Table 3.

Effects of hyperparameter variations through 10-fold cross validation in terms of prediction accuracy.

Parameters	Dt				Nr				Nc				Pd
	1	5	10	20	1	3	5	10	64	128	256	512	0	0.3	0.5	0.7
Training	0.9831	0.9861	0.9868	0.9865	0.9845	0.9868	0.9863	0.9860	0.9842	0.9848	0.9854	0.9868	0.9831	0.9868	0.9846	0.9815
Test	0.9807	0.9823	0.9828	0.9827	0.9806	0.9828	0.9826	0.9825	0.9819	0.9821	0.9823	0.9828	0.9827	0.9828	0.9820	0.9796

D_t, duplication time in the features enhancement module; N_r, the number of residual units in the deep feature learning module; N_c, the number of cells of fully connected neural network layer in deep feature learning module; P_d, dropout probability.