Table 5.

Results on the Korean native cattle data with different combinations of deepGBLUP components: (1) Deep learning networks ${\widehat{\mathbf{b}}}_{\text{deep}}$, (2) additive GBLUP ${\widehat{\mathbf{b}}}_{\text{a}}$, (3) dominance GBLUP ${\widehat{\mathbf{b}}}_{\text{d}}$, (4) epistasis GBLUP ${\widehat{\mathbf{b}}}_{\text{e}}$

Component				CWT	BF	EMA	MS
${\widehat{\mathbf{b}}}_{\text{deep}}$	${\widehat{\mathbf{b}}}_{\text{a}}$	${\widehat{\mathbf{b}}}_{\text{d}}$	${\widehat{\mathbf{b}}}_{\text{e}}$
✓				0.746 ± 0.017	0.661 ± 0.009	0.722 ± 0.014	0.622 ± 0.011
✓	✓			0.753 ± 0.015	0.673 ± 0.009	0.744 ± 0.016	0.666 ± 0.012
✓		✓		0.748 ± 0.017	0.659 ± 0.01	0.725 ± 0.014	0.623 ± 0.011
✓			✓	0.747 ± 0.016	0.671 ± 0.009	0.734 ± 0.016	0.646 ± 0.012
✓	✓	✓		0.755 ± 0.016	0.672 ± 0.009	0.746 ± 0.016	0.666 ± 0.012
✓	✓		✓	0.751 ± 0.015	0.673 ± 0.009	0.744 ± 0.017	0.672 ± 0.012
✓		✓	✓	0.748 ± 0.016	0.669 ± 0.009	0.736 ± 0.016	$\underset{\_}{0.647\pm 0.011}$
	✓	✓	✓	$\underset{\_}{0.725\pm 0.016}$	$\underset{\_}{0.639\pm 0.01}$	$\underset{\_}{0.722\pm 0.019}$	0.665 ± 0.014
✓	✓	✓	✓	0.752 ± 0.016	0.673 ± 0.009	0.746 ± 0.017	0.672 ± 0.012

The absence of a checkmark indicates that the corresponding component is excluded from the phenotype prediction. We highlight the best results in italic and the worst results in underline