Table 3.

Configuration of CNN on Myxobacterial pictures in MY22.

Structure	Feature Extraction Layers	Fully Connected	Accuracy
1	Convolution Layer (fi (2,16), Padding (3,3))	F (O(22))	65.46%
	Max-Pooling (Pol (2,2), Stir(2,2))
	Convolution Layer (fi (2,32), Padding (3,3))
	Max-Pooling (Pol (2,2), Stir(2,2))
	Convolution Layer (fi (2,64), Padding (3,3))
	Max-Pooling (Pol (2,2), Stir(2,2))
	Convolution Layer (fi (2,128), Padding (3,3))
2	Convolution Layer (fi (3,16), Padding (1,1))	F (O(22))	74.246%
	Max-Pooling (Pol (2,2)), Stir(2,2))
	Convolution Layer (fi (3,32), Padding (1,1))
	Max-Pooling (Pol (2,2), Stir(2,2))
	Convolution Layer (fi (3,64), Padding (1,1))
3	Convolution Layer (fi (3,16), Padding (1,1))	F (O(22))	77.24%
	Max-Pooling (Pol (2,2)), Stir(2,2))
	Convolution Layer (fi (3,32), Padding (1,1))
	Max-Pooling (Pol (2,2), Stir(2,2))
	Convolution Layer (fi (3,64), Padding (1,1))
	Max-Pooling (Pol (2,2), Stir(2,2))
	Convolution Layer (fi (3,128), Padding (1,1))

fi: Filter.

M: Max-Pooling Layer.

Pol: Poll-size.

Stir: Stride.

O: Output.