<ul><li>Injecting rule-based models like Random Forests into differentiable neural network frameworks is a challenge in machine learning.</li><li>A novel strategy is proposed to jointly train a Graph Transformer neural network on both experimental and synthetic molecular property targets.</li><li>The synthetic tasks, derived from XGBoost models trained on Osmordred molecular descriptors, improve performance significantly in molecular property prediction tasks.</li><li>The results indicate that synthetic task augmentation enhances neural model performance without requiring feature injection or pretraining.</li></ul>

All You Need Is Synthetic Task Augmentation

Discover more