Table 4 Comparison with the state-of-the-art
Ref | Method | Accuracy (%) | Precision (%) | Recall (%) |
|---|---|---|---|---|
Cai et al. [4],2023 | Self-training | 87.1 | 86.5 | 85.8 |
Dzien et al. [5], 2024 | Self-training | 84.8 | 84.2 | 83.5 |
Xie et al. [24], 2023 | Self-training | 79.5 | 79 | 78.3 |
Yang et al. [7], 2024 | co-training | 87 | 88 | 86 |
Tang et al. [6], 2024 | co-training | 81.2 | 82 | 80.5 |
Bai et al. [35], 2024 | co-training | 82 | 83 | 81 |
Sun et al. [11], 2023 | Graph-based training | 80.87 | 81.5 | 80 |
Miller et al. [10], 2024 | Graph-based training | 89 | 89.5 | 88 |
Miller et al. [9], 2009 | Graph-based training | 90.2 | 90.5 | 89.5 |
Li et al. [12], 2024 | pseudo label | 85 | 85 | 84 |
Li et al. [16], 2023 | Ensemble Learning | 86.2 | 87 | 85.5 |
You et al. [15], 2023 | Rethinking Semi-Supervised learning | 78.3 | 78 | 77 |
Salimans et al. [34], 2016 | Simple GAN | 89.7 | 78.1 | 81.5 |
Khosravan et al. [22], 2018 | Multi-task learning | 90.2 | 84 | 89 |
DDDG-GAN (proposed) | 92.56 | 90.12 | 95.87 | |