Gradient Neural Dynamics Based on Modified Error Function
Table 2: Experiment results for non-squared matrices
| m |
n |
Rank A |
Rank B |
Time |
GNN |
GGNN |
γ |
GNN CPU Time |
GGNN CPU Time |
| 25 |
20 |
20 |
20 |
10-6 |
26.45 |
12,85 |
100 |
9.25 |
3.42 |
| 25 |
20 |
2 |
2 |
10-6 |
105.80 |
59.11 |
100 |
1.17 |
1.55 |
| 25 |
20 |
20 |
2 |
10-6 |
53.10 |
48.76 |
100 |
1.37 |
2.08 |
| 25 |
20 |
10 |
2 |
10-6 |
72.35 |
43.07 |
100 |
1.51 |
1.00 |
| 25 |
20 |
10 |
10 |
10-6 |
33.65 |
28.49 |
100 |
1.58 |
2.16 |
| 25 |
20 |
20 |
10 |
10-6 |
27.58 |
17.22 |
100 |
1.14 |
2.07 |
| 10 |
8 |
8 |
8 |
10-3 |
6.19 |
4.26 |
1 |
0.55 |
1.12 |
| 10 |
8 |
2 |
2 |
10-3 |
11.87 |
10.22 |
1 |
0.52 |
0.91 |
| 10 |
8 |
8 |
2 |
10-3 |
6.63 |
6.18 |
1 |
0.70 |
0.90 |
| 10 |
8 |
5 |
2 |
10-3 |
7.45 |
7.39 |
1 |
1.44 |
0.91 |
| 10 |
8 |
5 |
5 |
10-3 |
7.37 |
5.88 |
1 |
0.59 |
0.86 |
| 10 |
8 |
8 |
5 |
10-3 |
6.33 |
5.38 |
1 |
0.60 |
0.89 |
< Back to article
