Related Experiment Video
Updated: Aug 6, 2026
09:44
Recording and Analyzing Multimodal Large-Scale Neuronal Ensemble Dynamics on CMOS-Integrated High-Density Microelectrode Array
Published on: March 8, 2024
NewRep: a classification model for online pattern recognition based on HD-EMG signals
Lizhi Pan1,2, Shiwei Chen1,2, Jianmin Li1,2
1The Key Laboratory of Mechanism Theory and Equipment Design of Ministry of Education, School of Mechanical Engineering, Tianjin University, 135 Yaguan Road, Jinnan District, Tianjin, 300350, China.
Medical & Biological Engineering & Computing
|July 21, 2026
Summary
A new neural network, NewRep, enhances electromyography (EMG) pattern recognition for myoelectric control. It significantly improves online gesture classification accuracy and reduces motion completion time for practical applications.
Area of Science:
- Biomedical Engineering
- Machine Learning
- Neuroscience
Background:
- High-density electromyography (HD-EMG) signals are crucial for pattern recognition (PR) in myoelectric control, especially for gesture recognition.
- While neural networks excel in EMG PR, improving recognition accuracy remains a challenge for real-time applications.
Purpose of the Study:
- To develop a novel neural network, NewRep, by integrating a self-attention mechanism into the RepViT architecture.
- To enhance the online classification performance of neural networks for EMG-based gesture recognition.
Main Methods:
- A new neural network, NewRep, was created by incorporating a self-attention mechanism into RepViT, a lightweight convolutional neural network.
- Online gesture classification was conducted using NewRep, RepViT, and linear discriminant analysis with 256-channel EMG data from 18 subjects performing 20 hand/wrist motions.
Main Results:
- NewRep demonstrated significant improvements over RepViT and linear discriminant analysis.
- Specifically, NewRep reduced motion completion time by 0.053 s and 0.117 s, respectively.
- Online classification accuracy increased by 2.91% and 6.24% compared to RepViT and linear discriminant analysis, respectively.
Conclusions:
- The proposed NewRep model effectively enhances real-time decoding capabilities in myoelectric control.
- NewRep offers a promising advancement for practical applications in prosthetic limb control and human-computer interaction.