APPLICATION OF DEEP LEARNING METHODS IN BRAIN-COMPUTER INTERFACE SYSTEMS

Abstract

Deep learning, has produced many successful methods and architectures. Some of which are currently considered state-of-the-art in the areas of image classification and natural language processing. This thesis focuses on developing deep learning methods to the area of EEG classification. First, we propose a classification framework by introducing a new representation of the EEG data from extending the FBCSP method and utilizing a convolutional neural network. Our framework outperforms the state-of-the-art on a four-class motor-imagery dataset by a significant seven percent increase in accuracy. We have also analyzed and visualized the network for a more in-depth understanding. Second, we extend the application of deep learning to transfer learning in brain-computer interfaces by training a model on multiple subjects. The classification accuracy results produced in this thesis are stunningly higher relative to simple machine learning algorithms.