Implicit Graph Neural Diffusion Networks: Convergence, Generalization, and Over-Smoothing

arXiv:2308.03306v2 Announce Type: replace
Abstract: Implicit Graph Neural Networks (GNNs) have achieved significant success in addressing graph learning problems recently. However, poorly designed implicit GNN layers may have limited adaptability to learn graph metrics, experience over-smoothing issues, or exhibit suboptimal convergence and generalization properties, potentially hindering their practical performance. To tackle these issues, we introduce a geometric framework for designing implicit graph diffusion layers based on a parameterized graph Laplacian operator. Our framework allows learning the metrics of vertex and …

abstract adaptability arxiv convergence cs.lg diffusion experience gnn gnns graph graph learning graph neural networks learn metrics networks neural networks performance practical success type