Experiments on neural networks with different configurations for electroencephalography (EEG) signal pattern classifications in imagination of direction

Here we present experimental results of classification methods for brain activity in the imagination of direction. We used a wireless portable electroencephalography (EEG) headset in our preceding study to collect EEG data from subjects in experiments, during which the subjects imagined arrows indicating one of the four directions: up, down, right, and left. The implemented classification methods consisted of a band-pass filter, fast Fourier transformation, principal component analysis, and neural network. We have applied neural networks with different configurations to the EEG data used in the preceding study in order to improve the classification rate. The experiments conducted in this study demonstrated some improvement results.