AUTHOR=Zhang Pengwei , Min Chongdan , Zhang Kangjia , Xue Wen , Chen Jingxia 

TITLE=Hierarchical Spatiotemporal Electroencephalogram Feature Learning and Emotion Recognition With Attention-Based Antagonism Neural Network

JOURNAL=Frontiers in Neuroscience

VOLUME=Volume 15 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2021.738167

DOI=10.3389/fnins.2021.738167

ISSN=1662-453X

ABSTRACT=Inspired by the neuroscience research results that human brain can produce dynamic response to different emotions, a new electroencephalogram (EEG) based human emotion classification model was proposed, named R2G-ST-BiLSTM, which uses a hierarchical neural network model to learn more discriminative spatio-temporal EEG features from local to global brain regions. Firstly, the bi-directional long and short-term memory (BiLSTM) network is used to obtain the internal spatial relationship of EEG signals on different channels within and between regions of the brain. Considering the different effects of various cerebral regions on emotions, the regional attention mechanism is introduced in the R2G-ST-BiLSTM model to determine the weight of different brain regions, which could enhance or weaken the contribution of each brain area to emotion recognition. Then a hierarchical BiLSTM network is again used to learn the spatiotemporal EEG features from regional to global brain areas, which are then input into an 
emotion classifier. Specially, we introduce a domain discriminator to work together with the classifier to reduce the domain offset between train and test data. Finally, we make experiment on EEG data of the large public DEAP data set. It is proved that our method achieves higher accuracy than that of the-state-of-the-art method.