Table 6.
Classification results of GAN/WGAN frameworks for the sampling rate reconstruction of the same sensitivity signals in the MI dataset.
| Subjects | Original data (a) | Spline data (b) | GAN up-sampling (c) | WGAN up-sampling (d) |
|---|---|---|---|---|
| MI-1 | 60.59 | 57.82 | 61.62 | 62.81 |
| MI-2 | 70.31 | 68.58 | 71.18 | 73.35 |
| MI-3 | 53.47 | 53.82 | 54.93 | 56.81 |
| MI-4 | 59.38 | 56.71 | 61.83 | 60.62 |
| MI-5 | 72.22 | 70.49 | 73.88 | 76.57 |
| MI-6 | 67.36 | 65.73 | 68.72 | 68.81 |
| MI-7 | 56.25 | 55.41 | 58.83 | 57.61 |
| MI-8 | 57.81 | 56.43 | 58.81 | 57.69 |
| MI-9 | 60.42 | 58.67 | 62.73 | 61.85 |
| AVG | 61.98 | 60.41 | 63.61 | 64.01 |
| T-test | – | a vs. b | c vs. a | d vs. a |
| p-value | – | **p <0.01 | **p <0.01 | **p <0.01 |
The bold text is the best performance.