Table 3.

ShallowNet architecture.

Layer	Input size	Output size	Kernels	Kernel size	Stride	Padding
Conv2d	1 × 3 × 300	10 × 3 × 300	10	(1, 21)	(1, 1)	(0, 10)
BatchNorm2d	10 × 3 × 300	10 × 3 × 300
Conv2d	10 × 3 × 300	15 × 1 × 300	15	(3, 1)	(1, 1)	(0, 0)
BatchNorm2d	15 × 1 × 300	15 × 1 × 300
Pow2	15 × 1 × 300	15 × 1 × 300
AvgPool2d	15 × 1 × 300	15 × 1 × 17		(1, 55)	(1, 15)	(0, 0)
Log	15 × 1 × 17	15 × 1 × 17
Dropout	15 × 1 × 17	15 × 1 × 17
Linear	255	2