Divergent Paths: Separating Homophilic and Heterophilic Learning for Enhanced Graph-level Representations

2025-04-06

AI Summary

Graph Neural Networks (GNNs) suffer from limited representational capacity on heterogeneous graphs for graph-level tasks. Method: This paper introduces, for the first time, a homophily/heterophily decoupling paradigm into graph-level representation learning. We propose a dual-path architecture: IntraNet performs class-aware intra-class convolution to model homophilous structures, while InterNet employs high-frequency-enhanced higher-order convolution to capture heterophilous inter-class relationships. A class-guided graph readout function and a gating-based fusion mechanism enable structure-aware feature aggregation. Crucially, class-partitioned graphs are constructed during preprocessing to jointly preserve semantic consistency and structural discriminability. Contribution/Results: Our method achieves significant improvements over baselines—including GCN, GIN, and GraphSAGE—on multiple graph classification benchmarks, with up to 12.7% accuracy gain on strongly heterogeneous datasets, empirically validating the efficacy of synergistic homophily-heterophily modeling for graph-level tasks.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Computer Vision: Representation Learning for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Graph Convolutional Networks (GCNs) are predominantly tailored for graphs displaying homophily, where similar nodes connect, but often fail on heterophilic graphs. The strategy of adopting distinct approaches to learn from homophilic and heterophilic components in node-level tasks has been widely discussed and proven effective both theoretically and experimentally. However, in graph-level tasks, research on this topic remains notably scarce. Addressing this gap, our research conducts an analysis on graphs with nodes' category ID available, distinguishing intra-category and inter-category components as embodiment of homophily and heterophily, respectively. We find while GCNs excel at extracting information within categories, they frequently capture noise from inter-category components. Consequently, it is crucial to employ distinct learning strategies for intra- and inter-category elements. To alleviate this problem, we separately learn the intra- and inter-category parts by a combination of an intra-category convolution (IntraNet) and an inter-category high-pass graph convolution (InterNet). Our IntraNet is supported by sophisticated graph preprocessing steps and a novel category-based graph readout function. For the InterNet, we utilize a high-pass filter to amplify the node disparities, enhancing the recognition of details in the high-frequency components. The proposed approach, DivGNN, combines the IntraNet and InterNet with a gated mechanism and substantially improves classification performance on graph-level tasks, surpassing traditional GNN baselines in effectiveness.

Problem

• Addresses GCNs' failure in heterophilic graph-level tasks
• Proposes separate learning for homophilic and heterophilic components
• Enhances graph classification via IntraNet and InterNet fusion

Innovation

• Separates homophilic and heterophilic graph learning
• Uses IntraNet for intra-category convolution
• Employs InterNet with high-pass filtering