Research Article
Neural Virtual Sensors for Adaptive Magnetic Localization of Autonomous Dataloggers
Table 6
Results for all experiments employing D2D and ML computation for path 2 in Figure
11.
| Data set | ISEL1 | BREW | HMI | Network type | RBF | SVR | RBF | SVR | RBF | SVR |
| | 139 | 150 | 333 | 492 | 162 | 216 |
| Results for test and training data set in [cm] | Loc. error | 0.90 | 0.84 | 3.52 | 3.51 | 2.86 | 2.21 | % | 0.32 | 0.3 | 0.95 | 0.94 | 4.69 | 3.62 | Loc. error | 0.80 | 0.65 | 4.13 | 4.47 | 3.46 | 3.30 | % | 0.29 | 0.23 | 1.11 | 1.2 | 5.67 | 5.41 | Max. loc. error | 5.04 | 3.13 | 37.04 | 35.62 | 21.41 | 21.41 | % | 1.82 | 1.13 | 9.96 | 9.58 | 35.1 | 35.1 |
| Results for test with test data set in [cm] | Loc. error | 1.05 | 0.91 | 5.60 | 6.24 | 4.84 | 4.36 | % | 0.38 | 0.33 | 1.51 | 1.68 | 7.93 | 7.15 | Loc. error | 0.78 | 0.64 | 5.01 | 4.99 | 4.26 | 3.82 | % | 0.28 | 0.23 | 1.35 | 1.34 | 6.98 | 6.26 | Max. loc. error | 5.04 | 3.13 | 37.04 | 35.62 | 21.41 | 21.41 | % | 1.82 | 1.13 | 9.96 | 9.58 | 35.1 | 35.1 |
|
|