Table 4.

10-fold cross validation performances on the first group of dataset

Method(Species)	Accuracy		Sensitivity		Specificity
	μ	σ	μ	σ	μ	σ
iNuc-PseKNC(CE)	86.90	x	90.30	x	83.55	x
iNuc-PseKNC(DM)	79.97	x	78,31	x	81.65	x
iNuc-PseKNC(HM)	86,27	x	87,86	x	84,70	x
DLNN-3(CE)	89.60	0.8	93.36	1.27	85.93	2,13
DLNN-3(DM)	85.54	1.13	87.60	2.55	83.42	2.65
DLNN-3(HM)	84.65	2.16	89.67	2.83	79.64	4.29
DLNN-5(CE)	89.62	2.45	93.04	3.68	86.34	5.54
DLNN-5(DM)	85.60	0.75	87.81	2.79	83.33	2.74
LNN-5(HM)	85.37	1.91	88.34	1,82	82.29	4.86

iNuc-PseKNC refers to the method introduced in [18]; CE, DM, HM refers to the datasets descried in Table 1; DLNN refers to the DLNN proposed in this paper and -3 or -5 refers to the kernel dimension in the first convolutional layer of the net. Best values are in bold