DAMC-Net: Dual-Branch Attention-Guided Multi-Scale Convolutional Network for EEG-Based Imagined Speech Decoding
Under review
Research approach
Classifies imagined-speech EEG using temporal and spatial multiscale branches, SimAM and log-power aggregation.
Results
A subject-dependent model for 15 participants and five words in BCI Competition Track 3. The implementation has 55,749 parameters; the manuscript is under review at AAAI.
All publications