Information bottleneck method
-
CORE: Data Augmentation for Link Prediction via Information Bottleneck
CORE applies Information Bottleneck principles to augment graph data for link prediction, simultaneously recovering missing edges and reducing noise to enhance model robustness.

