TL;DR本文提出了一种直接在超伪球面上进行图卷积运算的超伪球面图卷积网络(H2H-GCN),该网络采用保流形图卷积和 Einstein 中点算法来保留全局的超伪球面结构。在链接预测、节点分类和图分类任务中均实现了显著的性能提升。
Abstract
hyperbolic graph convolutional networks (GCNs) demonstrate powerful
representation ability to model graphs with hierarchical structure. Existing
hyperbolic GCNs resort to tangent spaces to realize graph convolution on
h