Table 2 Simulation results of "Hyperparameter optimization experiment ".
From: Research on improved convolutional wavelet neural network
Number of CSP | BPNN training times | BPNN mean square error | BPNN maximum error | RBFNN training times | RBFNN mean square error | RBFNN maximum error | WNN training times | WNN mean square error | WNN maximum error |
|---|---|---|---|---|---|---|---|---|---|
1 | 20,000 | 0.000686 | 0.078208 | 20,000 | 0.007004 | 0.281643 | 1403 | 0.000052 | 0.019990 |
2 | 20,000 | 0.000441 | 0.052301 | 16,391 | 0.000078 | 0.019912 | 20,000 | 0.000261 | 0.032628 |
3 | 20,000 | 0.000559 | 0.056556 | 20,000 | 0.068413 | 0.762847 | 3740 | 0.000071 | 0.019316 |
4 | 20,000 | 0.000623 | 0.070849 | 7401 | 0.000047 | 0.020000 | 25e85 | 0.000065 | 0.019895 |
5 | 20,000 | 0.000687 | 0.082878 | 20,000 | 0.002712 | 0.136289 | 16,836 | 0.000030 | 0.019999 |
6 | 20,000 | 0.000610 | 0.071138 | 20,000 | 0.013012 | 0.401105 | 20,000 | 0.002192 | 0.109309 |
7 | 20,000 | 0.000572 | 0.068204 | 9546 | 0.000050 | 0.015604 | 1270 | 0.000048 | 0.019945 |
8 | 20,000 | 0.000601 | 0.070306 | 20,000 | 0.014854 | 0.338185 | 20,000 | 0.000183 | 0.039474 |
9 | 20,000 | 0.000617 | 0.070071 | 20,000 | 0.006495 | 0.211676 | 3943 | 0.000043 | 0.019992 |
10 | 20,000 | 0.000707 | 0.083616 | 5185 | 0.000073 | 0.019370 | 20,000 | 0.000562 | 0.063996 |
Mean | No case success | No case success | No case success | 9631 | 0.000062 | 0.018722 | 4963 best | 0.000052 | 0.019856 |
Success rate | 0% | 40% | 60%best | ||||||
