Table 4.

CNN model architecture with input/output dimensions, filter size, and stride.

Layer	Input Dim	Padding	Output Dim	Filter Size	Output Dim	Maxpool, Stride	Output Dim
1	(1 × 282 × 40)	1	(1 × 284 × 42)	(1 × 3 × 3)	(16 × 40 × 282)	(2 × 2), 2	(16 × 20 × 141)
2	(16 × 20 × 141)	1	(16 × 22 × 143)	(16 × 3 × 3)	(32 × 20 × 141)	(4 × 4), 4	(32 × 5 × 35)
3	(32 × 5 × 35)	1	(32 × 7 × 37)	(32 × 3 × 3)	(64 × 5 × 35)	(4 × 4), 4	(64 × 1 × 8)
Flatten (64 × 1 × 8); final convolutional embedding length (1 × 512).