cases in Sec. 4.6 and Sec. 5 (offline controller interaction, prediction for unseen discharges), such future information is not available in advance. The quantitative gains in Tables 3 and 4 may largely reflect this oracle information, and the comparison with baselines, while internally fair, does not establish predictive capability in a realistic setting. Please either reformulate the known inputs to contain only pre-programmed control commands (PF coil voltages and IC), or explicitly reposition the method as a magnetic-signal synthesizer conditioned on scenario parameters and adjust the claims accordingly.
  2. [Sec. 4.2 and Sec. 4.6] The quantitative metrics in Table 3 are computed on non-autoregressive sliding-window tests, where each 1000-step prediction window is preceded by the true observed magnetic measurements (as described in Sec. 4.2). This evaluation measures the model's ability to predict one window given fresh sensor data, not its ability to evolve over extended periods, because the input window is refreshed with ground truth at every step. The only test of true long-horizon behavior is the autoregressive case study of Sec. 4.6, which reports no quantitative error metrics beyond a qualitative statement that errors 'do not diverge.' The central claim of 'long-sequence magnetic measurement evolution' (abstract, Sec. 1) is therefore not supported by the reported numbers. Please report quantitative autoregressive metrics, such as MAE or similarity as a function of prediction horizon, or revise the claim to specify the 1000-step conditioned setting.
  3. [Abstract and Table 3] The abstract claims 'superior evolution results compared to existing studies,' but the empirical comparison in Table 3 is limited to four generic time-series architectures (Transformer, LSTM, Flowformer, Non-stationary transformer). No plasma-specific discharge-prediction or magnetic-evolution models are included as baselines, so the claim of superiority over existing plasma studies is not demonstrated. Either add relevant plasma-specific baselines (e.g., discharge prediction models or equilibrium-reconstruction-based surrogates) or amend the claim to state that the method outperforms the generic time-series baselines considered here.
  4. [Sec. 3.3] The data augmentation is labeled 'physically consistent,' but the only justification offered is that the low-frequency trend (below 80 Hz) is preserved while high-frequency patches are exchanged between different discharges. No physical argument or quantitative validation is provided to show that the mixed high-frequency content is representative of plausible plasma dynamics; the mixing operation may introduce non-physical high-frequency artifacts that the model could exploit as a regularizer. Please either provide supporting evidence (e.g., verify that augmented signals remain within physically expected ranges or do not degrade physical constraints) or rename the method to avoid overclaiming.
minor comments (7)
  1. [Sec. 1, contributions] The list of contributions numbers only items 1) and 3), skipping 2); a numbering correction is needed.
  2. [Abstract] The phrase 'a incremental prediction' should be 'an incremental prediction'.
  3. [Sec. 4.1] The sentence 'By screening the discharges with longer durations and higher sampling frequencies no from EAST experiments' contains an extraneous 'no'.
  4. [Sec. 3.3] The task is supervised, not 'self-supervised,' because the model trains on labeled future targets; please correct this terminology.
  5. [Sec. 3.1.1] The phrase 'In subsequent chapters' should be 'In subsequent sections'.
  6. [Sec. 4.3] The sentence 'Specifically, it attains a Sim of 94.87% and a Rela of 99.09%' does not match any row in Table 3 (the closest values are in the PaMMA-Net 'Original' row, but those are Sim 93.68% and Rela 98.68%); please verify and correct these numbers.
  7. [Tables 3 and 4] No error bars or multiple-run statistics are reported, which makes it difficult to judge whether small differences in Corr (e.g., 99.53% vs. 99.66% in Table 3) are statistically meaningful; adding variance over seeds or runs would strengthen the empirical claims.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the magnetic-measurement evolution is an empirically benchmarked supervised mapping; the known-future-input oracle is a disclosed limitation, not a self-referential derivation.

full rationale

The central result is an empirical supervised-learning benchmark. Eq. (1) defines the prediction as \hat Y = f(X, Z), and all baselines (Transformer, LSTM, Flowformer, Non-stationary transformer) are evaluated on the same input/output pairs, so the reported relative gains are not manufactured by the model definition. The incremental accumulative prediction (Eq. 2) reparameterizes the target as a sum of predicted increments plus the observed initial value y_t; it does not feed the target back as an input, so it is a training/optimization device rather than a circular construction. The CutMix-style augmentation follows external algorithms (CutMix, Mixup) and is applied jointly to input and target as a regularizer; it does not fit the test predictions. The known-future-input set Z includes reconstructed macroscopic plasma variables (Ip, WMHD, βp, li) as disclosed in Table 1 and Sec. 4.6, which makes the task an oracle-conditional emulator and raises a legitimate external-validity question for free-running discharge prediction; however, the paper does not claim to predict those variables, and the output Y is a separate 72-channel diagnostic signal evaluated against held-out measured data. Thus the prediction is not equivalent to its inputs by construction. The handful of references to prior EAST-group work (e.g., Refs. [17,18]) are contextual related-work citations and are not load-bearing for the model architecture or the quantitative claims. The main weaknesses are task-design overclaiming and lack of a fully autoregressive quantitative benchmark, which are correctness/scope concerns rather than circularity.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on several modeling choices: the definition of known future inputs, the learnability assumption of the plasma response, and the physical-consistency assumption of the spectrogram augmentation. None of these are machine-checked or externally validated beyond the reported test-set metrics.

free parameters (5)
  • Observation window length M = 1000 steps
    Chosen by hand; determines the context length fed to the transformer.
  • Prediction horizon N = 1000 steps
    Chosen by hand; the model predicts 1000 steps per inference.
  • Frequency cutoff H0 for augmentation = 80 Hz
    Chosen based on spectrogram analysis to separate the fundamental frequency from power supply harmonics.
  • Edge margin W0 for augmentation = 0.1W
    Chosen to avoid spectrum edge effects in the CutMix bounding box.
  • Mixup alpha = not specified
    The ratio lambda is sampled from Beta(alpha, alpha), but the value of alpha is not reported.
assumptions (4)
  • domain assumption Future control signals are known in advance and, together with observed measurements, determine future magnetic measurements.
    Task definition (Sec. 3.1.1) assumes Z is known future input; the model learns f(X, Z) -> Y.
  • domain assumption The mapping from observed measurements and control inputs to future magnetic measurements is learnable from historical discharge data and generalizes to unseen shots.
    Supervised training on EAST shots assumes stationarity of the plasma response across discharges.
  • ad hoc to paper Data augmentation by mixing high-frequency spectrogram patches preserves the physical plausibility of plasma magnetic signals.
    Sec. 3.3 states the augmentation is 'physically consistent' based on spectrogram analysis, but no explicit validation against plasma physics is provided.
  • standard math Resampling all signals to 1000 Hz with linear interpolation preserves the information content needed for prediction.
    Preprocessing in Sec. 4.1; linear interpolation is a standard assumption for missing data filling, but can introduce artifacts.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PaMMA-Net: Plasmas magnetic measurement evolution based on data-driven incremental accumulative prediction." pith.science (2026). https://pith.science/paper/3L54ZCVX

@misc{pith2026250114003,
  author       = {Pith},
  title        = {Pith review of: PaMMA-Net: Plasmas magnetic measurement evolution based on data-driven incremental accumulative prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3L54ZCVX}},
  note         = {Machine review of arXiv:2501.14003}
}
read the original abstract

An accurate evolution model is crucial for effective control and in-depth study of fusion plasmas. Evolution methods based on physical models often encounter challenges such as insufficient robustness or excessive computational costs. Given the proven strong fitting capabilities of deep learning methods across various fields, including plasma research, this paper introduces a deep learning-based magnetic measurement evolution method named PaMMA-Net (Plasma Magnetic Measurements Incremental Accumulative Prediction Network). This network is capable of evolving magnetic measurements in tokamak discharge experiments over extended periods or, in conjunction with equilibrium reconstruction algorithms, evolving macroscopic parameters such as plasma shape. Leveraging a incremental prediction approach and data augmentation techniques tailored for magnetic measurements, PaMMA-Net achieves superior evolution results compared to existing studies. The tests conducted on real experimental data from EAST validate the high generalization capability of the proposed method.

Figures

Figures reproduced from arXiv: 2501.14003 by the authors.

Figure 1
Figure 1. Workflow and model architecture of PaMMA-Net. The left half presents [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. (a) Distribution of BP channel 1 within the dataset, (b) Distribution of the increment of BP channel 1 within the dataset. 3.2.2. State fusion prediction As defined in Sec. 3.1, the proposed evolution model incorporates observed magnetic measurements as input, thereby implicitly modeling the relationships among various signal channels and furnishing an initialization for incremental prediction. However, this state f… view at source ↗
Figure 3
Figure 3. The principle and efficacy of the physically consistent data augmentation. [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Learning behaviors of PaMMA-Net with and without incremental design. [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: The evolution similarity metrics and its distribution of all discharges within [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Examples of the non-autoregressive evolution outcomes throughout the entire [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]
Figure 7
Figure 7. Figure 7: The evolution of LCFS derived from the combination of magnetic measurement [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]
Figure 8
Figure 8. Figure 8: An example of the auto-regressive evolution outcomes throughout the entire [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 38 canonical work pages

  1. [1]

    Ongena J, Koch R, Wolf R and Zohm H 2016 Nature Physics 12 398–410

    work page 2016

  2. [2]

    2007 Plasma Science and Technology9 312

    Batchelor D, Beck M, Becoulet A, Budny R, Chang C, Diamond P, Dong J, Fu G, Fukuyama A, Hahm T et al. 2007 Plasma Science and Technology9 312

    work page 2007

  3. [3]

    Vaswani A 2017 Advances in Neural Information Processing Systems

    work page 2017

  4. [4]

    2023 arXiv preprint arXiv:2303.08774

    Achiam J, Adler S, Agarwal S, Ahmad L, Akkaya I, Aleman F L, Almeida D, Altenschmidt J, Altman S, Anadkat S et al. 2023 arXiv preprint arXiv:2303.08774

    arXiv 2023

  5. [5]

    Sharadga H, Hajimirza S and Balog R S 2020 Renewable Energy 150 797–807

    work page 2020

  6. [6]

    Tang Y, Song Z, Zhu Y, Yuan H, Hou M, Ji J, Tang C and Li J 2022 Neurocomputing 512 363–380

    work page 2022

  7. [7]

    2023 Nuclear Fusion63 094001

    Guo B, Chen D, Rea C, Wu M, Shen B, Granetz R, Zhang Z, Huang Y, Duan Y, Zeng L et al. 2023 Nuclear Fusion63 094001

    work page 2023

  8. [8]

    2022 Plasma Physics and Controlled Fusion64 074001

    Lao L, Kruger S, Akcay C, Balaprakash P, Bechtel T, Howell E, Koo J, Leddy J, Leinhauser M, Liu Y et al. 2022 Plasma Physics and Controlled Fusion64 074001

    work page 2022

Show all 47 references
  1. [9]

    Wan C, Yu Z, Pau A, Liu X and Li J 2022 Nuclear Fusion62 126060

    2022

  2. [10]

    2014 Nuclear Fusion54 043018

    Falchetto G L, Coster D, Coelho R, Scott B D, Figini L, Kalupin D, Nardon E, Nowak S, Alves L L, Artaud J F et al. 2014 Nuclear Fusion54 043018

    2014

  3. [11]

    Zwingmann W 2003 Nuclear Fusion43 842

    2003

  4. [12]

    2008 Plasma (2007)(greifswald, germany) aip conf Proc vol 993

    Zwingmann W et al. 2008 Plasma (2007)(greifswald, germany) aip conf Proc vol 993

    2007

  5. [13]

    Westerhof E, Peeters A G and Schippers W 1992 Relax-a computer code for the study of collisional and wave driven relaxation of the electron distribution function in toroidal geometry Tech. rep. FOM-Instituut voor Plasmafysica

    1992

  6. [14]

    L¨ utjens H, Bondeson A and Sauter O 1996 Computer physics communications97 219–260

    1996

  7. [15]

    Huysmans G, Sharapov S, Mikhailovskii A and Kerner W 2001 Physics of Plasmas8 4292–4305

    2001

  8. [16]

    Liu Y, Bondeson A, Fransson C M, Lennartson B and Breitholtz C 2000 Physics of Plasmas 7 3681–3690

    2000

  9. [17]

    2021 Nuclear Fusion61 066034

    Hu W, Rea C, Yuan Q, Erickson K, Chen D, Shen B, Huang Y, Xiao J, Chen J, Duan Y et al. 2021 Nuclear Fusion61 066034

    2021

  10. [18]

    Guo B H, Chen D L, Shen B, Rea C, Granetz R S, Zeng L, Hu W H, Qian J P, Sun Y W and Xiao B J 2021 Plasma Physics and Controlled Fusion63 115007 PaMMA-Net: Plasmas magnetic measurement evolution model 20

    2021

  11. [19]

    Yoshino R 2003 Nuclear fusion 43 1771

    2003

  12. [20]

    Clayton D, Tritz K, Stutman D, Bell R, Diallo A, LeBlanc B and Podest` a M 2013Plasma Physics and Controlled Fusion55 095015

  13. [21]

    Wan C, Yu Z, Wang F, Liu X and Li J 2021 Nuclear Fusion61 066015

    2021

  14. [22]

    Wan C, Yu Z, Pau A, Sauter O, Liu X, Yuan Q and Li J 2023 Nuclear Fusion63 056019

    2023

  15. [23]

    Wan C, Bai S, Yu Z, Yuan Q, Huang Y, Liu X, Hu Y and Li J 2024 Nuclear Fusion64 026014

    2024

  16. [24]

    Joung S, Kim J, Kwak S, Bak J, Lee S, Han H, Kim H, Lee G, Kwon D and Ghim Y C 2019 Nuclear Fusion60 016034

    2019

  17. [25]

    Yang B, Liu Z, Song X and Li X 2020 Plasma Physics and Controlled Fusion62 125022

    2020

  18. [26]

    Wakatsuki T, Suzuki T, Hayashi N, Oyama N and Ide S 2019 Nuclear Fusion59 066022

    2019

  19. [27]

    Seo J, Na Y S, Kim B, Lee C, Park M, Park S and Lee Y 2021 Nuclear Fusion61 106010

    2021

  20. [28]

    2022 Nature 602 414–419

    Degrave J, Felici F, Buchli J, Neunert M, Tracey B, Carpanese F, Ewalds T, Hafner R, Abdolmaleki A, de Las Casas D et al. 2022 Nature 602 414–419

    2022

  21. [29]

    Seo J, Kim S, Jalalvand A, Conlin R, Rothstein A, Abbate J, Erickson K, Wai J, Shousha R and Kolemen E 2024 Nature 626 746–751

    2024

  22. [30]

    Hochreiter S 1997 Neural Computation MIT-Press

    1997

  23. [31]

    Bai S, Kolter J Z and Koltun V 2018 arXiv preprint arXiv:1803.01271

    2018 arXiv

  24. [32]

    Salinas D, Flunkert V, Gasthaus J and Januschowski T 2020 International Journal of Forecasting 36 1181–1191

    2020

  25. [33]

    Wu H, Hu T, Liu Y, Zhou H, Wang J and Long M 2022 arXiv preprint arXiv:2210.02186

    2022 arXiv

  26. [34]

    Challu C, Olivares K G, Oreshkin B N, Ramirez F G, Canseco M M and Dubrawski A 2023 Nhits: Neural hierarchical interpolation for time series forecasting AAAI vol 37 pp 6989–6997

    2023

  27. [35]

    Oreshkin B N, Carpov D, Chapados N and Bengio Y 2019 arXiv preprint arXiv:1905.10437

    2019 arXiv

  28. [36]

    Zeng A, Chen M, Zhang L and Xu Q 2023 Are transformers effective for time series forecasting? AAAI vol 37 pp 11121–11128

    2023

  29. [37]

    Zhou H, Zhang S, Peng J, Zhang S, Li J, Xiong H and Zhang W 2021 Informer: Beyond efficient transformer for long sequence time-series forecasting AAAI vol 35 pp 11106–11115

    2021

  30. [38]

    Zhou T, Ma Z, Wen Q, Wang X, Sun L and Jin R 2022 Fedformer: Frequency enhanced decomposed transformer for long-term series forecasting ICML pp 27268–27286

    2022

  31. [39]

    Wu H, Xu J, Wang J and Long M 2021 Advances in Neural Information Processing Systems34 22419–22430

    2021

  32. [40]

    Nie Y, Nguyen N H, Sinthong P and Kalagnanam J 2022 arXiv preprint arXiv:2211.14730

    2022 arXiv

  33. [41]

    Liu Y, Hu T, Zhang H, Wu H, Wang S, Ma L and Long M 2023 arXiv preprint arXiv:2310.06625

    2023 arXiv

  34. [42]

    Lim B, Arık S ¨O, Loeff N and Pfister T 2021 International Journal of Forecasting37 1748–1764

    2021

  35. [43]

    Lao L, Ferron J, Groebner R, Howl W, John H S, Strait E and Taylor T 1990 Nuclear Fusion30 1035

    1990

  36. [44]

    Yun S, Han D, Oh S J, Chun S, Choe J and Yoo Y 2019 Cutmix: Regularization strategy to train strong classifiers with localizable features Proceedings of the IEEE/CVF international conference on computer visionpp 6023–6032

    2019

  37. [45]

    Zhang H 2017 arXiv preprint arXiv:1710.09412

    2017 arXiv

  38. [46]

    Wu H, Wu J, Xu J, Wang J and Long M 2022 arXiv preprint arXiv:2202.06258

    2022 arXiv

  39. [47]

    Liu Y, Wu H, Wang J and Long M 2022 Advances in Neural Information Processing Systems35 9881–9893

    2022

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.