推荐指数:#paper/⭐
发表于:ICML24
这不和from Trainable Negative Depth to Edge Heterophily in Graph一样吗?
24年也是同样思想的文章,SLOG An Inductive Spectral Graph Neural Network Beyond Polynomial Filter.这篇虽然也用了负度,但是人家的切入点更好(大图负度).
个人感觉创新性一般般,但是理论很足.
Mitigating Oversmoothing Through Reverse ProcessofGNNs for Heterophilic Graphs
于 2024-08-14 18:36:59 首次发布