Protein Property Prediction on Multi-task Experiment (150 train points, test)
0.852Spearman CorrelationCLOCK-GP+CNN-ZS
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| CLOCK-GP+CNN-ZSTraining points=150, Kernel=non-linear2026.06 | 0.852 | 0.338 | 0.88 | 0.469 | |
| CNN-EMB-OHTraining points=150, Features=CNN neural features + one-hot2026.06 | 0.838 | 0.374 | 0.863 | 0.502 | |
| CLOCK-GP-LIN+CNN-ZSTraining points=150, Kernel=linear2026.06 | 0.837 | 0.36 | 0.862 | 0.502 | |
| CNN-ZS-OHTraining points=150, Features=CNN zero-shot + one-hot2026.06 | 0.826 | 0.388 | 0.853 | 0.52 | |
| CLOCK-GPTraining points=150, Kernel=non-linear2026.06 | 0.826 | 0.362 | 0.857 | 0.508 | |
| CLOCK-GP-LINTraining points=150, Kernel=linear2026.06 | 0.743 | 0.424 | 0.771 | 0.641 | |
| CNN-ZSTraining points=150, Features=CNN zero-shot predictions2026.06 | 0.724 | 1.679 | 0.754 | 1.811 | |
| Global CLOCK-GPTraining points=150, Kernel=non-linear2026.06 | 0.72 | 0.462 | 0.751 | 0.657 | |
| Ridge-EMB-OHTraining points=150, Features=Mean-pooled Chroma features + one-hot2026.06 | 0.717 | 0.499 | 0.74 | 0.671 | |
| Global CLOCK-GP-LINTraining points=150, Kernel=linear2026.06 | 0.698 | 0.48 | 0.722 | 0.698 | |
| LOCK-GPTraining points=150, Kernel=non-linear2026.06 | 0.689 | 0.505 | 0.712 | 0.704 | |
| LOCK-GP-LINTraining points=150, Kernel=linear2026.06 | 0.67 | 0.513 | 0.695 | 0.72 | |
| Ridge-EMBTraining points=150, Features=Mean-pooled Chroma features2026.06 | 0.643 | 0.567 | 0.672 | 0.74 | |
| Ridge-OHTraining points=150, Features=One-hot sequences2026.06 | 0.621 | 0.567 | 0.629 | 0.774 |