Table 5.
Performance comparison between multi-task [10] and meta-learning neural networks
| Assay ID | Multi-task | Proposed (Meta-learning) |
|---|---|---|
| 1851_1a2 | 0.938 | 0.943 |
| 1851_2c19 | 0.903 | 0.908 |
| 1851_2c9 | 0.907 | 0.908 |
| 1851_2d6 | 0.861 | 0.892 |
| 1851_3a4 | 0.897 | 0.920 |
| 1915 | 0.750 | 0.764 |
| 2358 | 0.751 | 0.807 |
| 463213 | 0.676 | 0.694 |
| 463215 | 0.654 | 0.634 |
| 488912 | 0.816 | 0.700 |
| 488915 | 0.873 | 0.739 |
| 488917 | 0.894 | 0.841 |
| 488918 | 0.842 | 0.801 |
| 492992 | 0.829 | 0.862 |
| 504607 | 0.670 | 0.726 |
| 624504 | 0.889 | 0.904 |
| 651739 | 0.825 | 0.809 |
| 651744 | 0.900 | 0.909 |
| 652065 | 0.792 | 0.832 |
The mean AUC values for both neural networks are shown (bold: top AUC on each dataset)