Related Experiment Videos
A Gaussian mixture model based classification scheme for myoelectric control of powered upper limb prostheses
Yonghong Huang1, Kevin B Englehart, Bernard Hudgins
1Department of Electrical and Computer Engineering and the Institute of Biomedical Engineering, University of New Brunswick, Fredericton, Canada.
IEEE Transactions on Bio-Medical Engineering
|November 16, 2005
Summary
Gaussian mixture models (GMMs) offer robust limb motion classification from myoelectric signals. This study optimizes GMM configurations, achieving high accuracy with low computational cost for prosthetic control.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Machine Learning
Background:
- Myoelectric signals are crucial for prosthetic limb control.
- Accurate classification of limb movements from these signals is challenging.
- Optimizing machine learning models is key to improving prosthetic functionality.
Purpose of the Study:
- To introduce and evaluate Gaussian mixture models (GMMs) for multiple limb motion classification.
- To optimize the configuration of GMMs for enhanced performance.
- To compare GMMs against other common classification algorithms.
Main Methods:
- Utilized continuous myoelectric signals from 12 subjects.
- Investigated algorithmic aspects of GMMs: model order selection, variance limiting, and data segmentation.
- Evaluated various feature sets, including time-domain and autoregressive features.
- Applied a majority vote rule for postprocessing results.
Main Results:
- GMMs demonstrated exceptional classification accuracy for limb motion.
- The optimized GMM system proved robust with a low computational load.
- Postprocessing with a majority vote rule enhanced performance.
- GMMs outperformed linear discriminant analysis, linear perceptron, and multilayer perceptron networks.
Conclusions:
- GMMs provide a highly accurate and robust method for myoelectric limb motion classification.
- Optimized GMM configurations offer a computationally efficient solution.
- This approach holds significant potential for advanced prosthetic limb control systems.