传感技术学报2025，Vol.38Issue(9)：1587-1596,10.DOI:10.3969/j.issn.1004-1699.2025.09.007
基于局部-全局时空特征学习的脑电情绪识别
EEG Emotion Recognition Based on Local-Global Spatiotemporal Feature Learning
摘要
Abstract
By analyzing the electroencephalogram(EEG)signal features,rapid and accurate recognition of human emotional states can be achieved.In emotion recognition tasks,there exists uncertainty in the continuous time variation of emotional intensity.At the same time,different functional areas of the cerebral cortex play different roles in tasks,therefore,the extraction of local spatial information is of ut-most importance.To address these issues,a spatial-temporal feature learning network for EEG emotion recognition,termed SPGAT is proposed.SPGAT employs a parallel approach to learn spatiotemporal features and subsequently fused spatial-temporally.Graph atten-tion networks are used to capture the relationship between local and global functional areas,and BiLSTM and attention modules are ex-ploited to capture the relationship between time duration and emotional intensity.Extensive experiments on the EEG emotion dataset,i.e.,DEAP,demonstrate the superiority of the proposed model over other advanced methods for EEG emotion recognition.Additionally,the visualization illustrates the functional impact of different brain regions on emotion recognition.关键词
脑电信号/情绪识别/图注意力网络/深度学习Key words
electroencephalogram signal/emotion recognition/graph attention network/deep learning分类
信息技术与安全科学引用本文复制引用
胡佳仑,代成龙,李光辉..基于局部-全局时空特征学习的脑电情绪识别[J].传感技术学报,2025,38(9):1587-1596,10.基金项目
江苏省自然科学基金项目(BK20210455) （BK20210455）
国家自然科学基金项目(62106087) （62106087）
