Pith. sign in

REVIEW 3 major objections 6 minor 93 references

SSSUMO: Real-Time Semi-Supervised Submovement Decomposition

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A semi-supervised neural method, SSSUMO, decomposes velocity traces into discrete minimum-jerk submovements in real time and reports higher reconstruction accuracy than peak-detector and Scattershot baselines on both synthetic and human…

desk verdict Solid synthetic benchmark and real speed win, but the human-data claim rests on a self-training loop that the reported metrics cannot validate. read the letter →

arxiv 2507.08028 v1 pith:FRE3ZMDE submitted 2025-07-08 cs.HC cs.AIcs.CV

Evgenii Rudakov , Jonathan Shock , Otto Lappi , Benjamin Ultan Cowley This is my paper
classification cs.HCcs.AIcs.CV
keywords semi-supervisedlearningsubmovementdecompositionmotorcontroltemporalconvolutionalnetworksyntheticdatagenerationpseudo-labelingreal-timemotionanalysisminimum-jerkmodel
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

SSSUMO is a semi-supervised neural method that decomposes a one-dimensional velocity time series into discrete submovements, each described by an onset time, a duration, and a displacement, by learning to superimpose minimum-jerk velocity primitives. The paper claims that on synthetic benchmarks this learned decomposer beats a tuned peak-detector heuristic and the standard Scattershot optimizer on every overlap and noise condition for almost every metric, and on seven human-motion datasets it achieves the highest reconstruction $R^2$ in every task and noise condition. The method's central move is to train on synthetic traces generated from minimum-jerk priors and then iteratively resample those traces from the statistical distributions of the model's own pseudo-labels on unlabeled human data, so no hand-labeled submovements are ever needed. If the claim holds, the practical barriers to submovement analysis—lack of ground-truth labels and the hours-long runtime of optimization-based decomposition—are removed, enabling real-time decomposition that could support adaptive interfaces, rehabilitation monitoring, and motor-control research.

What carries the argument

The argument is carried by three coupled modules plus an input transform. The Detector is a temporal fully-convolutional network with a receptive field of about 1.65 seconds that maps a velocity sequence to three parallel outputs: onset probability, duration, and displacement at every time step. The Reconstructor is a partially differentiable module that turns predicted onsets into minimum-jerk velocity primitives, sums them into a reconstructed signal, and lets reconstruction error backpropagate into duration and displacement predictions. The bootstrapped refinement loop estimates kernel-density distributions of displacement, duration given displacement, onset-to-onset interval given displacement, and next-displacement dependency from the model's pseudo-labels on unlabeled human data, then regenerates synthetic training traces from those distributions and mixes them 50/50 with the original cold-start data. The signed tangential velocity transform flips the sign of speed at detected direction reversals, so opposing submovements contribute negative pulses instead of being merged into one positive peak.

What would settle it

Generate synthetic movement traces whose submovements are drawn from a realistic kernel-density distribution with known onset, duration, and displacement labels, then run the full pseudo-label refinement loop on them as if they were unlabeled human data; if the model's final predicted distributions drift systematically away from the planted values, or if additional refinement iterations increase parameter error, the bootstrap is entrenching bias. A complementary check: the paper reports that a 10 Hz low-pass filter already explains 89–99.8% of variance in the human datasets, so any real-data validation of decomposition correctness needs task-specific labels or independent neural correlates, not reconstruction $R^2$ alone.

Watch

Extended reading notes

Core claim

The paper's central claim is that submovement decomposition is better posed as a learned, semi-supervised inference problem than as an offline optimization problem. The authors argue that a temporal fully-convolutional detector, pretrained on labeled synthetic minimum-jerk traces and then refined through a pseudo-label bootstrap on unlabeled human movement, recovers onset, duration, and displacement parameters more accurately than both the peak-detector heuristic and the Scattershot optimizer on synthetic ground truth, and reconstructs human velocity signals with higher $R^2$ across all seven tested tasks and all three noise conditions. They also claim this accuracy comes with linear-time inference at about $6.7\times10^{-4}$ seconds per input second on a CPU, roughly 5,000 times faster than Scattershot, which they present as the difference between offline analysis and real-time use.

Load-bearing premise

The load-bearing premise is that the model's own pseudo-labels on unlabeled human movement are trustworthy enough to regenerate its training data: the refinement loop builds onset-interval, duration, and displacement distributions exclusively from the model's predictions, so a systematic decomposition bias in the base model can be entrenched rather than corrected, and the reported human-data metrics (reconstruction $R^2$ and submovement rate) would not reveal it.

Editorial extensions

If this is right

  • Submovement decomposition can be performed online on a laptop CPU, opening closed-loop human-computer interaction, adaptive interfaces, and rehabilitation feedback that optimization-based methods cannot support.
  • A model trained without any hand-labeled human submovements can outperform both heuristic peak detection and optimization-based decomposition in reconstruction fidelity on real movement data, including fast noisy tasks like handwriting.
  • Out-of-domain fine-tuning, where the target task is held out during adaptation, still improves performance on the hardest datasets, indicating the method transfers across task types rather than memorizing one.
  • On synthetic ground truth, parameter recovery ($F_1$, displacement $R^2$, duration $R^2$) degrades gracefully with submovement overlap and sensor noise, whereas both baselines' duration and displacement estimates collapse in high-overlap conditions.

Reading between the lines

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

  • Beyond the paper: the pseudo-label bootstrap is expectation-maximization-like, and that reading implies the 50% cold-start mixing and the false-positive penalties in Section 4.5 are what keep the loop from collapsing onto over-segmented solutions; ablating the mixture ratio and penalty strengths is a direct way to test how much of the reported gain comes from regularization rather than from the da
  • Beyond the paper: a planted-submovement validation—embedding known min-jerk pulses into human-like traces and running the full loop—would separate genuine adaptation from bias entrenchment, a test the paper does not perform.
  • Beyond the paper: because a 10 Hz low-pass filter already reaches 89–99.8% $R^2$ on most human datasets, the real-data comparisons certify signal reconstruction, not decomposition correctness; confirming the decomposition itself would require independent labels, such as time-locked neural correlates of submovements.
  • Beyond the paper: if the learned representations are as informative as the reconstruction accuracy suggests, stripping the output heads could turn the network into a feature extractor for motor-control studies, an application the authors propose but do not test.
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 / 6 minor

Summary. The manuscript proposes SSSUMO, a fully convolutional semi-supervised method for decomposing one-dimensional signed tangential velocity signals into overlapping minimum-jerk submovements. A temporal fully-convolutional detector is pre-trained on synthetic traces generated from minimum-jerk primitives, then refined through an iterative loop that pseudo-labels unlabeled human motion data, estimates KDE distributions of submovement parameters, regenerates synthetic training data from those distributions, and retrains the model. The detector outputs onset probability, duration, and displacement, and a partially differentiable reconstructor enables end-to-end training with a reconstruction loss. The method is evaluated on a synthetic benchmark with controlled overlap and noise and a domain-shifted test distribution, and on seven human motion datasets, where it reports higher reconstruction R2 than a Peak Detector and a Scattershot baseline, while running in well under a millisecond per input second on CPU. Training code, benchmark code, and pretrained weights are released.

Significance. If the results hold, the paper makes a substantial methodological contribution: it brings modern deep learning to an ill-posed inverse problem that has been dominated by exponential-time or polynomial-time optimization, with an honest synthetic benchmark design, disclosed baseline tuning, ablations that isolate design choices, and a public code release. The real-time throughput (6.7e-4 s per input second on CPU, Table 5) and robust parameter recovery on synthetic data (Table 2) are practically significant for adaptive human-computer interaction, rehabilitation, and motor control research. However, the strongest human-data claim—state-of-the-art decomposition accuracy—is not yet supported by the reported evidence, because reconstruction R2 and submovement rate cannot verify the correctness of the decomposition parameters. The synthetic comparison is independent and strong, but the test distribution still shares the generative family with the training data. I regard the contribution as valuable and the synthetic results as credible, but the human-data accuracy claims need additional validation before they should be accepted at face value.

major comments (3)
  1. [Section 4.4, Section 4.5, Section 6.2 (Table 3), Table 1, Figure 9]
  2. [Section 4.3, Section 5.3]
  3. [Section 4.4]
minor comments (6)
  1. [Abstract]
  2. [Section 3.2, Equation (4)]
  3. [Section 5.1]
  4. [Figure 9]
  5. [References]
  6. [Section 4.5]

Circularity Check

2 steps flagged · score 4.0 of 10

Fine-tuning loop estimates training distributions from the model's own pseudo-labels and validates adaptation via the same reconstruction objective, making the human-data accuracy claim partly self-referential; the synthetic ground-truth benchmark remains independent.

  1. fitted input called prediction [Section 4.4 'Bootstrapped Distribution Refinement'; Table 3 and Section 6.2 'Human-Motion Evaluation']
    "The process begins by applying the pre-trained model to unlabeled human motion data, generating pseudo-labels that represent the model’s current best estimate of submovement decomposition. From these predictions, we extract and estimate the statistical distributions of key submovement characteristics including displacement, duration, and temporal relationships between consecutive submovements. ... During training, we regenerate synthetic data after each epoch using the updated distribution estimates. ... This convergence is evidenced by improved reconstruction performance."

    The KDE distributions P(dn), P(Tn|dn), P(In|dn) and P(dn|dn-1) are estimated exclusively from the Detector's own pseudo-labels on unlabeled human data, and the next synthetic training set is sampled from those distributions. Human-data success is then reported as reconstruction R2, which is the same quantity the reconstruction loss L_MSE(v, \hat v) optimizes during fine-tuning. The paper itself concedes that the loop can 'reinforce decomposition into an increasingly larger number of submovements' and that 'this would lead to higher reconstruction accuracy, it would ultimately result in implausible decompositions' (Section 4.5). Table 1 also shows that a 10 Hz low-pass filter attains 89.3-99.8% R2, so reconstruction R2 barely constrains which decomposition is chosen.

  2. self definitional [Section 4.5 'Training Procedure and Loss Functions', fine-tuning loss definitions]
    "During the fine-tuning, there is a risk that the training could enter a loop of reinforcing decomposition into an increasingly larger number of submovements. ... The ratio gamma penalizes the model for producing more false positives by comparing the number of detected submovement onsets |O_hat| to the number of ground-truth submovement onsets |O|. Duration and displacement losses are computed over the set of ground-truth submovement onset timestamps O."

    In the fine-tuning phase on human data there is no human ground-truth onset set. The 'ground-truth' onsets O entering gamma, beta_t, and the duration/displacement MSE losses are synthetic labels sampled from the Label Sampler, which draws from KDEs estimated from the model's own pseudo-labels. Thus the anti-over-segmentation penalty only enforces agreement with the model-derived distribution: it cannot detect or correct a systematic bias in the base model's decomposition, even though the paper identifies over-segmentation as the exact failure mode it is meant to prevent.

full rationale

The synthetic benchmark is genuinely independent and is the main reason the score is not higher. In Section 6.1/Table 2, the model is tested on synthetic traces with ground-truth onset, duration, and displacement labels, including a shifted displacement distribution different from training, and it outperforms Peak Detector and Scattershot on the parameter-recovery metrics; this is external, held-out evidence not reducible to the fine-tuning loop. The ablation study in Section 5.3/Figure 8 similarly isolates real design choices. I found no load-bearing self-citation: the authors' own steering and flow datasets are used only as benchmark data, and the min-jerk and optimization machinery is cited from Rohrer and Hogan, not from the present authors. The circularity is localized to the human-data adaptation: pseudo-labels from the model determine the KDE distributions that regenerate the next training batch, and the same reconstruction objective is then cited as evidence of convergence. The Section 4.5 safeguards are also defined relative to synthetic 'ground-truth' onsets that originate from the same pseudo-label-derived distributions. Accordingly, the human-data R2 gains in Tables 3-4 should be read as evidence of self-consistency within the loop rather than proof that the recovered submovement parameters are correct; the paper is candid about the over-segmentation risk, but still presents the human-data R2 as 'state-of-the-art accuracy' without an external decomposition check.

Assumptions & free parameters 9 free parameters · 6 assumptions · 1 invented entities

The central claim rests primarily on the minimum-jerk and additivity axioms for submovement shape, on the hand-chosen synthetic generator priors, and, for the human-data adaptation, on the unvalidated premise that pseudo-labels from the model itself are trustworthy. The STV input representation is an invented construction with only internal evidence.

free parameters (9)
  • synthetic-onset-interval prior = U(0, 1.5) x previous duration
    Controls overlap distribution in generated training data; hand-chosen from literature range (Section 4.3).
  • synthetic-duration prior = U(85 ms, 1000 ms)
    Training-data generator hyperparameter (Section 4.3); affects what the detector can learn.
  • synthetic-displacement distribution = polynomial-weighted Gaussian, unspecified
    Generator's displacement model is not fully specified in the text (Section 4.3).
  • STV reversal threshold = 90 degrees
    Angle beyond which a direction reversal is encoded as a sign flip (Section 3.2); choice affects the input representation.
  • detection threshold = p >= 0.5
    Peak Detector converts onset probabilities to binary onsets at 0.5 (Section 4.2).
  • cold-start mixing ratio = 50%
    Proportion of cold-start vs distribution-adapted synthetic data during fine-tuning (Section 4.4).
  • receptive field = 99 samples, 1.65 s
    Chosen by hyperparameter optimization (Section 4.1); ablation shows it matters (F1 0.85 vs 0.80).
  • BCE positive-class weight alpha = not reported
    Class-imbalance weight in L_BCE (Section 4.5); value not given in the text.
  • adaptive false-positive penalty gamma/beta = gamma = max(1, |O_hat|/|O|), beta_t in {0.25, 0.5, 1}
    Anti-over-segmentation scheduling in fine-tuning (Section 4.5).
assumptions (6)
  • domain assumption Movement velocity profiles are superpositions of submovement primitives.
    Core model in Section 2.1 and Section 4.2 (Primitive Composer).
  • domain assumption Submovement velocity shape is minimum-jerk, parameterized only by onset, duration, and displacement.
    Section 4.2; limits the model to symmetric profiles; authors note a learnable-shape extension in Section 7.3.
  • domain assumption Signed tangential velocity with a 90-degree reversal threshold preserves the information needed for decomposition.
    Section 3.2; authors note it loses positional information (Section 7.3).
  • ad hoc to paper The model's own pseudo-labels on unlabeled human data are informative enough to guide synthetic-data refinement.
    Section 4.4; no human ground truth validates this premise.
  • domain assumption Additive Gaussian noise models realistic measurement noise.
    Sections 4.3 and 7.3; authors note real noise is sensor quantization and jitter.
  • domain assumption 60 Hz resampling preserves submovement information.
    Section 3.2; authors note sampling-rate limitations in Section 7.3.
invented entities (1)
  • Signed Tangential Velocity (STV)
    purpose: 1D signed input representation encoding direction reversals greater than 90 degrees.
    A preprocessing construction introduced in Section 3.2; no external validation, and the ablation shows removing it drops F1 from 0.85 to 0.77.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SSSUMO: Real-Time Semi-Supervised Submovement Decomposition." pith.science (2026). https://pith.science/paper/FRE3ZMDE

@misc{pith2026250708028,
  author       = {Pith},
  title        = {Pith review of: SSSUMO: Real-Time Semi-Supervised Submovement Decomposition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FRE3ZMDE}},
  note         = {Machine review of arXiv:2507.08028}
}
read the original abstract

This paper introduces a SSSUMO, semi-supervised deep learning approach for submovement decomposition that achieves state-of-the-art accuracy and speed. While submovement analysis offers valuable insights into motor control, existing methods struggle with reconstruction accuracy, computational cost, and validation, due to the difficulty of obtaining hand-labeled data. We address these challenges using a semi-supervised learning framework. This framework learns from synthetic data, initially generated from minimum-jerk principles and then iteratively refined through adaptation to unlabeled human movement data. Our fully convolutional architecture with differentiable reconstruction significantly surpasses existing methods on both synthetic and diverse human motion datasets, demonstrating robustness even in high-noise conditions. Crucially, the model operates in real-time (less than a millisecond per input second), a substantial improvement over optimization-based techniques. This enhanced performance facilitates new applications in human-computer interaction, rehabilitation medicine, and motor control studies. We demonstrate the model's effectiveness across diverse human-performed tasks such as steering, rotation, pointing, object moving, handwriting, and mouse-controlled gaming, showing notable improvements particularly on challenging datasets where traditional methods largely fail. Training and benchmarking source code, along with pre-trained model weights, are made publicly available at https://github.com/dolphin-in-a-coma/sssumo.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

93 extracted references · 54 canonical work pages

  1. [1]

    write newline

    " write newline " cite write " FUNCTION editor.postfix editor num.names #1 > "( )" "( )" if FUNCTION editor.trans.postfix editor num.names #1 > "( )" "( )" if FUNCTION trans.postfix translator num.names #1 > "( )" "( )" if FUNCTION authors.editors.reflist.apa5 'field := 'dot := field num.names 'numnames := numnames 'format.num.names := format.num.names na...

  2. [2]

    sn-aps.bst

    FUNCTION identify.aps.version "sn-aps.bst" " [2024/07/19 v1.1 APS bibliography style]" * top ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key keywords month note number organization pages publisher school series title type url volume year eprint archive archivePrefix primaryClass adsurl adsnote version lab...

    2024

  3. [3]

    write newline

    " write newline "" before.all 'output.state := FUNCTION if.digit duplicate "0" = swap duplicate "1" = swap duplicate "2" = swap duplicate "3" = swap duplicate "4" = swap duplicate "5" = swap duplicate "6" = swap duplicate "7" = swap duplicate "8" = swap "9" = or or or or or or or or or FUNCTION n.separate 't := "" #0 'numnames := t empty not t #-1 #1 subs...

  4. [4]

    : Fitts Law Revision: ilestz/ FittsLaw \_MSI

    botherref Al-Fawakhiri , N. : Fitts Law Revision: ilestz/ FittsLaw \_MSI. Zenodo (2024). 10.5281/zenodo.10642652 botherref

    work page doi:10.5281/zenodo.10642652 2024

  5. [5]

    : Components of complex movement that can be best studied using the sequence learning paradigm

    barticle Babu , R. : Components of complex movement that can be best studied using the sequence learning paradigm . Journal of Neurophysiology 124 ( 3 ), 645 -- 647 ( 2020 ) 10.1152/jn.00769.2019 . Publisher: American Physiological Society . Accessed 2025-03-24 barticle

    work page arXiv 2020

  6. [6]

    , Hinrichs , K

    bchapter Burgbacher , U. , Hinrichs , K. : Modeling Human Performance of Stroke - Based Text Entry . In: Proceedings of the 17th International Conference on Human - Computer Interaction with Mobile Devices and Services . MobileHCI '15 , pp. 137 -- 143 . Association for Computing Machinery , New York, NY, USA ( 2015 ). 10.1145/2785830.2785858 . https://doi...

    arXiv 2015

  7. [7]

    , Markkula , G

    barticle Benderius , O. , Markkula , G. : Evidence for a fundamental property of steering . Proceedings of the Human Factors and Ergonomics Society Annual Meeting 58 ( 1 ), 884 -- 888 ( 2014 ) 10.1177/1541931214581186 . Publisher: SAGE Publications Inc . Accessed 2025-03-24 barticle

    work page doi:10.1177/1541931214581186 2014

  8. [8]

    , Hosseini , S

    bchapter Cheema , N. , Hosseini , S. , Sprenger , J. , Herrmann , E. , Du , H. , Fischer , K. , Slusallek , P. : Dilated temporal fully-convolutional network for semantic segmentation of motion capture data . In: Proceedings of the Conference on Computer Animation . SCA Posters '18 , pp. 5 -- 6 . Eurographics Association , Goslar, DEU ( 2018 ). 10.2312/sc...

    work page doi:10.2312/sca.20181185 2018

Show all 93 references
  1. [9]

    , Rubiola , E

    barticle Cárdenas-Olaya , A.C. , Rubiola , E. , Friedt , J.-M. , Bourgeois , P.-Y. , Ortolano , M. , Micalizio , S. , Calosso , C.E. : Noise characterization of analog to digital converters for amplitude and phase noise measurements . Review of Scientific Instruments 88 ( 6 ),...

    2017 doi

  2. [10]

    : CogCarSim Gameplay Database (2025) 10.6084/m9.figshare.28714382.v1 botherref

    botherref Cowley , B. : CogCarSim Gameplay Database (2025) 10.6084/m9.figshare.28714382.v1 botherref

    2025 doi

  3. [11]

    , Palomäki , J

    barticle Cowley , B.U. , Palomäki , J. , Tammi , T. , Frantsi , R. , Inkilä , V.-P. , Lehtonen , N. , Pölönen , P. , Vepsäläinen , J. , Lappi , O. : Flow Experiences During Visuomotor Skill Acquisition Reflect Deviation From a Power - Law Learning Curve , but Not Overall Level...

    2019

  4. [12]

    , Riggs , S

    bchapter Clark , L. , Riggs , S. : VR - Based Kinematic Assessments : Examining the Effects of Task Properties on Arm Movement Kinematics . In: Extended Abstracts of the 2022 CHI Conference on Human Factors in Computing Systems . CHI EA '22 , pp. 1 -- 8 . Association for Compu...

    2022

  5. [13]

    , Krebs , H.I

    barticle Dipietro , L. , Krebs , H.I. , Fasoli , S.E. , Volpe , B.T. , Hogan , N. : Submovement changes characterize generalization of motor recovery after stroke . Cortex 45 ( 3 ), 318 -- 324 ( 2009 ) 10.1016/j.cortex.2008.02.008 . Accessed 2025-03-24 barticle

    2009 doi

  6. [14]

    , Plamondon , R

    barticle Djioua , M. , Plamondon , R. : The limit profile of a rapid movement velocity . Human Movement Science 29 ( 1 ), 48 -- 61 ( 2010 ) 10.1016/j.humov.2009.02.007 . Accessed 2025-05-01 barticle

    2010 doi

  7. [15]

    , Wisleder , D

    barticle Dounskaia , N. , Wisleder , D. , Johnson , T. : Influence of biomechanical factors on substructure of pointing movements . Experimental Brain Research 164 ( 4 ), 505 -- 516 ( 2005 ) 10.1007/s00221-005-2271-4 . Accessed 2025-03-24 barticle

    2005 doi

  8. [16]

    ACM Transactions on Computer-Human Interaction ( 2022 ) 10.1145/3524122

    barticle FischerFlorian , FleigArthur , KlarMarkus , MüllerJörg : Optimal Feedback Control for Modeling Human – Computer Interaction . ACM Transactions on Computer-Human Interaction ( 2022 ) 10.1145/3524122 . Publisher: ACMPUB27New York, NY . Accessed 2025-03-24 barticle

    2022 doi

  9. [17]

    , Jansen , A

    bchapter Findlater , L. , Jansen , A. , Shinohara , K. , Dixon , M. , Kamb , P. , Rakita , J. , Wobbrock , J.O. : Enhanced area cursors: reducing fine pointing demands for people with motor impairments . In: Proceedings of the 23nd Annual ACM Symposium on User Interface Softwa...

    2010

  10. [18]

    , Lee , G

    barticle Fradet , L. , Lee , G. , Dounskaia , N. : Origins of submovements during pointing movements . Acta Psychologica 129 ( 1 ), 91 -- 100 ( 2008 ) 10.1016/j.actpsy.2008.04.009 . Accessed 2025-03-24 barticle

    2008 doi

  11. [19]

    , Mattout , J

    barticle Friston , K. , Mattout , J. , Kilner , J. : Action understanding and active inference . Biological cybernetics 104 ( 1-2 ), 137 -- 160 ( 2011 ) 10.1007/s00422-011-0424-z . Accessed 2025-03-24 barticle

    2011 doi

  12. [20]

    : JasonFriedman /submovements: Code for decomposition of velocity data into submovements

    botherref Friedman , J. : JasonFriedman /submovements: Code for decomposition of velocity data into submovements. https://github.com/JasonFriedman/submovements Accessed 2025-04-30 botherref

    2025

  13. [21]

    , Bengio , Y

    bchapter Grandvalet , Y. , Bengio , Y. : Semi-supervised Learning by Entropy Minimization . In: Advances in Neural Information Processing Systems , vol. 17 . MIT Press , ??? ( 2004 ). https://proceedings.neurips.cc/paper_files/paper/2004/hash/96f2b50b5d3613adf9c27049b2a888c7-A...

    2004

  14. [22]

    , Lipinski , J

    barticle Grimme , B. , Lipinski , J. , Schöner , G. : Naturalistic arm movements during obstacle avoidance in 3D and the identification of movement primitives . Experimental Brain Research 222 ( 3 ), 185 -- 200 ( 2012 ) 10.1007/s00221-012-3205-6 . Number: 3 Publisher: Springer...

    2012 doi

  15. [23]

    , Overduin , S.A

    barticle Gowda , S. , Overduin , S.A. , Chen , M. , Chang , Y.-H. , Tomlin , C.J. , Carmena , J.M. : Accelerating Submovement Decomposition With Search - Space Reduction Heuristics . IEEE Transactions on Biomedical Engineering 62 ( 10 ), 2508 -- 2515 ( 2015 ) 10.1109/TBME.2015...

    2015

  16. [24]

    , Pouget-Abadie , J

    bchapter Goodfellow , I.J. , Pouget-Abadie , J. , Mirza , M. , Xu , B. , Warde-Farley , D. , Ozair , S. , Courville , A. , Bengio , Y. : Generative Adversarial Nets . In: Advances in Neural Information Processing Systems , vol. 27 . Curran Associates, Inc. , ??? ( 2014 ). http...

    2014

  17. [25]

    , de Carvalho , F

    barticle Hall , T.M. , de Carvalho , F. , Jackson , A. : A Common Structure Underlies Low - Frequency Cortical Dynamics in Movement , Sleep , and Sedation . Neuron 83 ( 5 ), 1185 -- 1199 ( 2014 ) 10.1016/j.neuron.2014.07.022 . Publisher: Elsevier . Accessed 2025-03-24 barticle

    2014 doi

  18. [26]

    : Brownian Process Repository: jameshermus/ brownianProcess

    botherref Hermus , J. : Brownian Process Repository: jameshermus/ brownianProcess . https://github.com/jameshermus/brownianProcess (2023) botherref

    2023

  19. [27]

    : An organizing principle for a class of voluntary movements

    barticle Hogan , N. : An organizing principle for a class of voluntary movements . Journal of Neuroscience 4 ( 11 ), 2745 -- 2754 ( 1984 ) 10.1523/JNEUROSCI.04-11-02745.1984 . Publisher: Society for Neuroscience Section: Articles . Accessed 2025-03-24 barticle

    1984

  20. [28]

    : The relationship between smoothness and performance during the practice of a lower limb obstacle avoidance task

    barticle Hreljac , A. : The relationship between smoothness and performance during the practice of a lower limb obstacle avoidance task . Biological Cybernetics 68 ( 4 ), 375 -- 379 ( 1993 ) 10.1007/BF00201862 . Accessed 2025-04-02 barticle

    1993 doi

  21. [29]

    , Sternad , D

    barticle Hogan , N. , Sternad , D. : Dynamic primitives of motor behavior . Biological Cybernetics 106 ( 11 ), 727 -- 739 ( 2012 ) 10.1007/s00422-012-0527-1 . Accessed 2025-03-24 barticle

    2012 doi

  22. [30]

    , Anquetil , E

    barticle Imbert , F. , Anquetil , E. , Soullard , Y. , Tavenard , R. : Mixture-of-experts for handwriting trajectory reconstruction from IMU sensors . Pattern Recognition 161 , 111231 ( 2025 ) 10.1016/j.patcog.2024.111231 barticle

    2025

  23. [31]

    , Bifulco , P

    barticle Iuppariello , L. , Bifulco , P. , Romano , M. , D'Addio , G. , Cesarelli , M. : A hybrid decomposition method to infer the sub-movements composition of planar reaching movements . Informatics in Medicine Unlocked 9 , 210 -- 218 ( 2017 ) 10.1016/j.imu.2017.09.004 . Acc...

    2017 doi

  24. [32]

    , Hersche , M

    bchapter Ingolfsson , T.M. , Hersche , M. , Wang , X. , Kobayashi , N. , Cavigelli , L. , Benini , L. : EEG - TCNet : An Accurate Temporal Convolutional Network for Embedded Motor - Imagery Brain – Machine Interfaces . In: 2020 IEEE International Conference on Systems , Man , ...

    2020

  25. [33]

    , Sakaguchi , Y

    barticle Inoue , Y. , Sakaguchi , Y. : A wavelet-based method for extracting intermittent discontinuities observed in human motor behavior . Neural Networks 62 , 91 -- 101 ( 2015 ) 10.1016/j.neunet.2014.05.004 . Accessed 2025-03-24 barticle

    2015 doi

  26. [34]

    , Li , J

    botherref Ji , Y. , Li , J. , Ye , H. , Wu , K. , Yao , K. , Xu , J. , Mo , L. , Zhang , M. : Test- Time Compute : from System -1 Thinking to System -2 Thinking . arXiv. arXiv:2501.02497 [cs] (2025). 10.48550/arXiv.2501.02497 . http://arxiv.org/abs/2501.02497 Accessed 2025-04-...

    2025 doi arXiv

  27. [35]

    , Fischer , F

    barticle Klar , M. , Fischer , F. , Fleig , A. , Bachinski , M. , Müller , J. : Simulating Interaction Movements via Model Predictive Control . ACM Trans. Comput.-Hum. Interact. 30 ( 3 ), 44 -- 14450 ( 2023 ) 10.1145/3577016 . Accessed 2025-03-24 barticle

    2023 doi

  28. [36]

    , Hadjiosif , A.M

    bchapter Krakauer , J.W. , Hadjiosif , A.M. , Xu , J. , Wong , A.L. , Haith , A.M. : Motor Learning . In: Terjung , R. (ed.) Comprehensive Physiology , 1 st edn., pp. 613 -- 663 . Wiley , ??? ( 2019 ). 10.1002/cphy.c170043 . https://onlinelibrary.wiley.com/doi/10.1002/cphy.c17...

    2019 doi

  29. [37]

    , Volpe , B

    barticle Krebs , H.I. , Volpe , B. , Hogan , N. : A working model of stroke recovery from rehabilitation robotics practitioners . Journal of NeuroEngineering and Rehabilitation 6 ( 1 ), 6 ( 2009 ) 10.1186/1743-0003-6-6 . Accessed 2025-03-24 barticle

    2009 doi

  30. [38]

    , Bicer , Y

    bchapter Lockwood , K. , Bicer , Y. , Asghari-Esfeden , S. , Zhu , T. , Furmanek , M. , Mangalam , M. , Strenge , G. , Imbiriba , T. , Yarossi , M. , Padir , T. , Erdogmus , D. , Tunik , E. : Leveraging Submovements for Prediction and Trajectory Planning for Human - Robot Hand...

    2022

  31. [39]

    : Pseudo- Label : The Simple and Efficient Semi - Supervised Learning Method for Deep Neural Networks (2013) botherref

    botherref Lee , D.-H. : Pseudo- Label : The Simple and Efficient Semi - Supervised Learning Method for Deep Neural Networks (2013) botherref

    2013

  32. [40]

    , Kirsch , R.F

    botherref Liao , J.Y. , Kirsch , R.F. : Characterizing and Predicting Submovements during Human Three - Dimensional Arm Reaches (2014) 10.1371/journal.pone.0103387 botherref

    2014 doi

  33. [41]

    , Kirsch , R.F

    botherref Liao , J.Y. , Kirsch , R.F. : Data S1 - Submovement Characterization: Characterizing and Predicting Submovements during Human Three-Dimensional Arm Reaches. PLOS ONE (2014). 10.1371/journal.pone.0103387.s001 botherref

    2014 doi

  34. [42]

    , Lee , S

    bchapter Lee , H. , Lee , S. , Nallapati , R. , Uh , Y. , Lee , B. : Characterizing and Quantifying Expert Input Behavior in League of Legends . In: Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems . CHI '24 , pp. 1 -- 21 . Association for Computing...

    2024

  35. [43]

    , Nancel , M

    bchapter Lee , B. , Nancel , M. , Kim , S. , Oulasvirta , A. : AutoGain : Gain Function Adaptation with Submovement Efficiency Optimization . In: Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems . CHI '20 , pp. 1 -- 12 . Association for Computing Ma...

    2020

  36. [44]

    , Abrams , R.A

    barticle Meyer , D.E. , Abrams , R.A. , Kornblum , S. , Wright , C.E. , Keith Smith , J.E. : Optimality in human motor performance: Ideal control of rapid aimed movements . Psychological Review 95 ( 3 ), 340 -- 370 ( 1988 ) 10.1037/0033-295X.95.3.340 . Place: US Publisher: Ame...

    1988 doi

  37. [45]

    , Boer , E

    barticle Markkula , G. , Boer , E. , Romano , R. , Merat , N. : Sustained sensorimotor control as intermittent decisions about prediction errors: computational framework and application to ground vehicle steering . Biological Cybernetics 112 ( 3 ), 181 -- 207 ( 2018 ) 10.1007/...

    2018 doi

  38. [46]

    , Daneault , J.-F

    barticle Miranda , J.G.V. , Daneault , J.-F. , Vergara-Diaz , G. , Torres , A.F.S.d.O.e. , Quixada , A.P. , Fonseca , M.d.L. , Vieira , J.P.B.C. , Santos , V.S. , Figueiredo , T.C. , Pinto , E.B. , Pena , N. , Bonato , P. : Complex Upper - Limb Movements Are Generated by Combi...

    2018 doi

  39. [47]

    , Gollee , H

    barticle Martín , J.A.A. , Gollee , H. , Müller , J. , Murray-Smith , R. : Intermittent Control as a Model of Mouse Movements . ACM Trans. Comput.-Hum. Interact. 28 ( 5 ), 35 -- 13546 ( 2021 ) 10.1145/3461836 . Accessed 2025-04-30 barticle

    2021 doi

  40. [48]

    : A model for the generation of movements requiring endpoint precision

    barticle Milner , T.E. : A model for the generation of movements requiring endpoint precision . Neuroscience 49 ( 2 ), 487 -- 496 ( 1992 ) 10.1016/0306-4522(92)90113-G . Accessed 2025-03-24 barticle

    1992 doi

  41. [49]

    , Kim , A.D

    barticle Meyer , T. , Kim , A.D. , Spivey , M. , Yoshimi , J. : Mouse tracking performance: A new approach to analyzing continuous mouse tracking data . Behavior Research Methods 56 ( 5 ), 4682 -- 4694 ( 2023 ) 10.3758/s13428-023-02210-5 . Number: 5 Publisher: Springer . Acces...

    2023 doi

  42. [50]

    , Yoshimi , J

    botherref Meyer , T. , Yoshimi , J. : Continuous Mouse Tracking Data (2021) 10.5281/zenodo.5181395 . Accessed 2024-12-05 botherref

    2021 doi

  43. [51]

    , Kashinath , K

    barticle O’Brien , T.A. , Kashinath , K. , Cavanaugh , N.R. , Collins , W.D. , O’Brien , J.P. : A fast and objective multidimensional kernel density estimation method: fastKDE . Computational Statistics & Data Analysis 101 , 148 -- 160 ( 2016 ) 10.1016/j.csda.2016.02.014 . Acc...

    2016 doi

  44. [52]

    , Djioua , M

    barticle Plamondon , R. , Djioua , M. : A multi-level representation paradigm for handwriting stroke generation . Human Movement Science 25 ( 4 ), 586 -- 607 ( 2006 ) 10.1016/j.humov.2006.07.004 . Accessed 2025-04-02 barticle

    2006 doi

  45. [53]

    , Gross , S

    botherref Paszke , A. , Gross , S. , Chintala , S. , Chanan , G. , Yang , E. , DeVito , Z. , Lin , Z. , Desmaison , A. , Antiga , L. , Lerer , A. : Automatic differentiation in PyTorch (2017). Accessed 2025-05-01 botherref

    2017

  46. [54]

    , Lee , B

    bchapter Park , E. , Lee , B. : An Intermittent Click Planning Model . In: Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems . CHI '20 , pp. 1 -- 13 . Association for Computing Machinery , New York, NY, USA ( 2020 ). 10.1145/3313831.3376725 . https:/...

    2020

  47. [55]

    : A kinematic theory of rapid human movements: Part I

    barticle Plamondon , R. : A kinematic theory of rapid human movements: Part I . Movement representation and generation . Biological Cybernetics 72 ( 4 ), 295 -- 307 ( 1995 ) 10.1007/BF00202785 . Accessed 2025-04-02 barticle

    1995 doi

  48. [56]

    , Lee , S

    bchapter Park , E. , Lee , S. , Ham , A. , Choi , M. , Kim , S. , Lee , B. : Secrets of Gosu : Understanding Physical Combat Skills of Professional Players in First - Person Shooters . In: Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems . CHI '21 ,...

    2021

  49. [57]

    , Roitman , A.V

    barticle Pasalar , S. , Roitman , A.V. , Ebner , T.J. : Effects of speeds and force fields on submovements during circular manual tracking in humans . Experimental Brain Research 163 ( 2 ), 214 -- 25 ( 2005 ) 10.1007/s00221-004-2169-6 . Num Pages: 214-25 Place: Heidelberg, Net...

    2005 doi

  50. [58]

    , Sobolewski , A

    barticle Pereira , M. , Sobolewski , A. , Millán , J.D.R. : Action Monitoring Cortical Activity Coupled to Submovements . eneuro 4 ( 5 ), 0241 -- 172017 ( 2017 ) 10.1523/ENEURO.0241-17.2017 . Accessed 2025-03-24 barticle

    2017 doi

  51. [59]

    , Tammi , T

    barticle Palomäki , J. , Tammi , T. , Lehtonen , N. , Seittenranta , N. , Laakasuo , M. , Abuhamdeh , S. , Lappi , O. , Cowley , B.U. : The link between flow and performance is moderated by task experience . Computers in Human Behavior 124 , 106891 ( 2021 ) 10.1016/j.chb.2021....

    2021

  52. [60]

    : Bochum Movement Data Repository: larslau/ Bochum \_movement\_data

    botherref Raket , L.L. : Bochum Movement Data Repository: larslau/ Bochum \_movement\_data. https://github.com/larslau/Bochum_movement_data (2016) botherref

    2016

  53. [61]

    , Grimme , B

    barticle Raket , L.L. , Grimme , B. , Schöner , G. , Igel , C. , Markussen , B. : Separating timing, movement conditions and individual differences in the analysis of human movement . PLOS Computational Biology 12 ( 9 ), 1 -- 27 ( 2016 ) 10.1371/journal.pcbi.1005092 barticle

    2016 doi

  54. [62]

    , Hogan , N

    barticle Rohrer , B. , Hogan , N. : Avoiding spurious submovement decompositions:a globally optimal algorithm. Biological Cybernetics 89 ( 3 ), 190 -- 199 ( 2003 ) 10.1007/s00422-003-0428-4 . Publisher: Springer Nature . Accessed 2025-03-24 barticle

    2003 doi

  55. [63]

    , Hogan , N

    barticle Rohrer , B. , Hogan , N. : Avoiding Spurious Submovement Decompositions II : A Scattershot Algorithm . Biological Cybernetics 94 ( 5 ), 409 -- 414 ( 2006 ) 10.1007/s00422-006-0055-y . Accessed 2025-03-24 barticle

    2006 doi

  56. [64]

    , Mohamed , S

    bchapter Rezende , D. , Mohamed , S. : Variational Inference with Normalizing Flows . In: Proceedings of the 32nd International Conference on Machine Learning , pp. 1530 -- 1538 . PMLR , ??? ( 2015 ). ISSN: 1938-7228 . https://proceedings.mlr.press/v37/rezende15.html Accessed ...

    2015

  57. [65]

    , Anquetil , E

    botherref Soullard , Y. , Anquetil , E. : IRISA - KIHT - S and KIHT -public Datasets – SHADoc (2024). https://www-shadoc.irisa.fr/irisa-kiht-s-and-kiht-public-datasets/ Accessed 2025-01-30 botherref

    2024

  58. [66]

    , Garcia-Romero , D

    bchapter Snyder , D. , Garcia-Romero , D. , Sell , G. , Povey , D. , Khudanpur , S. : X- Vectors : Robust DNN Embeddings for Speaker Recognition . In: 2018 IEEE International Conference on Acoustics , Speech and Signal Processing ( ICASSP ) , pp. 5329 -- 5333 ( 2018 ). 10.1109...

    2018

  59. [67]

    , Imbert , F

    barticle Swaileh , W. , Imbert , F. , Soullard , Y. , Tavenard , R. , Anquetil , E. : Online Handwriting Trajectory Reconstruction from Kinematic Sensors using Temporal Convolutional Network . International Journal on Document Analysis and Recognition ( 2023 ) 10.1007/s10032-0...

    2023 doi

  60. [68]

    , Krakauer , J.W

    barticle Shmuelof , L. , Krakauer , J.W. , Mazzoni , P. : How is a motor skill learned? Change and invariance at the levels of task success and trajectory control . Journal of Neurophysiology 108 ( 2 ), 578 -- 594 ( 2012 ) 10.1152/jn.00856.2011 . Publisher: American Physiologi...

    2012

  61. [69]

    , Röhm , A

    barticle Stelzer , F. , Röhm , A. , Vicente , R. , Fischer , I. , Yanchuk , S. : Deep neural networks using a single neuron: folded-in-time architecture using feedback-modulated delay loops . Nature Communications 12 ( 1 ), 5164 ( 2021 ) 10.1038/s41467-021-25427-4 . Publisher:...

    2021 doi

  62. [70]

    , Sattler , J.B

    barticle Stetter , L. , Sattler , J.B. , Marquardt , C. , Hermsdörfer , J. : Handwriting kinematics during learning to write with the dominant left hand in converted left-handers . Scientific Reports 13 ( 1 ), 2171 ( 2023 ) 10.1038/s41598-023-28911-7 . Publisher: Nature Publis...

    2023 doi

  63. [71]

    , Xu , W

    barticle Susilaradeya , D. , Xu , W. , Hall , T.M. , Galán , F. , Alter , K. , Jackson , A. : Extrinsic and intrinsic dynamics in movement intermittency . eLife 8 , 40145 ( 2019 ) 10.7554/eLife.40145 . Publisher: eLife Sciences Publications, Ltd . Accessed 2025-03-25 barticle

    2019 doi

  64. [72]

    , Hermus , J

    botherref Tessari , F. , Hermus , J. , Sugimoto-Dimitrova , R. , Hogan , N. : Crank Data Repository. https://www.dropbox.com/scl/fo/lbxsi4ce05hwjn6l1djhm/AAWrod-Son1sCXrz_vl50Ns/crank_data?dl=0&e=1&rlkey=cy5mq0sju3gqck6nz60hwtga3 (2023) botherref

    2023

  65. [73]

    , Hermus , J

    barticle Tessari , F. , Hermus , J. , Sugimoto-Dimitrova , R. , Hogan , N. : Brownian processes in human motor control support descending neural velocity commands . Scientific Reports 14 ( 1 ), 8341 ( 2024 ) 10.1038/s41598-024-58380-5 . Publisher: Nature Publishing Group . Acc...

    2024 doi

  66. [74]

    , Jordan , M

    bchapter Todorov , E. , Jordan , M. : A Minimal Intervention Principle for Coordinated Movement . In: Advances in Neural Information Processing Systems , vol. 15 . MIT Press , ??? ( 2002 ). https://papers.nips.cc/paper_files/paper/2002/hash/8c5f6ecd29a0eb234459190ca51c16dd-Abs...

    2002

  67. [75]

    , Tomassini , A

    barticle Torricelli , F. , Tomassini , A. , Pezzulo , G. , Pozzo , T. , Fadiga , L. , D'Ausilio , A. : Motor invariants in action execution and perception . Physics of Life Reviews 44 , 13 -- 47 ( 2023 ) 10.1016/j.plrev.2022.11.003 . Accessed 2025-02-06 barticle

    2023 doi

  68. [76]

    , Yan , J.H

    barticle Thomas , J.R. , Yan , J.H. , Stelmach , G.E. : Movement Substructures Change as a Function of Practice in Children and Adults . Journal of Experimental Child Psychology 75 ( 3 ), 228 -- 244 ( 2000 ) 10.1006/jecp.1999.2535 . Accessed 2025-03-24 barticle

  69. [77]

    , Gommers , R

    barticle Virtanen , P. , Gommers , R. , Oliphant , T.E. , Haberland , M. , Reddy , T. , Cournapeau , D. , Burovski , E. , Peterson , P. , Weckesser , W. , Bright , J. , Walt , S.J. , Brett , M. , Wilson , J. , Millman , K.J. , Mayorov , N. , Nelson , A.R.J. , Jones , E. , Kern...

    2020

  70. [78]

    : Structuring of Early Reaching Movements : A Longitudinal Study

    barticle Hofsten , C. : Structuring of Early Reaching Movements : A Longitudinal Study . Journal of Motor Behavior 23 ( 4 ), 280 -- 292 ( 1991 ) 10.1080/00222895.1991.9942039 . Publisher: Routledge . Accessed 2025-04-02 barticle

    1991

  71. [79]

    , Wessberg , J

    barticle Vallbo , A.B. , Wessberg , J. : Organization of motor output in slow finger movements in man. The Journal of Physiology 469 ( 1 ), 673 -- 691 ( 1993 ) 10.1113/jphysiol.1993.sp019837 . Publisher: John Wiley & Sons, Ltd . Accessed 2025-03-24 barticle

    1993 doi

  72. [80]

    : Accuracy of voluntary movement

    barticle Woodworth , R.S. : Accuracy of voluntary movement . The Psychological Review: Monograph Supplements 3 ( 3 ), 114 ( 1899 ) 10.1037/h0092992 . Place: US Publisher: The Macmillan Company barticle

    doi

  73. [81]

    , Soteropoulos , D.S

    barticle Williams , E.R. , Soteropoulos , D.S. , Baker , S.N. : Spinal interneuron circuits reduce approximately 10- Hz movement discontinuities by phase cancellation . Proceedings of the National Academy of Sciences 107 ( 24 ), 11098 -- 11103 ( 2010 ) 10.1073/pnas.0913373107 ...

    2010 doi

  74. [82]

    : Unsupervised word sense disambiguation rivaling supervised methods

    bchapter Yarowsky , D. : Unsupervised word sense disambiguation rivaling supervised methods . In: Proceedings of the 33rd Annual Meeting on Association for Computational Linguistics . ACL '95 , pp. 189 -- 196 . Association for Computational Linguistics , USA ( 1995 ). 10.3115/...

    1995

  75. [83]

    , Matic , A

    barticle Zago , M. , Matic , A. , Flash , T. , Gomez-Marin , A. , Lacquaniti , F. : The speed-curvature power law of movements: a reappraisal . Experimental Brain Research 236 ( 1 ), 69 -- 82 ( 2017 ) 10.1007/s00221-017-5108-z . Number: 1 Publisher: Springer . Accessed 2025-02...

    2017 doi

  76. [84]

    sn-basic.bst

    FUNCTION identify.basic.version "sn-basic.bst" " [2024/07/19 v1.1 bibliography style]" * top ENTRY address archive author booktitle chapter doi edition editor eid eprint howpublished institution journal key keywords month note number organization pages publisher school series ...

    2024

  77. [85]

    write newline

    " write newline "" before.all 'output.state := FUNCTION add.period duplicate empty 'skip "." * add.blank if FUNCTION if.digit duplicate "0" = swap duplicate "1" = swap duplicate "2" = swap duplicate "3" = swap duplicate "4" = swap duplicate "5" = swap duplicate "6" = swap dupl...

  78. [86]

    write newline

    " write newline "" before.all 'output.state := FUNCTION output.doi doi empty skip "doi:" doi * "" * output if FUNCTION format.archive archivePrefix empty "" archivePrefix ":" * if FUNCTION format.primaryClass primaryClass empty "" " [" primaryClass * "] " * if FUNCTION format....

  79. [87]

    write newline

    " write newline "" before.all 'output.state := FUNCTION string.to.integer 't := t text.length 'k := #1 'char.num := t char.num #1 substring 's := s is.num s "." = or char.num k = not and char.num #1 + 'char.num := while char.num #1 - 'char.num := t #1 char.num substring FUNCTI...

  80. [88]

    write newline

    " write newline "" before.all 'output.state := FUNCTION string.to.integer 't := t text.length 'k := #1 'char.num := t char.num #1 substring 's := s is.num s "." = or char.num k = not and char.num #1 + 'char.num := while char.num #1 - 'char.num := t #1 char.num substring FUNCTI...

  81. [89]

    sn-nature.bst

    FUNCTION identify.nature.version "sn-nature.bst" " [2024/07/19 v1.1 bibliography style]" * top ENTRY address archive author booktitle chapter edition editor eprint howpublished institution journal key keywords month note number organization pages publisher school series title ...

    2024

  82. [90]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  83. [91]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  84. [92]

    sn-vancouver-num.bst

    FUNCTION identify.vancouver.version "sn-vancouver-num.bst" " [2024/07/19 v1.1 Vancouver bibliography style]" * top ENTRY address assignee author booktitle chapter cartographer day edition editor howpublished institution inventor journal key keywords month note number organizat...

    2024

  85. [93]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.