现代电子技术2025，Vol.48Issue(18)：159-164,6.DOI:10.16652/j.issn.1004-373x.2025.18.024
融合注意力的双分支时空卷积脑电识别网络
Dual-branch spatio-temporal convolution EEG recognition network integrating attention
摘要
Abstract
In allusion to the problems of weak feature expression of motor image(MI)electroencephalography(EEG)signals and low accuracy of classification and recognition,a dual-branch spatio-temporal convolutional motor imagery recognition network integrating attention mechanism is proposed.In order to address the problem of few data,an improved sliding window technology is proposed for the data augmentation.The data is fed into the feature extraction network,and the dual-branch structure is used to simultaneously extract the feature in both temporal and spatial aspects.On the temporal branch,multi-scale temporal convolution is used to extract temporal features at different scales.On the spatial branch,deep spatial features are used to extract depth-separable convolution.The attention mechanism is used to dynamically assign weights to the fused features.The comparison results with other methods show that the proposed network model has better classification performance and generalization ability.关键词
脑电信号/运动想象/多尺度时序卷积/双分支结构/数据增强/注意力机制Key words
EEG signal/motor imagery/multi-scale temporal convolution/dual-branch structure/data augmentation/attention mechanism分类
信息技术与安全科学引用本文复制引用
谷学静,周记帆,郭志斌..融合注意力的双分支时空卷积脑电识别网络[J].现代电子技术,2025,48(18):159-164,6.基金项目
唐山市沉浸式虚拟环境基础仿真团队项目(18130221A) （18130221A）
