Deep Learning Structure for Cross-Domain Sentiment Classification Based on Improved Cross Entropy and Weight

Table 7

The variation of accuracy rate with the number of iterations under different loss functions.

Times loss function	Iteration
	1	2	3	4	5
Binary_crossentropy	82.68	82.39	81.57	81.27	80.76
Binary-optimize	81.88	81.52	83.02	82.99	82.21
Hinge	50.00	50.00	50.00	50.00	50.00
MAPE	50.00	50.00	50.00	50.00	50.00
MAE	54.78	75.38	77.25	78.04	78.81