Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Heterogeneous Graph Contrastive Learning with Spectral Augmentation

Jing Zhang, Xiaoqian Jiang, Yingjie Xie, Cangqi Zhou

TL;DR

This work addresses the gap in heterogeneous graph representation learning by incorporating spectral information through a novel spectral augmentation framework. The proposed SHCL model learns adaptive, meta-path–specific topology augmentations that maximize spectral distance, enabling the encoder to capture spectral invariance via a dual-aggregation scheme. Empirical results on DBLP, ACM, and Freebase show SHCL outperforms strong baselines and its ablations confirm the value of learned spectral augmentation. The approach highlights the importance of spectral perspective in heterogeneous graphs and offers a practical path toward more robust representations in real-world networks.$

Abstract

Heterogeneous graphs can well describe the complex entity relationships in the real world. For example, online shopping networks contain multiple physical types of consumers and products, as well as multiple relationship types such as purchasing and favoriting. More and more scholars pay attention to this research because heterogeneous graph representation learning shows strong application potential in real-world scenarios. However, the existing heterogeneous graph models use data augmentation techniques to enhance the use of graph structure information, which only captures the graph structure information from the spatial topology, ignoring the information displayed in the spectrum dimension of the graph structure. To address the issue that heterogeneous graph representation learning methods fail to model spectral information, this paper introduces a spectral-enhanced graph contrastive learning model (SHCL) and proposes a spectral augmentation algorithm for the first time in heterogeneous graph neural networks. The proposed model learns an adaptive topology augmentation scheme through the heterogeneous graph itself, disrupting the structural information of the heterogeneous graph in the spectrum dimension, and ultimately improving the learning effect of the model. Experimental results on multiple real-world datasets demonstrate substantial advantages of the proposed model.

Heterogeneous Graph Contrastive Learning with Spectral Augmentation

TL;DR

This work addresses the gap in heterogeneous graph representation learning by incorporating spectral information through a novel spectral augmentation framework. The proposed SHCL model learns adaptive, meta-path–specific topology augmentations that maximize spectral distance, enabling the encoder to capture spectral invariance via a dual-aggregation scheme. Empirical results on DBLP, ACM, and Freebase show SHCL outperforms strong baselines and its ablations confirm the value of learned spectral augmentation. The approach highlights the importance of spectral perspective in heterogeneous graphs and offers a practical path toward more robust representations in real-world networks.$

Abstract

Heterogeneous graphs can well describe the complex entity relationships in the real world. For example, online shopping networks contain multiple physical types of consumers and products, as well as multiple relationship types such as purchasing and favoriting. More and more scholars pay attention to this research because heterogeneous graph representation learning shows strong application potential in real-world scenarios. However, the existing heterogeneous graph models use data augmentation techniques to enhance the use of graph structure information, which only captures the graph structure information from the spatial topology, ignoring the information displayed in the spectrum dimension of the graph structure. To address the issue that heterogeneous graph representation learning methods fail to model spectral information, this paper introduces a spectral-enhanced graph contrastive learning model (SHCL) and proposes a spectral augmentation algorithm for the first time in heterogeneous graph neural networks. The proposed model learns an adaptive topology augmentation scheme through the heterogeneous graph itself, disrupting the structural information of the heterogeneous graph in the spectrum dimension, and ultimately improving the learning effect of the model. Experimental results on multiple real-world datasets demonstrate substantial advantages of the proposed model.
Paper Structure (16 sections, 22 equations, 3 figures, 4 tables)

This paper contains 16 sections, 22 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Graph Contrastive Learning
  4. Heterogeneous graphs represent learning
  5. Preliminary
  6. The Proposed Method
  7. Overview of the Model
  8. Parametric Topology Augmentation Scheme in Heterogeneous Graphs
  9. Spectral Augmentation Algorithms for Heterogeneous Graphs
  10. Experiments
  11. Datasets
  12. Experimental Setup
  13. Experimental Results and Discussion
  14. Results of Comparison Experiment
  15. Ablation experiment
  16. ...and 1 more sections

Figures (3)

  • Figure 1: Examples and related concepts of heterogeneous graphs.
  • Figure 2: Illustration of the proposed SHCL.
  • Figure 3: Comparison of different SHCL model variants.

Theorems & Definitions (5)

  • Definition 1
  • Definition 2
  • Definition 3
  • Definition 4
  • Definition 5