Synthetic Data Reveals Generalization Gaps in Correlated Multiple Instance Learning

Open in new window