Pith. sign in

REVIEW 2 major objections 5 minor 66 references

Scaling and Distilling Transformer Models for sEMG

T0 review · 2 major / 5 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 109-million-parameter plain transformer decodes unseen users' typing from wrist sEMG at 30.5 character error rate, and a 2.2-million-parameter distilled version stays within 1.5 points.

desk verdict A well-documented empirical scaling and logit-distillation study on emg2qwerty with real gains, but the cross-user claim is built on only eight test users and seed-level error bars. read the letter →

arxiv 2507.22094 v1 pith:6RD3SCTW submitted 2025-07-29 eess.AS cs.AIcs.HCcs.LG

Nicholas Mehlman , Jean-Christophe Gagnon-Audet , Michael Shvartsman , Kelvin Niu , Alexander H. Miller , Shagun Sodhani This is my paper
classification eess.AScs.AIcs.HCcs.LG
keywords surfaceelectromyographytransformerscalingknowledgedistillationlogitcross-usergeneralizationemg2qwertycharactererrorrateCTCloss
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that the standard deep-learning recipe — train a plain transformer as large as compute allows, then shrink it with logit distillation — works for surface electromyography, the electrical muscle recordings that can drive hands-free interfaces. On the emg2qwerty typing benchmark, a 2.2M-parameter causal transformer with a learned convolutional featurizer lowers character error rate (CER) on users never seen during training from 55.57 for the TDS-ConvNet baseline to 35.9, and a 109M-parameter version reaches 30.5; across 20 architectures the improvement is nearly log-linear in parameter count up to that size. Distilling the 109M model's per-timestep output probabilities into the 2.2M student gives 31.9 CER, within 1.5 points of the teacher and 4 points better than training that student from scratch, and the distilled students also personalize better on individual users. If the claims hold, practitioners can stop hand-designing sEMG architectures and instead buy accuracy with scale and pay it back with the simplest possible compression.

What carries the argument

The load-bearing mechanism is a causal transformer encoder fed by a small convolutional featurizer that downsamples the 2 kHz raw sEMG stream to 100 Hz features; only the encoder depth (2-10 layers) and width (hidden size 128-1024) are varied, with the feed-forward dimension fixed at four times the hidden size and the CTC loss supervising the transcription. The compression mechanism is logit distillation: the student's per-timestep output distribution is pushed toward the teacher's temperature-2 soft targets by a cross-entropy term with weight 0.5, combined with the task loss, which is what lets a 2.2M student inherit the 109M teacher's inter-class knowledge. The emg2qwerty benchmark itself, with 108 users and 346 hours of naturalistic typing, is the third supporting piece because it gives the scaling curve enough data to appear.

What would settle it

Hold out a different set of eight users from the same dataset and repeat the 20-architecture scaling grid and the distillation protocol; the central claim fails if held-out CER stops improving between 2.2M and 109M parameters, or if the distilled 2.2M student lands more than 1.5 points above its teacher.

Watch

Extended reading notes

Core claim

The central discovery is that parameter count, rather than architectural novelty, is the main lever for cross-user sEMG performance once the task is large and naturalistic enough. Scaling only the depth and width of a vanilla transformer with a learned convolutional featurizer — from roughly 0.6M to 109M parameters — moves held-out-user CER on emg2qwerty from about 42 to 30.5 following an almost log-linear curve, and the best 109M model beats the TDS-ConvNet baseline by about 25 points absolute. The same 109M model, used as a teacher for logit distillation (cross-entropy on temperature-2 soft targets added to the CTC loss for unsegmented sequence transcription), transfers most of its accuracy to a 2.2M student: 31.9 CER versus the teacher's 30.5, a gap below 1.5 points at 50x fewer parameters and a 4.7x faster inference time. Distilled students also make better starting points for per-user personalization, improving CER from 9.7 to 8.6 for the Tiny student and from 7.9 to 7.1 for the Small student over the same architectures trained without the teacher signal.

Load-bearing premise

All cross-user conclusions rest on the assumption that the eight held-out users and their two held-out test sessions per user represent the broader population of sEMG users; if that group is unrepresentative or its sessions leak recording conditions into training, the scaling and distillation gains could be overstated.

Editorial extensions

If this is right

  • Cross-user sEMG transcription accuracy is not capped at the sub-10M-parameter regime: larger plain transformers keep improving up to at least 109M parameters on the same fixed dataset.
  • A 50x smaller student can recover most of the large model's accuracy, so the trade-off between edge-device latency and accuracy can be managed by distillation rather than by architecture search.
  • For small models practical on edge devices, training with a teacher logit signal is better than supervised training alone: 31.9 vs 35.9 CER for the 2.2M model and 32.7 vs 35.2 for the 5.4M model.
  • Distilled students fine-tune better on a single user's data, so the benefit of the teacher persists through personalization, not only in zero-shot evaluation: Tiny 8.6 vs 9.7 and Small 7.1 vs 7.9 CER.
  • A distilled 5.4M model runs about 4.7x faster than the 109M teacher per 4-second window (5.7 vs 27.0 ms), which is the difference between a model that fits a real-time interface and one that does not.

Reading between the lines

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

  • The paper leaves implicit that the same scaling-and-distillation recipe should be tested on other sEMG tasks; a natural extension would run the identical architecture grid on a gesture-recognition dataset with held-out users, where the scaling slope is likely shallower because isolated gestures carry less sequential structure than typing.
  • Because the teacher's soft labels encode inter-class similarity, the 50x compression limit is probably sensitive to the distillation temperature and weight; varying temperature while holding everything else fixed would map the trade-off between student capacity and inherited structure, which the paper does not do.
  • The results suggest, but do not prove, that even larger teachers would keep helping smaller students; a straightforward test is to train a teacher beyond 109M parameters or an ensemble of teachers and check whether the distilled 2.2M student's 1.5-point gap shrinks further.
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

2 major / 5 minor

Summary. The paper investigates whether vanilla transformer models can be scaled up on the emg2qwerty surface-EMG typing benchmark and whether logit distillation can compress them for deployment. Using a fixed convolutional featurizer followed by a causal transformer encoder, the authors train 20 architectures ranging from about 0.6M to 134M parameters and evaluate them on the benchmark's 8 held-out users in both zero-shot cross-user and per-user personalization settings. They report that a 2.2M-parameter Tiny transformer already outperforms the published TDS-ConvNet baseline by about 20 percentage points CER (35.9 vs 55.57), that a 109M-parameter Large transformer improves further to 30.5 CER, and that distilling the Large teacher into the Tiny student yields 31.9 CER, within 1.5 absolute points of the teacher with roughly 50x fewer parameters. Ablations attribute most of the gain to the learned raw-sEMG featurizer over log-spectrogram features, with a smaller additional contribution from the transformer encoder. The appendix provides the full model grid, hyperparameters, and code release.

Significance. If the results hold, they are practically significant for sEMG-based human-computer interfaces: they identify a regime where plain transformer scaling and the simplest logit distillation produce large gains over prior CNN baselines, and they offer a concrete recipe for deploying small real-time models. The manuscript is strong on reproducibility: it reports multiple seeds (6 for supervised, 3 for distillation/personalization), documents all hyperparameters in the appendix, releases code, reports the full architecture grid, and includes featurizer/encoder ablations that separate architecture from feature-input effects. The main weakness is statistical: all cross-user conclusions rest on only 8 test users, and no per-user variability is reported, which limits the population-level support for the headline claims.

major comments (2)
  1. [§3.1, Tables 2–3, §4.2–4.3] The paper's central cross-user claims are supported only by seed-level means of pooled CER over exactly 8 held-out users. All reported uncertainties (Tables 2, 3, 21, 22) are standard deviations across training seeds of the aggregate CER, not across users. Because CER is pooled over characters and users, one atypical user can dominate the mean, and sEMG is known for high inter-individual variability (Section 2.1). The manuscript should report per-user CER, paired user-level comparisons, and/or bootstrap confidence intervals over users. Without these, the headline gaps (Large vs Tiny: 5.4 CER; distilled Tiny vs supervised Tiny: 4.0 CER) cannot be distinguished from effects driven by a few responsive users, so the cross-user generalization claim is not yet statistically established.
  2. [§3.2, Table 2] The TDS-ConvNet baseline is taken as a single published number (55.57 CER) without standard deviation or a statement about whether it was re-run with the same protocol. The headline '20% absolute improvement' depends on this number. Although Table 4's own TDS re-implementation partially addresses this, please report the baseline's variance or explicitly state that it is a single published result.
minor comments (5)
  1. [Table 3 caption] The caption says 'Personalized models are personalized from the distilled student,' but the table includes both Standard and Distilled personalization columns; please rephrase to indicate that personalization is applied to checkpoints from both origins.
  2. [§3.2.2, Eq. (1)] The distillation temperature and alpha are chosen after tuning, but no sensitivity results are shown; a short table over alpha and temperature would make the choice's robustness transparent.
  3. [§4.2, Figure 3] The 'nearly log-linear' scaling claim is based on Pareto-front points only; the full grid (Table 21, Figure 7) contains non-monotonicity (e.g., 512-hidden 4-layer at 36.65 vs 256-hidden 4-layer at 34.89), so the main text should note that the curve is a Pareto-front summary rather than all models.
  4. [§3.2, Table 2] The TDS baseline lacks a standard deviation; please state explicitly whether it is a single published result and, if possible, provide variance from multiple runs.
  5. [Throughout] There are several typos, including 'contermporary' in Section 1, 'F eaturizer' in the Section 5 header, and 'V anilla' in Table 6; these should be cleaned up.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the scaling and distillation claims are measured against external held-out ground truth, not enforced by construction.

full rationale

The paper's central claims—that vanilla transformer performance on sEMG improves with scale up to 109M parameters and that logit distillation into 50x-smaller students preserves most of the gain—are empirical measurements on the emg2qwerty benchmark. The evaluation is external to the model definitions: Section 3.1 fixes 8 held-out test users and Section 3.3 defines CER as Levenshtein error against ground-truth character sequences. In Table 3, the distilled student's 31.9 CER is computed on that held-out test set after training with Eq. (1), which combines teacher logits with the CTC task loss; neither the teacher logits nor the student parameters enter the ground-truth CER computation, so the reported 4.0-point gain over the 35.9 supervised Tiny model is not forced by construction. Using the authors' own Large model as the teacher is appropriate for a distillation claim and does not define the outcome: Table 22 shows the smallest distilled students fail to reach the teacher's CER (e.g., 41.56 vs 30.38 for the 2-layer/128-dim student), so the student could and did underperform. Baseline numbers from Sivakumar et al. (2024) are external comparison points, not load-bearing justifications of the scaling or distillation conclusions. Hyperparameters and checkpoints are selected on validation splits and then reported on held-out test users, a standard protocol rather than a circular fit. The absence of per-user error bars is a statistical generalizability limitation, not circularity: the same held-out ground-truth labels would adjudicate a user-level analysis. No load-bearing step reduces to its own inputs by definition or by a self-citation chain.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

No new theoretical entities are introduced. The ledger lists experimental protocol assumptions that the empirical claims depend on, rather than fitted physical constants.

free parameters (2)
  • distillation loss weight alpha = 0.5
    Chosen by validation sweeps over [0.1, 2]; the reported 50x distillation benefit depends on this choice being near-optimal.
  • distillation temperature = 2
    Standard logit-distillation temperature from Hinton et al.; the exact value affects the soft-target distribution and therefore the student results.
assumptions (3)
  • domain assumption The 8-user held-out test split of emg2qwerty is representative of cross-user generalization.
    All cross-user scaling and distillation claims are measured on this single test group; if these users are atypical, the conclusions may not transfer. See Section 3.1.
  • domain assumption The TDS-ConvNet baseline numbers from Sivakumar et al. are directly comparable to the transformer results despite different input featurization.
    The paper uses the same train/validation/test splits but changes the input from spectrograms to raw sEMG with a learned featurizer; the ablation in Table 4 partially addresses this, but the headline SOTA gain is a joint featurizer-architecture difference.
  • domain assumption Post-hoc selection of learning rate and checkpoint on the validation set does not overfit the validation set enough to bias reported test CER.
    Appendix B.2 describes selecting the best validation learning rate and the epoch with lowest validation CER before reporting test results; this is standard practice but can inflate apparent gains when many architectures are compared.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Scaling and Distilling Transformer Models for sEMG." pith.science (2026). https://pith.science/paper/6RD3SCTW

@misc{pith2026250722094,
  author       = {Pith},
  title        = {Pith review of: Scaling and Distilling Transformer Models for sEMG},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6RD3SCTW}},
  note         = {Machine review of arXiv:2507.22094}
}
read the original abstract

Surface electromyography (sEMG) signals offer a promising avenue for developing innovative human-computer interfaces by providing insights into muscular activity. However, the limited volume of training data and computational constraints during deployment have restricted the investigation of scaling up the model size for solving sEMG tasks. In this paper, we demonstrate that vanilla transformer models can be effectively scaled up on sEMG data and yield improved cross-user performance up to 110M parameters, surpassing the model size regime investigated in other sEMG research (usually <10M parameters). We show that >100M-parameter models can be effectively distilled into models 50x smaller with minimal loss of performance (<1.5% absolute). This results in efficient and expressive models suitable for complex real-time sEMG tasks in real-world environments.

Figures

Figures reproduced from arXiv: 2507.22094 by the authors.

Figure 1
Figure 1. The emg2qwerty task: participants type on a keyboard while sEMG activity is recorded from both hands. The goal is to map from sequences of sEMG signals to sequences of characters. Figure cropped from https://github.com/facebookresearch/emg2qwerty, licensed CC BY-NC-SA. Improving on the above, Sivakumar et al. (2024) released a dataset that represents a significant advancement over existing sEMG benchmarks in terms o… view at source ↗
Figure 2
Figure 2. An illustration of the distillation pro [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Scaling curve of transformers on the emg2qwerty dataset, showing the benefit of model size across 3 orders of magnitude from <1M to over 100M parameters, and the benefit of distillation. A few things are of note: [a] even the smallest transformer we consider here (about 600K parameters) outperforms the 5.3M parameter TDS-ConvNet baseline (55.57 CER), in spite of having almost an order of magnitude fewer parameters; … view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Supervised training sample training curves for the [PITH_FULL_IMAGE:figures/full_fig_p026_4.png]
Figure 5
Figure 5. Figure 5: Distillation training sample training curves for the [PITH_FULL_IMAGE:figures/full_fig_p028_5.png]
Figure 6
Figure 6. Figure 6: Personalization training sample training curves for the [PITH_FULL_IMAGE:figures/full_fig_p029_6.png]
Figure 7
Figure 7. Figure 7: Supervised learning and distilled results on the [PITH_FULL_IMAGE:figures/full_fig_p030_7.png]
Figure 8
Figure 8. Figure 8: Performance improvement from distillation over supervised learning on the [PITH_FULL_IMAGE:figures/full_fig_p030_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

66 extracted references · 50 canonical work pages

  1. [1]

    u cahid G \

    Ahmet Alkan and M \"u cahid G \"u nay. Identification of emg signals using discriminant analysis and svm classifier. Expert systems with Applications, 39 0 (1): 0 44--47, 2012

    work page 2012

  2. [2]

    Advancing muscle-computer interfaces with high-density electromyography

    Christoph Amma, Thomas Krings, Jonas B\" o er, and Tanja Schultz. Advancing muscle-computer interfaces with high-density electromyography. In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems, CHI '15, pp.\ 929–938, New York, NY, USA, 2015. Association for Computing Machinery. ISBN 9781450331456. doi:10.1145/2702123.27025...

    arXiv 2015

  3. [3]

    Electromyography data for non-invasive naturally-controlled robotic hand prostheses

    M Atzori, A Gijsberts, C Castellini, B Caputo, AG Hager, S Elsig, G Giatsidis, F Bassetto, and H M \"u ller. Electromyography data for non-invasive naturally-controlled robotic hand prostheses. sci data. 2014; 1: 140053, 2014

    work page 2014

  4. [4]

    Deep learning with convolutional neural networks applied to electromyography data: A resource for the classification of movements for prosthetic hands

    Manfredo Atzori, Matteo Cognolato, and Henning Müller. Deep learning with convolutional neural networks applied to electromyography data: A resource for the classification of movements for prosthetic hands. Frontiers in Neurorobotics, 10, 2016. ISSN 1662-5218. doi:10.3389/fnbot.2016.00009. URL https://www.frontiersin.org/journals/neurorobotics/articles/10...

    work page arXiv 2016

  5. [5]

    Benalcazar, Lorena Barona, Leonardo Valdivieso, Xavier Aguas, and Jonathan Zea

    Marco E. Benalcazar, Lorena Barona, Leonardo Valdivieso, Xavier Aguas, and Jonathan Zea. Emg-epn-612 dataset, November 2020. URL https://doi.org/10.5281/zenodo.4421500

    work page doi:10.5281/zenodo.4421500 2020

  6. [6]

    Deep learning for processing electromyographic signals: A taxonomy-based survey

    Domenico Buongiorno, Giacomo Donato Cascarano, Irio De Feudis , Antonio Brunetti, Leonarda Carnimeo, Giovanni Dimauro, and Vitoantonio Bevilacqua. Deep learning for processing electromyographic signals: A taxonomy-based survey. Neurocomputing, 452: 0 549--565, 2021. ISSN 0925-2312. doi:https://doi.org/10.1016/j.neucom.2020.06.139. URL https://www.scienced...

    work page doi:10.1016/j.neucom.2020.06.139 2021

  7. [7]

    Machine-learning approaches for recognizing muscle activities involved in facial expressions captured by multi-channels surface electromyogram

    Yi Cai, Yifan Guo, Haotian Jiang, and Ming-Chun Huang. Machine-learning approaches for recognizing muscle activities involved in facial expressions captured by multi-channels surface electromyogram. Smart Health, 5: 0 15--25, 2018

    work page 2018

  8. [8]

    Cross-layer distillation with semantic calibration

    Defang Chen, Jian-Ping Mei, Yuan Zhang, Can Wang, Zhe Wang, Yan Feng, and Chun Chen. Cross-layer distillation with semantic calibration. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pp.\ 7028--7036, 2021

    work page 2021

Show all 66 references
  1. [9]

    Continuous motion finger joint angle estimation utilizing hybrid semg-fmg modality driven transformer-based deep learning model

    Zhouping Chen, Hong Wang, Haonan Chen, and Tao Wei. Continuous motion finger joint angle estimation utilizing hybrid semg-fmg modality driven transformer-based deep learning model. Biomedical Signal Processing and Control, 85: 0 105030, 2023. ISSN 1746-8094. doi:https://doi.or...

    2023

  2. [10]

    Chowdhury, Mamun Bin Ibne Reaz, Md

    Reaz H. Chowdhury, Mamun Bin Ibne Reaz, Md. Alauddin Ali, Ahmad Abu Bakar, Kasturi Chellappan, and Teck Guan Chang. Surface electromyography signal processing and classification techniques. Sensors (Basel), 13 0 (9): 0 12431--12466, Sep 17 2013. doi:10.3390/s130912431

    2013 doi

  3. [11]

    K. C. Cochrane-Snyman, T. J. Housh, C. M. Smith, E. C. Hill, N. D. Jenkins, R. J. Schmidt, and G. O. Johnson. Inter-individual variability in the patterns of responses for electromyography and mechanomyography during cycle ergometry using an rpe-clamp model. Eur J Appl Physiol...

    2016 doi

  4. [12]

    A generic noninvasive neuromotor interface for human-computer interaction

    CTRL-labs at Reality Labs . A generic noninvasive neuromotor interface for human-computer interaction. bioRxiv, 2024. doi:10.1101/2024.02.23.581779. URL https://www.biorxiv.org/content/early/2024/07/23/2024.02.23.581779

    2024 doi

  5. [13]

    Improved network and training scheme for cross-trial surface electromyography (semg)-based gesture recognition

    Qingfeng Dai, Yongkang Wong, Mohan Kankanhali, Xiangdong Li, and Weidong Geng. Improved network and training scheme for cross-trial surface electromyography (semg)-based gesture recognition. Bioengineering, 10 0 (9): 0 1101, 2023

    2023

  6. [14]

    Machine learning for detection of muscular activity from surface emg signals

    Francesco Di Nardo, Alessandro Nocera, Alessandro Cucchiarelli, Sandro Fioretti, and Chiara Morbidoni. Machine learning for detection of muscular activity from surface emg signals. Sensors (Basel), 22 0 (9): 0 3393, Apr 2022. doi:10.3390/s22093393

    2022 doi

  7. [15]

    Big data in myoelectric control: large multi-user models enable robust zero-shot emg-based discrete gesture recognition

    Ethan Eddy, Evan Campbell, Scott Bateman, and Erik Scheme. Big data in myoelectric control: large multi-user models enable robust zero-shot emg-based discrete gesture recognition. Frontiers in Bioengineering and Biotechnology, 12, 2024. ISSN 2296-4185. doi:10.3389/fbioe.2024.1...

    2024

  8. [16]

    Electromyography signal classification using deep learning

    Mekia Shigute Gaso, Selcuk Cankurt, and Abdulhamit Subasi. Electromyography signal classification using deep learning. In 2021 16th International Conference on Electronics Computer and Computation (ICECCO), pp.\ 1--6, 2021. doi:10.1109/ICECCO53203.2021.9663803

    2021

  9. [17]

    Godoy, Gustavo J

    Ricardo V. Godoy, Gustavo J. G. Lahr, Anany Dwivedi, Tharik J. S. Reis, Paulo H. Polegato, Marcelo Becker, Glauco A. P. Caurin, and Minas Liarokapis. Electromyography-based, robust hand motion classification employing temporal multi-channel vision transformers. IEEE Robotics a...

    2022

  10. [18]

    Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks

    Alex Graves, Santiago Fern\' a ndez, Faustino Gomez, and J\" u rgen Schmidhuber. Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks. In Proceedings of the 23rd International Conference on Machine Learning, ICML '06, pp.\ 3...

    2006

  11. [19]

    Sequence-to-sequence speech recognition with time-depth separable convolutions

    Awni Hannun, Ann Lee, Qiantong Xu, and Ronan Collobert. Sequence-to-sequence speech recognition with time-depth separable convolutions. arXiv preprint arXiv:1904.02619, 2019

    1904 arXiv

  12. [20]

    Surface emg pattern recognition using long short-term memory combined with multilayer perceptron

    Yunan He, Osamu Fukuda, Nan Bu, Hiroshi Okumura, and Nobuhiko Yamaguchi. Surface emg pattern recognition using long short-term memory combined with multilayer perceptron. In 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC...

    2018

  13. [21]

    Hinton, Oriol Vinyals, and Jeffrey Dean

    Geoffrey E. Hinton, Oriol Vinyals, and Jeffrey Dean. Distilling the knowledge in a neural network. ArXiv, abs/1503.02531, 2015. URL https://api.semanticscholar.org/CorpusID:7200347

    2015 arXiv

  14. [22]

    Paying more attention to attention: improving the performance of convolutional neural networks via attention transfer

    Nikos Komodakis and Sergey Zagoruyko. Paying more attention to attention: improving the performance of convolutional neural networks via attention transfer . In ICLR , Paris, France, June 2017. URL https://enpc.hal.science/hal-01832769

    2017

  15. [23]

    Knowledge distilled ensemble model for semg-based silent speech interface

    Wenqiang Lai, Qihan Yang, Ye Mao, Endong Sun, and Jiangnan Ye. Knowledge distilled ensemble model for semg-based silent speech interface. In IEEE EUROCON 2023-20th International Conference on Smart Technologies, pp.\ 117--122. IEEE, 2023

    2023

  16. [24]

    Fithubert: Going thinner and deeper for knowledge distillation of speech self-supervised learning, 2022

    Yeonghyeon Lee, Kangwook Jang, Jahyun Goo, Youngmoon Jung, and Hoirin Kim. Fithubert: Going thinner and deeper for knowledge distillation of speech self-supervised learning, 2022. URL https://arxiv.org/abs/2207.00555

    2022 arXiv

  17. [25]

    Gesture recognition using surface electromyography and deep learning for prostheses hand: state-of-the-art, challenges, and future

    Wei Li, Ping Shi, and Hongliu Yu. Gesture recognition using surface electromyography and deep learning for prostheses hand: state-of-the-art, challenges, and future. Frontiers in neuroscience, 15: 0 621885, 2021

    2021

  18. [26]

    Integration of convolutional neural network and vision transformer for gesture recognition using semg

    Xiaoguang Liu, Lijian Hu, Liang Tie, Li Jun, Xiaodong Wang, and Xiuling Liu. Integration of convolutional neural network and vision transformer for gesture recognition using semg. Biomedical Signal Processing and Control, 98: 0 106686, 2024. ISSN 1746-8094. doi:https://doi.org...

    2024

  19. [27]

    Decoupled weight decay regularization

    I Loshchilov. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017

    2017 arXiv

  20. [28]

    SGDR : Stochastic gradient descent with warm restarts

    Ilya Loshchilov and Frank Hutter. SGDR : Stochastic gradient descent with warm restarts. In International Conference on Learning Representations, 2017. URL https://openreview.net/forum?id=Skq89Scxx

    2017

  21. [29]

    An embedded electromyogram signal acquisition device

    Changjia Lu, Xin Xu, Yingjie Liu, Dan Li, Yue Wang, Wenhao Xian, Changbing Chen, Baichun Wei, and Jin Tian. An embedded electromyogram signal acquisition device. Sensors, 24 0 (13), 2024. ISSN 1424-8220. doi:10.3390/s24134106. URL https://www.mdpi.com/1424-8220/24/13/4106

    2024 doi

  22. [30]

    Behnam Mokhlesabadifarahani and V. K. Gunjan. EMG Signals Characterization in Three States of Contraction by Fuzzy Network and Feature Extraction. Springer Singapore Pte. Limited, 2015th edition, 2015. doi:10.1007/978-981-287-320-0

    2015 doi

  23. [31]

    Transformer-based hand gesture recognition from instantaneous to fused neural decomposition of high-density emg signals

    Mansooreh Montazerin, Elahe Rahimian, Farnoosh Naderkhani, S Farokh Atashzar, Svetlana Yanushkevich, and Arash Mohammadi. Transformer-based hand gesture recognition from instantaneous to fused neural decomposition of high-density emg signals. Scientific reports, 13 0 (1): 0 11...

    2023

  24. [32]

    Personal authentication by lips emg using dry electrode and cnn

    Shion Morikawa, Shin-ichi Ito, Momoyo Ito, and Minoru Fukumi. Personal authentication by lips emg using dry electrode and cnn. In 2018 IEEE International Conference on Internet of Things and Intelligence System (IOTAIS), pp.\ 180--183, 2018. doi:10.1109/IOTAIS.2018.8600859

    2018

  25. [33]

    BioPatRec: A modular research platform for the control of artificial limbs based on pattern recognition algorithms

    Max Ortiz-Catalan, Rickard Brånemark, and Bo Håkansson. BioPatRec: A modular research platform for the control of artificial limbs based on pattern recognition algorithms . Source Code for Biology and Medicine, 8: 0 11, 2013. doi:10.1186/1751-0473-8-11

    2013 doi

  26. [34]

    Emg based hand gesture recognition using deep learning

    Mehmet Akif Ozdemir, Deniz Hande Kisa, Onan Guren, Aytug Onan, and Aydin Akan. Emg based hand gesture recognition using deep learning. In 2020 Medical Technologies Congress (TIPTEKNO), pp.\ 1--4, 2020. doi:10.1109/TIPTEKNO50054.2020.9299264

    2020

  27. [35]

    Specaugment: A simple data augmentation method for automatic speech recognition

    Daniel S Park, William Chan, Yu Zhang, Chung-Cheng Chiu, Barret Zoph, Ekin D Cubuk, and Quoc V Le. Specaugment: A simple data augmentation method for automatic speech recognition. arXiv preprint arXiv:1904.08779, 2019 a

    1904 arXiv

  28. [36]

    Relational knowledge distillation

    Wonpyo Park, Dongju Kim, Yan Lu, and Minsu Cho. Relational knowledge distillation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp.\ 3967--3976, 2019 b

    2019

  29. [37]

    Dphubert: Joint distillation and pruning of self-supervised speech models, 2023

    Yifan Peng, Yui Sudo, Shakeel Muhammad, and Shinji Watanabe. Dphubert: Joint distillation and pruning of self-supervised speech models, 2023. URL https://arxiv.org/abs/2305.17651

    2023 arXiv

  30. [38]

    Robert Plonsey and Robert C. Barr. Bioelectricity: A quantitative approach. Springer, 2007

    2007

  31. [39]

    Efficiently scaling transformer inference

    Reiner Pope, Sholto Douglas, Aakanksha Chowdhery, Jacob Devlin, James Bradbury, Jonathan Heek, Kefan Xiao, Shivani Agrawal, and Jeff Dean. Efficiently scaling transformer inference. Proceedings of Machine Learning and Systems, 5: 0 606--624, 2023

    2023

  32. [40]

    Estimating finger joint angles by surface emg signal using feature extraction and transformer-based deep learning model

    Nur Achmad Sulistyo Putro, Cries Avian, Setya Widyawan Prakosa, Muhammad Izzuddin Mahali, and Jenq-Shiou Leu. Estimating finger joint angles by surface emg signal using feature extraction and transformer-based deep learning model. Biomedical Signal Processing and Control, 87: ...

    2024

  33. [41]

    Farokh Atashzar, and Arash Mohammadi

    Elahe Rahimian, Soheil Zabihi, Amir Asif, Dario Farina, S. Farokh Atashzar, and Arash Mohammadi. Temgnet: Deep transformer-based decoding of upperlimb semg for hand gestures recognition, 2021. URL https://arxiv.org/abs/2109.12379

    2021 arXiv

  34. [42]

    Enhancing gesture classification using active emg band and advanced feature extraction technique

    Parul Rani, Sidharth Pancholi, Vikash Shaw, Manfredo Atzori, and Sanjeev Kumar. Enhancing gesture classification using active emg band and advanced feature extraction technique. IEEE Sensors Journal, 24 0 (4): 0 5246--5255, 2024. doi:10.1109/JSEN.2023.3344700

    2024

  35. [43]

    Fitnets: Hints for thin deep nets

    Adriana Romero, Nicolas Ballas, Samira Ebrahimi Kahou, Antoine Chassang, Carlo Gatta, and Yoshua Bengio. Fitnets: Hints for thin deep nets. ICLR, abs/1412.6550, 2015. URL https://api.semanticscholar.org/CorpusID:2723173

    2015 arXiv

  36. [44]

    Demonstrating the feasibility of using forearm electromyography for muscle-computer interfaces

    T Scott Saponas, Desney S Tan, Dan Morris, and Ravin Balakrishnan. Demonstrating the feasibility of using forearm electromyography for muscle-computer interfaces. In Proceedings of the SIGCHI conference on human factors in computing systems, pp.\ 515--524, 2008

    2008

  37. [45]

    Multi-speaker speech synthesis from electromyographic signals by soft speech unit prediction

    Kevin Scheck and Tanja Schultz. Multi-speaker speech synthesis from electromyographic signals by soft speech unit prediction. In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.\ 1--5. IEEE, 2023

    2023

  38. [46]

    wav2vec: Unsupervised pre-training for speech recognition

    Steffen Schneider, Alexei Baevski, Ronan Collobert, and Michael Auli. wav2vec: Unsupervised pre-training for speech recognition. arXiv preprint arXiv:1904.05862, 2019

    1904 arXiv

  39. [47]

    Multiple kernel learning svm-based emg pattern classification for lower limb control

    Qingshan She, Zhizeng Luo, Ming Meng, and Ping Xu. Multiple kernel learning svm-based emg pattern classification for lower limb control. In 2010 11th International Conference on Control Automation Robotics & Vision, pp.\ 2109--2113. IEEE, 2010

    2010

  40. [48]

    Personal authentication and hand motion recognition based on wrist emg analysis by a convolutional neural network

    Ryohei Shioji, Shin ichi Ito, Momoyo Ito, and Minoru Fukumi. Personal authentication and hand motion recognition based on wrist emg analysis by a convolutional neural network. 2018 IEEE International Conference on Internet of Things and Intelligence System (IOTAIS), pp.\ 184--...

    2018

  41. [49]

    EMG2QWERTY: A Large Dataset with Baselines for Touch Typing using Surface Electromyography

    Viswanath Sivakumar, Jeffrey Seely, Alan Du, Sean Bittner, Adam Berenzweig, Anuoluwapo Bolarinwa, Alex Gramfort, and Michael Mandel. EMG2QWERTY: A Large Dataset with Baselines for Touch Typing using Surface Electromyography . In Proceedings of the Neural Information Processing...

    2024

  42. [50]

    Understanding and improving knowledge distillation

    Jiaxi Tang, Rakesh Shivanna, Zhe Zhao, Dong Lin, Anima Singh, Ed H Chi, and Sagar Jain. Understanding and improving knowledge distillation. arXiv preprint arXiv:2002.03532, 2020

    2002 arXiv

  43. [51]

    Similarity-preserving knowledge distillation

    Frederick Tung and Greg Mori. Similarity-preserving knowledge distillation. In Proceedings of the IEEE/CVF international conference on computer vision, pp.\ 1365--1374, 2019

    2019

  44. [52]

    Attention is all you need

    A Vaswani, N Shazeer, N Parmar, J Uszkoreit, L Jones, A Gomez, L Kaiser, and I Polosukhin. Attention is all you need. In NIPS, 2017

    2017

  45. [53]

    Deep neural network frontend for continuous emg-based speech recognition

    Michael Wand and J \"u rgen Schmidhuber. Deep neural network frontend for continuous emg-based speech recognition. In Interspeech, pp.\ 3032--3036, 2016

    2016

  46. [54]

    Exploring effective distillation of self-supervised speech models for automatic speech recognition, 2023

    Yujin Wang, Changli Tang, Ziyang Ma, Zhisheng Zheng, Xie Chen, and Wei-Qiang Zhang. Exploring effective distillation of self-supervised speech models for automatic speech recognition, 2023. URL https://arxiv.org/abs/2210.15631

    2023 arXiv

  47. [55]

    Lightweight transformer for semg gesture recognition with feature distilled variational information bottleneck

    Zefeng Wanga, Bingbing Hub, Junfeng Yaoa, and Jinsong Sua. Lightweight transformer for semg gesture recognition with feature distilled variational information bottleneck. 2024

    2024

  48. [56]

    L. N. Wimalasena, J. F. Braun, M. R. Keshtkaran, D. Hofmann, J. Á. Gallego, C. Alessandro, M. C. Tresch, L. E. Miller, and C. Pandarinath. Estimating muscle activation from emg using deep learning-based dynamical systems models. J Neural Eng, 19 0 (3), May 2022. doi:10.1088/17...

    2022 doi

  49. [57]

    Emg-based estimation of limb movement using deep learning with recurrent convolutional neural networks

    Peng Xia, Jie Hu, and Yinghong Peng. Emg-based estimation of limb movement using deep learning with recurrent convolutional neural networks. Artificial Organs, 42 0 (5): 0 E67--E77, 2018. doi:https://doi.org/10.1111/aor.13004. URL https://onlinelibrary.wiley.com/doi/abs/10.111...

    2018 doi

  50. [58]

    Training shallow and thin networks for acceleration via knowledge distillation with conditional adversarial networks, 2018

    Zheng Xu, Yen-Chang Hsu, and Jiawei Huang. Training shallow and thin networks for acceleration via knowledge distillation with conditional adversarial networks, 2018. URL https://arxiv.org/abs/1709.00513

    2018 arXiv

  51. [59]

    Emgbench: Benchmarking out-of-distribution generalization and adaptation for electromyography

    Jehan Yang, Maxwell Soh, Vivianna Lieu, Douglas Weber, and Zackory Erickson. Emgbench: Benchmarking out-of-distribution generalization and adaptation for electromyography. Advances in Neural Information Processing Systems, 37: 0 50313--50342, 2024

    2024

  52. [60]

    Trahgr: Transformer for hand gesture recognition via electromyography

    Soheil Zabihi, Elahe Rahimian, Amir Asif, and Arash Mohammadi. Trahgr: Transformer for hand gesture recognition via electromyography. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 31: 0 4211--4224, 2023. doi:10.1109/TNSRE.2023.3324252

    2023

  53. [61]

    Cross modality knowledge distillation between a-mode ultrasound and surface electromyography

    Jia Zeng, Yixuan Sheng, Yicheng Yang, Ziliang Zhou, and Honghai Liu. Cross modality knowledge distillation between a-mode ultrasound and surface electromyography. IEEE Transactions on Instrumentation and Measurement, 71: 0 1--9, 2022

    2022

  54. [62]

    Feasibility analysis of semg recognition via channel-wise transformer

    Jiaxuan Zhang, Yuki Matsuda, Manato Fujimoto, Hirohiko Suwa, and Keiichi Yasumoto. Feasibility analysis of semg recognition via channel-wise transformer. In 2022 IEEE 11th Global Conference on Consumer Electronics (GCCE), pp.\ 105--106. IEEE, 2022

    2022

  55. [63]

    Movement recognition via channel-activation-wise semg attention

    Jiaxuan Zhang, Yuki Matsuda, Manato Fujimoto, Hirohiko Suwa, and Keiichi Yasumoto. Movement recognition via channel-activation-wise semg attention. Methods, 218: 0 39--47, 2023 a

    2023

  56. [64]

    Lst-emg-net: Long short-term transformer feature fusion network for semg gesture recognition

    Wenli Zhang, Tingsong Zhao, Jianyi Zhang, and Yufei Wang. Lst-emg-net: Long short-term transformer feature fusion network for semg gesture recognition. Frontiers in Neurorobotics, 17, 2023 b . ISSN 1662-5218. doi:10.3389/fnbot.2023.1127338. URL https://www.frontiersin.org/jour...

    2023

  57. [65]

    Decoupled knowledge distillation

    Borui Zhao, Quan Cui, Renjie Song, Yiyu Qiu, and Jiajun Liang. Decoupled knowledge distillation. In Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition, pp.\ 11953--11962, 2022

    2022

  58. [66]

    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.