Author(s)

Dawei Yan

Corresponding Author

Dawei Yan

Abstract

Motor Imagery (MI) is a key research direction in brain–computer interface (BCI) systems, with widespread applications in neural rehabilitation and brain-controlled prosthetics. To address the challenges of low signal-to-noise ratio and strong temporal dependencies in electroencephalography (EEG) signals, this paper proposes an MI classification method based on spiking neural networks (SNN). By leveraging temporal encoding and neural dynamic modeling, the proposed approach effectively captures time-dependent features in EEG data. Experiments conducted on the BCI Competition IV 2008 dataset demonstrate that the method achieves an accuracy of 79.6%, representing a 5.2% improvement over conventional convolutional neural networks (CNNs), thereby validating the effectiveness and advantages of SNNs for MI classification task.

Keywords

Motor Imagery, Brain–Computer Interface, Electroencephalography, Spiking Neural Networks