A cross-dataset adaptive domain selection transfer learning framework for motor imagery-based brain-computer...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Minimizing calibration time is crucial for practical brain-computer interfaces (BCIs) using motor imagery (MI).
• Transfer learning (TL) shows promise in reducing MI-BCI calibration, but subject data distribution variations impact its effectiveness.
• Addressing data variability is key to enhancing TL performance in MI-BCIs.

Purpose of the Study:

• Propose a novel cross-dataset adaptive domain selection transfer learning framework for MI-BCIs.
• Enhance the efficiency and accuracy of MI-BCIs by optimizing TL through domain selection and data alignment.
• Improve the practical applicability of MI-BCIs by reducing reliance on extensive subject-specific training data.

Main Methods:

• Developed a framework integrating domain selection, data alignment, and an enhanced Common Spatial Pattern (CSP) algorithm.
• Utilized a large dataset of 109 subjects as the source domain, identifying aligned subjects using maximum mean discrepancy.
• Employed Euclidean alignment and multiple composite CSP for feature extraction, with Support Vector Machine for final classification.

Main Results:

• Achieved classification accuracies of 75.05% and 76.82% in two cross-dataset experiments, outperforming existing methods.
• Demonstrated the effectiveness of the adaptive domain selection strategy in handling data distribution variations.
• Validated the framework's ability to maintain high classification performance with reduced calibration.

Conclusions:

• The proposed transfer learning framework significantly optimizes MI-BCI implementation by reducing calibration needs while preserving high accuracy.
• This approach enhances the real-world viability of BCIs by making them more accessible and user-friendly.
• Future research can further refine domain adaptation techniques for even greater MI-BCI performance.