Research Article
Fusion of Deep Features from 2D-DOST of fNIRS Signals for Subject-Independent Classification of Motor Execution Tasks
Table 6
The accuracy of MobileNet in different scenarios.
| | Scenario | Number of channels | Fusion scheme | | Early | Late | Joint |
| | (RHT, LHT, FT) | 4 | 79.01 (80.26 ± 3.38) | 75.46 (77.29 ± 4.15) | 81.08 (83.41 ± 3.84) | | 8 | 82.23 (83.75 ± 2.62) | 81.90 (83.03 ± 3.44) | 87.74 (89.53 ± 3.09) | | 16 | 88.05 (88.93 ± 2.14) | 86.36 (88.35 ± 2.19) | 91.12 (93.02 ± 1.95) |
| | (RHT, LHT) | 4 | 85.49 (86.93 ± 2.72) | 80.92 (83.49 ± 3.13) | 86.02 (88.73 ± 2.72) | | 8 | 88.91 (89.93 ± 2.36) | 87.26 (89.07 ± 2.56) | 92.82 (94.09 ± 2.14) | | 16 | 94.01 (95.03 ± 1.95) | 91.91 (93.41 ± 2.28) | 97.15 (98.73 ± 1.41) |
| | (RHT, FT) | 4 | 83.29 (84.83 ± 3.08) | 80.26 (81.99 ± 3.11) | 85.93 (87.25 ± 3.05) | | 8 | 87.54 (88.28 ± 2.34) | 86.05 (87.26 ± 2.77) | 92.06 (94.01 ± 2.26) | | 16 | 92.71 (93.82 ± 1.87) | 90.98 (92.18 ± 2.49) | 95.65 (96.67 ± 1.98) |
| | (LHT, FT) | 4 | 82.17 (83.25 ± 3.12) | 79.15 (81.01 ± 3.24) | 85.19 (87.25 ± 3.09) | | 8 | 87.90 (89.16 ± 2.68) | 84.09 (85.17 ± 3.02) | 91.01 (92.13 ± 2.33) | | 16 | 90.88 (92.09 ± 2.16) | 89.91 (91.63 ± 2.47) | 93.87 (95.66 ± 2.12) |
|
|
The accuracy of the modified MobileNet is given in parentheses.
|
