Using Antonyan Vardan Transform and Extreme Learning Machines for Accurate sEMG Signal Classification

Area of Science:

• Biomedical Engineering
• Machine Learning
• Signal Processing

Background:

• Surface electromyography (sEMG) is crucial for understanding muscle activity.
• Accurate classification of sEMG signals is essential for prosthetic control and human-computer interfaces.
• Existing methods often require complex feature engineering or computationally intensive models.

Purpose of the Study:

• To evaluate an adapted Antonyan Vardan Transform (AVT) combined with Extreme Learning Machines (ELM) for sEMG-based finger movement classification.
• To assess the system's performance on a standard PC and an embedded Raspberry Pi platform.
• To compare the proposed system's efficacy against existing literature benchmarks.

Main Methods:

• Utilized an adapted Antonyan Vardan Transform (AVT) for feature extraction from sEMG data.
• Employed an Extreme Learning Machines (ELM) classifier for signal processing.
• Analyzed data from 12 assays, comprising three repetitions by four volunteers, covering six finger movements and a rest state.
• Conducted a sample-by-sample output label comparison for comprehensive system analysis.

Main Results:

• The integrated AVT-ELM system demonstrated high performance in classifying six finger movements and a rest state from sEMG data.
• The system achieved mean accuracy rates exceeding 94%.
• Performance was comparable or superior to existing solutions in the literature, even with a simpler model architecture.

Conclusions:

• The proposed AVT-ELM system offers an efficient and accurate solution for sEMG-based gesture recognition.
• The system's ability to perform well on both PC and embedded platforms highlights its practical applicability.
• This approach provides a competitive alternative for applications requiring real-time sEMG signal processing.