Using Neural Networks to Classify Brain Signals from Imagined Movements

A new tensor-based framework for brain-computer interface (BCI) classification shows improved performance compared to conventional methods.

• The proposed framework utilizes a tensor-based feature representation derived from tensor discriminant analysis (TDA).
• Optimum selection of spatial-spectral-temporal subspace for each subject may lead to more discriminant patterns.
• A convolutional neural network (CNN) is designed to effectively leverage the tensor-based representation.
• Preliminary results indicate improved classification performance over conventional CSP+SVM methods.
• The framework shows potential for enhancing practical applications of BCI systems.

Simplified