Table 2.
The decoding performance for each method.
| Method | Decoding performance (Average) | |||
|---|---|---|---|---|
| Accuracy ± std | Kappa | Recall | F1-score | |
| CSP + SVM (Antony et al., 2022) | 56.73% ± 6.34% | 0.087 | 0.163 | 0.261 |
| ANN (Yacine et al., 2022) | 61.43% ± 9.66% | 0.368 | 0.788 | 0.686 |
| CNN (Mehrdad and Salimi, 2023) | 63.43% ± 8.62% | 0.282 | 0.364 | 0.414 |
| 3DCNN (Li and Ruan, 2021) | 75.72% ± 5.82% | 0.506 | 0.646 | 0.704 |
| EEGNet (Lawhern et al., 2018) | 63.39% ± 10.34% | 0.278 | 0.511 | 0.571 |
| DeepConvNet (Chaudhary et al., 2019) | 62.56% ± 6.03% | 0.248 | 0.987 | 0.729 |
| ShallowConNet (Milanes Hermosilla et al., 2021) | 67.83% ± 9.93% | 0.363 | 0.777 | 0.674 |
| P-3DCNN | 86.69% ± 3.35% | 0.751 | 0.826 | 0.864 |