. 2016 Aug 8;11(8):e0160147. doi: 10.1371/journal.pone.0160147

Table 7. Comparison of recognition effect between dynamic supervised and unsupervised CEERM based on different numbers of RBM layers.

Model+Abstract Layers	R	P	F	FI	FIR(%)	FCI	FCIR(%)
DBN1+POSL+DPL+LENL+LOCL+DISL+TWFL	80.4	78.5	79.44	0	0	0	0
DBN2+POSL+DPL+LENL+LOCL+DISL+TWFL	89.5	83.6	86.45	7.01	8.83	5.56	6.43
DBN3+POSL+DPL+LENL+LOCL+DISL+TWFL	89.6	84.7	87.08	0.63	0.73	3.34	3.84
DBN4+POSL+DPL+LENL+LOCL+DISL+TWFL	90.32	86	88.11	1.03	1.18	2.94	3.34
DBN5+POSL+DPL+LENL+LOCL+DISL+TWFL	90.3	85	87.57	−0.54	−0.61	2.93	3.35
DBN6+POSL+DPL+LENL+LOCL+DISL+TWFL	89.7	84.8	87.18	−0.39	−0.44	4.02	4.61
DBN7+POSL+DPL+LENL+LOCL+DISL+TWFL	89.1	84.2	86.58	−0.6	−0.69	4.36	5.04
DBN8+POSL+DPL+LENL+LOCL+DISL+TWFL	90.1	82.4	86.08	−0.5	−0.58	4.74	5.51

Note: F_d is the F measure of the dynamic supervised classifier, F_u is the F measure of the unsupervised classifier, FCI is difference between F_d and F_u, FCIR is equal to $\frac{F_{d} - F_{u}}{F_{u}}$ .