Table 4.

Parameter setting of the proposed method.

Parameter Setting	Resnet	Vit	ECA-Net	MLP
Conv Layer 1	$CNN(3),144$	/	/	/
Conv Layer 2	$\left[\begin{array}{c}CNN(1),144\\ CNN(3),144\\ CNN(1),256\end{array}\right]\times 3$	/	/	/
Conv Layer 3	$\left[\begin{array}{c}CNN(1),128\\ CNN(3),128\\ CNN(1),512\end{array}\right]\times 4$	/	/	/
Conv Layer 4	$\left[\begin{array}{c}CNN(1),256\\ CNN(3),256\\ CNN(1),1024\end{array}\right]\times 6$	/	/	/
Conv Layer 5	$\left[\begin{array}{c}CNN(1),512\\ CNN(3),512\\ CNN(1),2048\end{array}\right]\times 3$	/	/	/
Patch Size	/	(4 × 20)	/	/
Head Number	/	12	/	/
Encoder Layer Number	/	8	/	/
Embedding Dropout	/	0.1	/	/
Dropout	/	0.1	/	/
Kernel Size	/	/	2	/
Linear Layer Size	/	/	/	$\left[\begin{array}{c}FC(1024)\\ FC(2048)\\ FC(8)\end{array}\right]$
Dropout	/	/	/	0.2

Note: CNN (a),(b) are convolutional layers, where (a) is the size of convolutional kernel, (b) is the kernel number; FC (d) is fully connected layer, where (d) is the size of FC layer.