DrugOOD
A graph-level distribution shift benchmark dataset in the domain of drug discovery. It comprises multiple sub-datasets covering distinct shift types such as scaffold shift and molecule size shift, and is used to evaluate out-of-distribution generalization methods.
Timeline (last 90 days)
On a controlled synthetic dataset and the DrugOOD dataset, SCARCE achieved state-of-the-art OOD generalization performance among label-free methods.
DrugOOD is a graph-level distribution shift benchmark dataset in the drug discovery domain, containing multiple sub-datasets covering different shift types (such as scaffold shift, molecule size shift, etc.).
In DrugOOD, each molecule is represented as a graph, with nodes as atoms and edges as chemical bonds, and the task is to predict molecular properties (such as toxicity).
On DrugOOD, SCARCE does not use class labels and trains using only the environment partition, achieving the best performance compared with label-free baselines.
All Facts (4)
On a controlled synthetic dataset and the DrugOOD dataset, SCARCE achieved state-of-the-art OOD generalization performance among label-free methods.
50%UnverifiedDrugOOD is a graph-level distribution shift benchmark dataset in the drug discovery domain, containing multiple sub-datasets covering different shift types (such as scaffold shift, molecule size shift, etc.).
50%UnverifiedIn DrugOOD, each molecule is represented as a graph, with nodes as atoms and edges as chemical bonds, and the task is to predict molecular properties (such as toxicity).
50%UnverifiedOn DrugOOD, SCARCE does not use class labels and trains using only the environment partition, achieving the best performance compared with label-free baselines.
50%