Analisando o impacto do espectro do sinal de EEG na abordagem via Geometria Riemanniana

In this work we investigate the use of a Gaussian membership function as a technique to improve the steps of feature extraction and classification in brain-computer interface (BCI) systems based on motor imagery (IM). The main idea of this approach is to filter the spectral information of the electroencephalogram (EEG) signal via parameterized covariance matrices to highlight features that contribute to signal classification through a classifier based on Riemann’s distance. The results, in relation to the accuracy performance, acquired in this work arevalidated from dataset 2a of the IV International ICM Competition. The results obtained suggest that the spectral filtering performed using the Riemann Geometry approach can positively affect the performance of the ICM system, increasing its flexibility.