Discovering Invariant Neighborhood Patterns for Heterophilic Graphs

arXiv:2403.10572v1 Announce Type: new
Abstract: This paper studies the problem of distribution shifts on non-homophilous graphs Mosting existing graph neural network methods rely on the homophilous assumption that nodes from the same class are more likely to be linked. However, such assumptions of homophily do not always hold in real-world graphs, which leads to more complex distribution shifts unaccounted for in previous methods. The distribution shifts of neighborhood patterns are much more diverse on non-homophilous graphs. We propose a novel …

abstract arxiv assumptions class cs.lg cs.si distribution graph graph neural network graphs however leads network neural network nodes paper patterns studies type world