Electroencephalogram Emotion Classification Method Based on Spatial Hierarchical Attention Mechanism

Different odor sources result in various emotions and induce various Electroencephalogram (EEG) signals in the brain. Therefore, the classification of the EEG signals is important to understand brain activity. To solve this issue, an emotion classification method using olfactory EEG signals based on the convolutional neural network and spatial layered attention mechanism is proposed in this work. The convolutional neural network is employed to extract features of EEG signals from multiple vector spaces. The attention mechanism is used to fuse features, which can realize the judgment of human emotions. Experiments are conducted on Kato's dataset and compared with traditional machine algorithms using accuracy, precision, recall, and F1 scores. The performance is improved by 16.80%, 18.30%, 18.72%, and 24.80% respectively, which verifies the effectiveness of the proposed approach.