Learning cross-regional dependence of EEG with convolutional neural networks for emotion classification

Seong-Eun Moon, Soobeom Jang, Jong-Seok Lee

Feb 12, 2018 (modified: Feb 12, 2018) ICLR 2018 Workshop Submission readers: everyone
  • Abstract: Electroencephalography (EEG) has received much attention because it provides comprehensive information of human perception and is relatively space- and cost-effective compared with the other functional brain imaging methods. In this paper, we present an approach to model undirected and directed cross-regional dependence of EEG signals with convolutional neural networks (CNNs). It considers the brain connectivity that measures simultaneous activation of different brain regions. Furthermore, the spatial arrangement of EEG electrodes is examined. We verify the effectiveness of the method for EEG-based emotion recognition.
  • TL;DR: We design connectivity matrix to model cross-regional dependence of EEG signals with CNNs.
  • Keywords: convolutional neural network, electroencephalography, brain connectivity