Table 4.
Performance comparisons with other approaches.
| Classification methods | DEAP dataset | SEED dataset (%) | |
|---|---|---|---|
| Valence (%) | Arousal (%) | ||
| CNN + statistical methods (Tripathi et al., 2017) | 81.4 | 73.4 | / |
| Gaussian Bayes (Koelstra et al., 2012) | 57.6 | 62.0 | / |
| Deep SAE + RSP (Zhang et al., 2017) | 73.1 | 80.8 | / |
| BDGLS + DE (Wang et al., 2018) | / | / | 93.7 |
| DGCNN + DE (Zhang et al., 2016) | / | / | 90.4 |
| GP + LVM (García et al., 2016) | 88.3 | 90.6 | / |
| BLSTM + DE (Wang et al., 2019) | / | / | 94.96 |
| Physe Space Dynamics (Soroush et al., 2019) | 84.6 | 87.4 | / |
| SDEL + PCA (Ullah et al., 2019) | 82.8 | 74.5 | / |
| This work [PCC] | 89.49 | 92.86 | 96.77 |