Table 1 5-Fold cross-validation performances of methods on Martin dataset
Method | Accuracy (%) | Precision (%) | Recall (%) | Specificity (%) | F1-Score (%) | MCC (%) |
|---|---|---|---|---|---|---|
PIPR (2019) | 80.84 ± 0.44 | 81.44 ± 0.69 | 81.55 ± 0.85 | 80.32 ± 0.67 | 81.43 ± 0.45 | 61.69 ± 0.89 |
FSNN-LGBM (2021) | 96.49 ± 0.13 | 96.03 ± 0.26 | 97.23 ± 0.04 | 95.69 ± 0.29 | 96.62 ± 0.12 | 92.98 ± 0.25 |
GcForestPPI (2021) | 89.26 | 88.95 | 89.71 | NA | 88.33 | 78.57 |
MARPPI (2023) | 91.80 ± 1.16 | 90.69 ± 2.68 | 94.51 ± 1.13 | 91.22 ± 1.25 | NA | 83.74 ± 2.32 |
HNSPPI (2023) | 93.21 ± 0.35 | 88.47 ± 0.53 | 99.39 ± 0.21 | NA | 93.59 ± 0.32 | 93.21 ± 0.35 |
EresCNN (2023) | 87.89 | 87.84 | 87.96 | NA | 87.90 | 75.81 |
Our xCAPT5 | 97.27 ± 0.12 | 97.30 ± 0.24 | 97.07 ± 0.20 | 97.44 ± 0.11 | 97.18 ± 0.25 | 94.82 ± 0.20 |