Table 4.
Online classification results of walking phase classifiers with MLP networks.
| Number of MLP Hidden Nodes | CSR (Error Rate) | Maximum Continuous Error Width (Frames) | Mean and Std. Continuous Error Width (Frames) | Number of Unstable Regions |
|---|---|---|---|---|
| 5 | 91.46% (8.54%) | 37 | 8.89 ± 8.18 | 3 |
| 10 | 90.38% (9.62%) | 37 | 10.02 ± 9.22 | 12 |
| 15 | 91.08% (8.92%) | 37 | 9.29 ± 8.73 | 4 |
| 20 | 90.85% (9.15%) | 37 | 9.52 ± 8.91 | 7 |
| 25 | 88.35% (11.65%) | 37 | 12.13 ± 11.20 | 15 |
| 30 | 91.51% (8.49%) | 37 | 8.84 ± 8.62 | 1 |
| 35 | 90.51% (9.49%) | 37 | 9.87 ± 9.06 | 9 |
| 40 | 91.14% (8.86%) | 37 | 9.22 ± 8.68 | 6 |
| 45 | 91.44% (8.56%) | 37 | 8.90 ± 8.22 | 3 |
| 50 | 90.82% (9.18%) | 37 | 9.56 ± 8.67 | 4 |