Modeling Tree-like Heterophily on Symmetric Matrix Manifolds

Yang Wu,Liang Hu,Juncheng Hu

DOI: https://doi.org/10.3390/e26050377

IF: 2.738

2024-04-30

Entropy

Abstract:Tree-like structures, characterized by hierarchical relationships and power-law distributions, are prevalent in a multitude of real-world networks, ranging from social networks to citation networks and protein–protein interaction networks. Recently, there has been significant interest in utilizing hyperbolic space to model these structures, owing to its capability to represent them with diminished distortions compared to flat Euclidean space. However, real-world networks often display a blend of flat, tree-like, and circular substructures, resulting in heterophily. To address this diversity of substructures, this study aims to investigate the reconstruction of graph neural networks on the symmetric manifold, which offers a comprehensive geometric space for more effective modeling of tree-like heterophily. To achieve this objective, we propose a graph convolutional neural network operating on the symmetric positive-definite matrix manifold, leveraging Riemannian metrics to facilitate the scheme of information propagation. Extensive experiments conducted on semi-supervised node classification tasks validate the superiority of the proposed approach, demonstrating that it outperforms comparative models based on Euclidean and hyperbolic geometries.

physics, multidisciplinary

What problem does this paper attempt to address?

This paper focuses on how to better model and understand the complex structure in networks that exhibit tree-like heterogeneity. Tree-like structures are common in networks such as social networks, citation networks, and protein-protein interaction networks. Although traditional Euclidean space and hyperbolic space have their advantages, they cannot capture the diversity of these structures perfectly.

To address this challenge, the paper proposes a new framework called Graph Convolutional Neural Network (GCN), which operates on symmetric matrix manifolds to provide a comprehensive geometric space for more effective modeling of tree-like heterogeneity. Specifically, they utilize projection techniques to generalize information propagation components (such as feature transformations, neighbor aggregation, and non-linear activation) to various Riemannian metrics, such as log-Euclidean metric (LEM) and log-Cholesky metric (LCM). This approach allows the representation of different substructures in continuous space, reducing distortion and improving expressive power.

Experimental results show that the proposed Riemannian GCN (RGCN) outperforms comparison models based on Euclidean and hyperbolic geometries in the semi-supervised node classification task. This suggests that RGCN can more accurately capture distances and structures in networks, especially in heterogeneous networks that contain smooth, tree-like, and cyclic substructures.

In summary, the paper addresses the problem of developing a more effective geometric representation method for networks with complex heterogeneity to improve the performance of node classification tasks. By introducing RGCN, the researchers provide a framework that operates on symmetric positive definite matrix manifolds, better adapting to and modeling the complexity of real-world networks.