A Multichannel Deep Belief Network for the Classification of EEG Data

Deep learning, and in particular Deep Belief Network (DBN), has recently witnessed increased attention from researchers as a new classification platform. It has been successfully applied to a number of classification problems, such as image classification, speech recognition and natural language processing. However, deep learning has not been fully explored in electroencephalogram (EEG) classification. We propose in this paper three implementations of DBNs to classify multichannel EEG data based on different channel fusion levels. In order to evaluate the proposed method, we used EEG data that has been recorded to study the modulatory effect of transcranial direct current stimulation. One of the proposed DBNs produced very promising results when compared to three well-established classifiers; which are Support Vector Machine (SVM), Linear Discriminant Analysis (LDA) and Extreme Learning Machine (ELM).