Deep EEG: Deep learning in biomedical signal processing with EEG applications

Abstract

Abstract Brain-computer interface (BCI) or brain-machine interface (BMI) devices are communication systems that provide a direct communication pathway (bidirectional information flow) between brain activity signals and an external computer or other device. This chapter introduces the fundamentals of deep convolutional neural networks (DCNNs) and the deployment of TensorFlow and Keras. This chapter investigates the process of obtaining electroencephalogram (EEG) data in real time using a BCI device. The data are already converted into the frequency domain using the fast Fourier transform (FFT) algorithm. The python programming language and some of its associated scientific and numeric libraries are then used to preprocess the data. The popular TensorFlow and Keras machine learning libraries are used to train a convolutional neural network to classify the EEG data into four different commands: forward, left, right, and stop. Finally the trained neural network is used with embedded systems to control a robotic car/wheelchair.