Table 5.

Comparison of CNN and CNNSA for train.

Epoch	CNN			CNNSA
	Objective	Top-1 error	Top-5 error	Objective	Top-1 error	Top-5 error
1	1.4835700	0.10676	0.52868	0.009493	0.00092	0.00168
2	1.0443820	0.03664	0.36148	0.013218	0.00094	0.00188
3	0.9158232	0.02686	0.31518	0.010585	0.00094	0.0017
4	0.8279042	0.02176	0.28358	0.008023	0.00096	0.00188
5	0.7749367	0.01966	0.26404	0.009285	0.00106	0.00186
6	0.7314783	0.01750	0.25076	0.013674	0.00102	0.00175
7	0.6968027	0.01566	0.23968	0.117740	0.0009	0.00168
8	0.6654411	0.01398	0.22774	0.011239	0.0011	0.0018
9	0.6440073	0.01320	0.21978	0.011338	0.00106	0.0018
10	0.6213060	0.01312	0.20990	0.009957	0.00116	0.0019
11	0.6024042	0.01184	0.20716	0.008434	0.00096	0.00176
12	0.5786811	0.01090	0.19954	0.009425	0.0011	0.00192
13	0.5684009	0.01068	0.19548	0.012485	0.00082	0.0018
14	0.5486258	0.00994	0.18914	0.012108	0.00098	0.00184
15	0.5347288	0.00986	0.18446	0.009675	0.0011	0.00186