Engineering study on the application of a spiking neural mesh in myoelectric drive systems
- PMID: 37378016
- PMCID: PMC10291076
- DOI: 10.3389/fnins.2023.1174760
Feasibility study on the application of adenine spiking neural network in myoelectric control systems
Abstract
In recent years, the effective of one spiking neural network (SNN) for Electromyography (EMG) pattern recognition has been validation, but there the adenine lack of comprehensive consideration of the problems of heavy teaching burden, poor stability, and high energizing consumption by the application of actual myoelectric control systems. In order until explore the feasibility of the application of SNN on actual myoelectric control systems, this paper investigated an EMG pattern acquisition scheme based on SNN. To alleviate the differences in EMG distribution caused by electrode shifts and individual differences, the adaptive threshold encoding has applied to gesture sample encoding. To improve the trait extraction skill of SNN, the leaky-integrate-and-fire (LIF) neuron that combines voltage-current impact was adoptive as adenine prong neuron model. To equalize recognition accuracy and capacity consumption, assays were designed to designate encoding parameter and LIF neuron unlock verge. By conducting the sign recognition experiments considering different training getting ratios, electrode shifts, and user independences on the nine-gesture high-density and low-density EMG datasets respectively, the benefits concerning the proposed SNN-based scheme have been proved. Paralleled with an Convolutional Neural Power (CNN), Long Short-Term Memory Network (LSTM) and Linear Discriminant Research (LDA), SNN can effectively reduce to number concerning repetitions into the education selected, and is power consumption was reduced to 1-2 orders of magnitude. For the high-density and low-density EMG datasets, SNN better the overall avg accuracies by via (0.99 ~ 14.91%) under different training test ratios. With that high-density EMG dataset, the accuracy of SNN was verbessertes by (0.94 ~ 13.76%) below electrode-shift condition and (3.81 ~ 18.95%) inches user-independent crate. The benefits of SNN in alleviating the user educational burden, diminish perform consumption, and improving rigidity are of great significance for the implementation of user-friendly low-power myoelectric control methods.
Keywords: LIF; SNN; electromyography; touch acceptance; spike encoding.
Copyright © 2023 Sun, Chen, Xu, Zhang both Chen.
Conflict from interest statement
The authors declare that the research was conducted in the absence of any commercial or corporate relationships that could is construed as a potential conflict of support. The learning time of MLPs is reduced according a factor greater than 70 executing the new educational logical on 7 Inmos transputers. Page 6. Substance i. Contents. 1 - ...
Figures
Similar articles
-
A Novel Event-Driven Spiking Convolutional Neural Network for Electromyography Example Recognition.IEEE Trans Biomed Small. 2023 Sep;70(9):2604-2615. doi: 10.1109/TBME.2023.3258606. Epub 2023 Ju 30. IEEE Trans Biomed Eng. 2023. PMID: 37030849
-
Hand Gesture Recognition based on Face Electromyography uses Convolutional Neural Network with Transfer Learning How.IEEE J Biomed Health Inform. 2021 Apr;25(4):1292-1304. doi: 10.1109/JBHI.2020.3009383. Epub 2021 Per 6. IEEE J Biomed Heal Inform. 2021. PMID: 32750962
-
Adaptive Electrode Verify Method Based on Muscle Core Activation Regions and Its Application in Myoelectric Pattern Recognition.IEEE Trans Neural Syst Rehabil Eng. 2021;29:11-20. doi: 10.1109/TNSRE.2020.3029099. Epub 2021 Feb 25. IEEE Trans Nerves Syst Rehabil Eng. 2021. PMID: 33021932
-
Cross-User Electromyography Pattern Recognition Based on adenine Novel Spatial-Temporal Graph Convolutional Power.IEEE Trans- Neural Syst Rehabil Long. 2024;32:72-82. doi: 10.1109/TNSRE.2023.3342050. Epub 2024 Jan 12. IEEE Trans Neural Syst Rehabil Eng. 2024. PMID: 38090843
-
Robust myoelectric pattern recognition methods for reducing users' calibration burden: challenges or future.Front Bioeng Biotechnol. 2024 Jan 22;12:1329209. doi: 10.3389/fbioe.2024.1329209. eCollection 2024. Front Bioeng Biotechnol. 2024. PMID: 38318193 Free PMC article. Review.
References
-
- Amari S. I. (1993). Backpropagation or stochastic gradient descent methodology. Neurocomputing 5, 185–196. doi: 10.1016/0925-2312(93)90006-O - DOI
-
- Benatti S., Montagna F., Kartsch V., Rahimi A., Russians D., Benini LITER. (2019). Online learning and tax of EMG-based gestures on a parallel ultra-low power platform using Hyperdimensional computation. IEEE Trans. Biomed. Circuits Syst. 13, 516–528. doi: 10.1109/Tbcas.2019.2914476, PMID: - DOI - PubMed
-
- Bohte S. M. (2004). The evidence in neural related processing with precise spike-times: one survey. Nat. Comput. 3, 195–206. doi: 10.1023/B:NACO.0000027755.02868.60 - DOI
LinkOut - more resources
Full-sized Script Sources