IncepA-EEGNet: 融合Inception网络和注意力机制的P300信号检测方法

许萌,王丹,李致远,陈远方

IncepA-EEGNet: P300 signal detection method based on fusion of Inception network and attention mechanism

Meng XU,Dan WANG,Zhi-yuan LI,Yuan-fang CHEN

表 4 不同CNN网络添加子模块对分类准确率的影响

Tab.4 Impact of adding sub-modules to different CNN networks on classification accuracy

添加模块 Acc 受试者 CNN-1 MCNN-1 MCNN-3 EEGNet 基础网络(Net) A 0.7037 0.6899 0.7038 0.7065 基础网络(Net) B 0.7065 0.6912 0.7037 0.7266 Net+Attention A 0.7092 0.6906 0.7091 0.7141 Net+Attention B 0.7185 0.7154 0.7192 0.7399 Net+Inception-v1 A 0.7100 0.6965 0.7103 0.7174 Net+Inception-v1 B 0.7222 0.7276 0.7203 0.7476 Net+Attention +Inception-v1 A 0.7186 0.7084 0.7258 0.7553 Net+Attention +Inception-v1 B 0.7454 0.7384 0.7478 0.7914