Channel Selection Method Based on CNNSE for EEG Emotion Recognition
作者
Zhongmin Wang,Hui Song,Shu-Yuan Hu,Ge Liu
标识
DOI:10.1109/iske47853.2019.9170317
摘要
Emotion recognition using electroencephalogram (EEG) signals has become a very important research direction in the field of artificial intelligence. However, traditional emotion recognition using all channels of EEG signals may cause computational complexity and interference information, thus it is not suitable for daily-life emotion recognition. This paper proposes a novel channel selection framework, called convolutional neural network with Squeeze-Excitation block (CNNSE), to select an optimal EEG channel set for emotion recognition. First, EEGLAB is used to obtain EEG timefrequency images. And the weights of all EEG channels are recalibrated by the CNNSE module. Finally, channel reduction is conducted by the analysis of obtained weight value. The experiments use publicly available dataset DEAP. It can be seen from the experimental results that the proposed method could maintain a higher recognition accuracy, and reduce EEG channels effectively.