Fig. 4
From: Advancing BCI with a transformer-based model for motor imagery classification
Confusion matrices of the classification results for four models: EEGEncoder, ATCNet, TCNetFusion, and EEGTCNet. These matrices represent the average performance across nine subjects. The EEGEncoder model demonstrates superior accuracy in the ’Foot’ category compared to the other models, while the ’Tongue’ category shows relatively lower performance.
