Protein Property Prediction on 50 training points (test)
0.8Spearman CorrCLOCK-GP+CNN-ZS
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CLOCK-GP+CNN-ZSTraining points=50, Kernel=non-linear2026.06 | 0.8 | 0.407 | 0.832 | 0.553 | |
| CNN-EMB-OHTraining points=50, Features=CNN neural features + one-hot2026.06 | 0.796 | 0.433 | 0.824 | 0.571 | |
| CNN-ZS-OHTraining points=50, Features=CNN zero-shot + one-hot2026.06 | 0.791 | 0.439 | 0.819 | 0.578 | |
| CLOCK-GP-LIN+CNN-ZSTraining points=50, Kernel=linear2026.06 | 0.778 | 0.435 | 0.805 | 0.595 | |
| CLOCK-GPTraining points=50, Kernel=non-linear2026.06 | 0.755 | 0.446 | 0.795 | 0.61 | |
| CNN-ZSTraining points=50, Features=CNN zero-shot predictions2026.06 | 0.724 | 1.688 | 0.754 | 1.82 | |
| Ridge-EMB-OHTraining points=50, Features=Mean-pooled Chroma features + one-hot2026.06 | 0.576 | 0.611 | 0.604 | 0.8 | |
| Global CLOCK-GPTraining points=50, Kernel=non-linear2026.06 | 0.529 | 0.611 | 0.571 | 0.829 | |
| Ridge-EMBTraining points=50, Features=Mean-pooled Chroma features2026.06 | 0.52 | 0.657 | 0.55 | 0.847 | |
| CLOCK-GP-LINTraining points=50, Kernel=linear2026.06 | 0.509 | 0.581 | 0.582 | 0.853 | |
| LOCK-GPTraining points=50, Kernel=non-linear2026.06 | 0.498 | 0.643 | 0.53 | 0.852 | |
| Global CLOCK-GP-LINTraining points=50, Kernel=linear2026.06 | 0.484 | 0.618 | 0.534 | 0.862 | |
| LOCK-GP-LINTraining points=50, Kernel=linear2026.06 | 0.478 | 0.641 | 0.514 | 0.86 | |
| Ridge-OHTraining points=50, Features=One-hot sequences2026.06 | 0.457 | 0.673 | 0.474 | 0.884 |