[Retracted] Gradient Descent Optimization in Deep Learning Model Training Based on Multistage and Method Combination Strategy

Table 12

Performance of the proposed method, SGD with decay.

	ResNet-20 on Cafri-10		LSTM on IMDB
	Val-loss	Val-acc	Val-loss	Val-acc
(SGD + d) + SGD	1.0363	0.6929	0.6888	0.5778
(SGD + d) + (SGD + M)	0.7713	0.7875	0.4271	0.8044
(SGD + d) + (SGD + d)	1.1088	0.6678	0.6888	0.5871
(SGD + d) + (SGD + M + d)	0.8374	0.7716	0.3906	0.8250
(SGD + d) + RMSprop	1.2161	0.7117	0.4230	0.8371
(SGD + d) + (RMSprop + d)	1.2169	0.6751	0.4650	0.8346
(SGD + d) + Adam	0.8920	0.7580	0.8291	0.7968
(SGD + d) + (Adam + d)	1.1164	0.6890	0.7607	0.8131

The bold values represent the best results.