基于通道加权的多模态特征融合用于EEG疲劳驾驶检测
|
程文鑫,闫光辉,常文文,吴佰靖,黄亚宁
|
Channel-weighted multimodal feature fusion for EEG-based fatigue driving detection
|
Wenxin CHENG,Guanghui YAN,Wenwen CHANG,Baijing WU,Yaning HUANG
|
|
表 3 混合所有被试数据的消融实验结果 |
Tab.3 Ablation experimental results using mixed data from all subjects |
|
方法 | A | | P | | S | | M | | F1 | D1 | D2 | D1 | D2 | D1 | D2 | D1 | D2 | D1 | D2 | 未计算DE特征 | 0.665 1 | 0.702 4 | | 0.801 5 | 0.777 1 | | 0.702 2 | 0.731 9 | | 0.334 9 | 0.297 6 | | 0.755 3 | 0.731 0 | 未计算通道贡献度 | 0.861 9 | 0.825 5 | 0.896 4 | 0.865 1 | 0.876 6 | 0.799 0 | 0.138 1 | 0.174 5 | 0.889 5 | 0.878 9 | 仅使用GASF | 0.917 4 | 0.891 7 | 0.951 1 | 0.904 4 | 0.917 6 | 0.867 4 | 0.082 6 | 0.108 3 | 0.934 1 | 0.920 7 | 仅使用GADF | 0.897 2 | 0.902 2 | 0.908 7 | 0.912 5 | 0.923 0 | 0.903 3 | 0.102 8 | 0.097 8 | 0.915 8 | 0.902 2 | 未融合GRU | 0.906 4 | 0.899 3 | 0.922 9 | 0.909 9 | 0.924 7 | 0.895 4 | 0.093 6 | 0.100 7 | 0.923 9 | 0.914 8 | 未加入MSA | 0.904 0 | 0.882 4 | 0.925 9 | 0.921 3 | 0.915 6 | 0.908 7 | 0.096 0 | 0.117 6 | 0.921 1 | 0.907 7 | 本研究方法 | 0.933 7 | 0.907 8 | 0.955 9 | 0.924 6 | 0.934 5 | 0.919 7 | 0.066 3 | 0.092 2 | 0.945 1 | 0.922 1 |
|
|
|