A Bridge Structure 3D Representation for Deep Neural Network and Its Application in Frequency Estimation

Table 3

Results of all models on the test set.

No.			Total loss	1^st	2^nd	3^rd	4^th	5^th	Speed
1	0	0	0.3867	0.0471	0.0577	0.0750	0.0945	0.1124	8
				15.5%	13.9%	12.3%	12.7%	12.4%
2	0	1	0.3901	0.0472	0.0585	0.0758	0.0947	0.1139	8
				15.3%	14.1%	12.6%	12.6%	12.6%
3	0	3	0.3898	0.0472	0.0577	0.0752	0.0951	0.1146	9
				16.0%	14.1%	11.7%	12.8%	12.6%
4	0	5	0.3793	0.0460	0.0564	0.0729	0.0902	0.1138	9
				14.3%	13.0%	11.8%	11.6%	12.3%
5	10	0	0.3773	0.0462	0.0577	0.0736	0.0916	0.1082	9
				15.2%	13.9%	11.9%	12.1%	11.8%
6	10	1	0.3766	0.0445	0.0563	0.0723	0.0916	0.1119	9
				14.3%	13.5%	11.8%	12.1%	12.5%
7	10	3	0.3758	0.0441	0.0573	0.0730	0.0926	0.1089	9
				14.1%	13.4%	11.7%	12.2%	11.9%
8	10	5	0.3738	0.0441	0.0569	0.0729	0.0907	0.1093	10
				13.9%	13.3%	12.1%	11.8%	11.8%
9	20	0	0.3771	0.0458	0.0568	0.0732	0.0918	0.1094	10
				14.4%	13.5%	11.9%	12.1%	11.8%
10	20	1	0.3788	0.0454	0.0560	0.0752	0.0917	0.1105	10
				14.2%	13.4%	12.6%	12.3%	12.1%
11	20	3	0.3721	0.0447	0.0568	0.0737	0.0894	0.1075	10
				14.2%	13.5%	12.1%	11.7%	11.7%
12	20	5	0.3687	0.0473	0.0540	0.0705	0.0912	0.1057	12
				14.5%	12.7%	11.5%	12.3%	11.5%
13	—	—	0.4995	0.0740	0.0827	0.0956	0.1137	0.1336	6
				26.9%	22.7%	17.4%	15.9%	15.2%

Note: all the units of error are Hz; the unit of speed is milliseconds per data