A two-stage co-adversarial perturbation to mitigate out-of-distribution generalization of large-scale graph

Yili Wang, Haotian Xue, Xin Wang

Abstract: Highlights•Article explores GNN training challenges, especially in large-scale datasets.•Study analyzes GNN behavior, highlighting a tendency to be trapped in sharp local minima.•Authors introduce co-adversarial perturbation optimization.