Table 1.

Performance comparisons of different word segmentation methods without Seq-GAN on the training dataset.

Layer	Word Segmentation Method	$\mathbf{Acc}$	$\mathbf{Sn}$	$\mathbf{Sp}$	$\mathbf{MCC}$
First Layer (Enhancer Identification)	Overlapped 3-Gram Word Segmentation	0.788	0.791	0.786	0.577
	Non-Overlapped 3-Gram Word Segmentation	0.773	0.777	0.769	0.546
	Word Segmentation Based on Statistics	0.784	0.765	0.803	0.569
Second Layer (Enhancer Strength Identification)	Overlapped 3-Gram Word Segmentation	0.675	0.714	0.635	0.350
	Non-Overlapped 3-Gram Word Segmentation	0.659	0.715	0.602	0.320
	Word Segmentation Based on Statistics	0.675	0.737	0.613	0.353