Table 2.
Offline 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 | 97.70% (2.30%) | 10 | 2.20 ± 1.78 | 3 |
| 10 | 97.73% (2.27%) | 8 | 2.17 ± 1.56 | 4 |
| 15 | 97.71% (2.29%) | 8 | 2.19 ± 1.62 | 8 |
| 20 | 97.73% (2.28%) | 10 | 2.17 ± 1.63 | 4 |
| 25 | 97.76% (2.24%) | 9 | 2.15 ± 1.65 | 3 |
| 30 | 97.85% (2.15%) | 10 | 2.05 ± 1.55 | 4 |
| 35 | 97.76% (2.24%) | 9 | 2.15 ± 1.66 | 4 |
| 40 | 97.78% (2.22%) | 9 | 2.12 ± 1.61 | 3 |
| 45 | 97.74% (2.26%) | 8 | 2.17 ± 1.56 | 5 |
| 50 | 97.71% (2.29%) | 8 | 2.18 ± 1.40 | 6 |