Researcher Jaime Fenando Delgado Saa. Electronic Engineer University: Universidad del Norte Group: Robótica y Sistemas Inteligentes First, the signals are separated by trials in a 3D matrix and the the EOG interference is atenuated using linear regression, after that the singals are filtered using a highpass filter (4Hz). This algorithm use spectral features of EEG and mu and betha bands (for C3 and C4) as inputs for a neural network used as classifier. the output appear at time 4 and its continuous fron there unthil the end of each trial.