Brain-computer interface (BCI) technologies enable direct communications between humans and computers by analyzing EEG signals. One of the challenges with traditional methods in classification tasks is receiving unsatisfactory recognition effects from EEG signals. In recent years, deep learning has drawn a great deal of attentions in diverse research fields, and could provide a novel solution for learning robust representations from EEG signals. In this chapter, we firstly introduce the basic concepts of deep learning techniques and two commonly used structures in time series analysis, namely, convolutional neural network and recurrent neural network. Then, we provide the applications of these two DL models to focus on the eye state detection task, which both achieve excellent recognition effects and are expected to be useful for broader applications in BCI systems.
KeywordsFatigue detection EEG analysis Brain-computer interface Deep learning
- Golmohammadi M, et al. Gated recurrent networks for seizure detection. In: Signal Processing in Medicine and Biology Symposium (SPMB). 2017. p. 1–5.Google Scholar
- Graves A, Schmidhuber J. Framewise phoneme classification with bidirectional LSTM networks. In: International Joint Conference on Neural Networks. 2005. 2047–2052.Google Scholar
- Gulli A, Pal S. Deep Learning with Keras. Birmingham: Packt Publishing Ltd; 2017.Google Scholar
- Ioffe S, Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In: International Conference on Machine Learning. 2015. p. 448–456.Google Scholar
- Nair V., Hinton GE. Rectified linear units improve restricted boltzmann machines. In: International conference on machine learning. 2010. 807–814.Google Scholar
- Salazar-Gomez AF, et al. Correcting robot mistakes in real time using EEG signals. In: IEEE International Conference on Robotics and Automation (ICRA). 2017. p. 6570–6577.Google Scholar