Graphs Generalization under Distribution Shifts

Qin Tian, Wenjun Wang, Chen Zhao +3

Traditional machine learning methods heavily rely on the independent and identically distribution assumption, which imposes limitations when the test distribution deviates from the…