Table 4.

Hyperparameters of selected CNN models from each population

Data	DE No.	Activation	No. layers	No. filters	Filter size	Epoch	FCL	Optimizer	Dropout	L2	Pooling
SP	1	linear	1	110	19	25	17	Adamax	0.197	0.21	Average
SP	2	Elu	1	16	15	32	110	rmsprop	0.146	0.03	Average
SP	3	Elu	1	15	8	44	79	rmsprop	0.692	0.02	Average
SP	4	linear	1	59	20	24	49	adamax	0.496	0.23	max
SP	5	linear	1	109	13	27	109	adam	0.827	0.01	Average
SC	1	linear	1	116	20	30	16	adam	0.370	0.10	Average
SC	2	linear	1	87	12	25	12	adam	0.086	0.13	Average
SC	3	linear	1	32	8	42	24	adam	0.250	0.19	Average
SC	4	linear	1	79	20	44	27	adamax	0.666	0.06	Max
SC	5	linear	1	98	16	40	153	adam	0.151	0.17	Average
RP	1	elu	2	[51,113]	18	22	50	adam	0.277	0.67	Average
RP	2	relu	3	[24,81,121]	12	27	268	adam	0.067	0.11	Average
RP	3	elu	2	[64,112]	13	45	278	adam	0.021	0.87	Average
RP	4	relu	3	[44,73,106]	13	47	326	adam	0.008	0.18	Average
RP	5	elu	3	[41,71,128]	5	41	238	adam	0.051	0.35	Average

SP, simulated pig dataset; SC, simulated cattle dataset; RP, real pig dataset; DE No., differential evolution of different data partitions; No. layers, number of convolutional layers; No. filters, number of filters applied based on No. layers; FCL, size (number of neurons) of the fully connected layer after flattened layer.