Table 12 Performance of RBFNN models 2–6.
RBFNN model type | Width | Training data accuracy (%) | Test data accuracy (%) | Average MSE |
|---|---|---|---|---|
RBFNN model 2 for 25 RBF units | 0.15 | 97.6879 | 96.7391 | 0.0162 |
0.20 | 97.3988 | 95.6522 | 0.0163 | |
0.25 | 97.6879 | 95.6522 | 0.0164 | |
RBFNN model 3 for 25 RBF units | 0.15 | 98.9130 | 98.5549 | 0.0108 |
0.20 | 98.5549 | 97.8261 | 0.0161 | |
0.25 | 97.8261 | 96.5318 | 0.0420 | |
RBFNN model 4 for 30 RBF units | 0.15 | 98.9130 | 98.8439 | 0.0144 |
0.20 | 98.8439 | 96.7391 | 0.0222 | |
0.25 | 96.7391 | 94.5087 | 0.0419 | |
RBFNN model 5 for 35 RBF units | 0.15 | 98.8439 | 97.8261 | 0.0148 |
0.20 | 98.2659 | 94.5652 | 0.0161 | |
0.25 | 97.6879 | 95.6522 | 0.0186 | |
RBFNN model 6 for 35 RBF units | 0.15 | 100.0000 | 99.1329 | 0.0131 |
0.20 | 100.0000 | 98.8439 | 0.0168 | |
0.25 | 99.1329 | 98.9130 | 0.0173 |
