Reproduce “Long-term stable Electromyography classification using Canonical Correlation Analysis” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Elisa Donati, Simone Benatti, Enea Ceolini, Giacomo Indiveri

A novel statistical method enhances long-term stability of electromyography (EMG) classification for hand gestures, maintaining approximately 95% accuracy over ten days with a single-shot calibration requiring only two repetitions per gesture. This approach leverages Canonical Correlation Analysis to compensate for signal variability, mitigating performance degradation without extensive retraining.

Key insights

  • Temporal variability in sEMG signals, often manifesting as shifts and rotations in feature space (e.g., due to electrode shifts), can be effectively compensated for by a linear transformation.
  • Canonical Correlation Analysis (CCA) successfully identifies and applies these optimal linear transformations to maximize the correlation between sEMG data acquired on different days from the same subject.
  • Maintaining robust, long-term sEMG classification accuracy is achievable with minimal daily calibration effort, offering a computationally efficient alternative to data-intensive deep learning models.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.