Related Experiment Video
Updated: May 25, 2026
06:34
A Single-Channel and Non-Invasive Wearable Brain-Computer Interface for Industry and Healthcare
Published on: July 7, 2023
Hierarchical domain adaptation for SEMG signal classification across multiple subjects
Rita Chattopadhyay1, Narayanan C Krishnan, Sethuraman Panchanathan
1#Center for Cognitive Ubiquitous Computing, School of Computing, Informatics, and Decision Systems Engineering, Arizona State University, Tempe, Arizona 85287, USA.
Summary
This study introduces a new method to improve automated classification of Surface Electromyogram (SEMG) signals. The hierarchical sample selection technique enhances subject-independent accuracy in SEMG analysis.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Rehabilitation Technology
Background:
- Surface Electromyogram (SEMG) signals exhibit significant inter-subject variability, hindering generalized automated classification.
- Accurate SEMG classification is crucial for applications in prosthetics, diagnostics, and human-computer interaction.
Purpose of the Study:
- To develop a domain adaptation methodology to overcome inter-subject variability in SEMG data.
- To improve the accuracy and generalizability of automated SEMG signal classification across different individuals.
Main Methods:
- A hierarchical sample selection methodology was proposed to select relevant training data.
- Samples were chosen based on similarity to the target subject at multiple granularity levels.
- The framework was validated using SEMG data from 8 subjects during a fatiguing exercise protocol.
Main Results:
- The proposed domain adaptation method significantly improved subject-independent SEMG classification accuracy.
- Accuracy increased by 21% to 23% compared to approaches without domain adaptation.
- Improvements of 14% to 20% were observed over existing state-of-the-art domain adaptation techniques.
Conclusions:
- The hierarchical sample selection domain adaptation approach effectively addresses inter-subject variability in SEMG data.
- This methodology enhances the robustness and generalizability of automated SEMG classification systems.
- The findings suggest a promising direction for developing more reliable SEMG-based technologies.
