41 / 2017-05-15 15:41:45

Analyze EEG signals with Convolutional Neural Network based on Power Spectrum Feature Selection

14270

Brain Computer Interface refers to establish a direct communication and control channels between the human brain and the computer. The essential part of the BCI system are feature extraction and classification. In this paper, we proposed an improvement architecture based on power spectrum and CNN, furthermore, two methods of power spectrum feature extraction is discussed. The result shows the identification system can efficiently detect ones’ right/left-hand Motor Imagination and has high classification accuracy. And the accuracy rate of up to 82%.