Validation-Guided Active Learning Improves Property Predictions Under Hidden Distribution Shift
Researchers introduced CADA-OOD, an active learning framework using validation error relevance to guide sample selection under hidden distribution shifts in ...

