Scaffold Split
A strict generalization evaluation protocol in molecular machine learning that partitions datasets by Murcko scaffold groups, ensuring that training and test molecules have different scaffolds and thus better reflecting a model's generalization to novel chemical space.
Timeline (last 90 days)
The study conducts a systematic evaluation across four GNN architectures, four prediction tasks, and two explainers, and uses scaffold split to ensure the rigor of generalization testing.
Scaffold split groups molecules by their Murcko scaffolds, ensuring that training-set and test-set molecules have different scaffold types, and is a rigorous generalization evaluation protocol in molecular machine learning.
All Facts (2)
The study conducts a systematic evaluation across four GNN architectures, four prediction tasks, and two explainers, and uses scaffold split to ensure the rigor of generalization testing.
50%UnverifiedScaffold split groups molecules by their Murcko scaffolds, ensuring that training-set and test-set molecules have different scaffold types, and is a rigorous generalization evaluation protocol in molecular machine learning.
50%