Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Beyond Redundancy: Information-aware Unsupervised Multiplex Graph Structure Learning

Zhixiang Shen, Shuo Wang, Zhao Kang

TL;DR

The proposed Information-aware Unsupervised Multiplex Graph Fusion framework (InfoMGF) uses graph structure refinement to eliminate irrelevant noise and simultaneously maximizes view-shared and view-unique task-relevant information, thereby tackling the frontier of non-redundant multiplex graph.

Abstract

Unsupervised Multiplex Graph Learning (UMGL) aims to learn node representations on various edge types without manual labeling. However, existing research overlooks a key factor: the reliability of the graph structure. Real-world data often exhibit a complex nature and contain abundant task-irrelevant noise, severely compromising UMGL's performance. Moreover, existing methods primarily rely on contrastive learning to maximize mutual information across different graphs, limiting them to multiplex graph redundant scenarios and failing to capture view-unique task-relevant information. In this paper, we focus on a more realistic and challenging task: to unsupervisedly learn a fused graph from multiple graphs that preserve sufficient task-relevant information while removing task-irrelevant noise. Specifically, our proposed Information-aware Unsupervised Multiplex Graph Fusion framework (InfoMGF) uses graph structure refinement to eliminate irrelevant noise and simultaneously maximizes view-shared and view-unique task-relevant information, thereby tackling the frontier of non-redundant multiplex graph. Theoretical analyses further guarantee the effectiveness of InfoMGF. Comprehensive experiments against various baselines on different downstream tasks demonstrate its superior performance and robustness. Surprisingly, our unsupervised method even beats the sophisticated supervised approaches. The source code and datasets are available at https://github.com/zxlearningdeep/InfoMGF.

Beyond Redundancy: Information-aware Unsupervised Multiplex Graph Structure Learning

TL;DR

The proposed Information-aware Unsupervised Multiplex Graph Fusion framework (InfoMGF) uses graph structure refinement to eliminate irrelevant noise and simultaneously maximizes view-shared and view-unique task-relevant information, thereby tackling the frontier of non-redundant multiplex graph.

Abstract

Unsupervised Multiplex Graph Learning (UMGL) aims to learn node representations on various edge types without manual labeling. However, existing research overlooks a key factor: the reliability of the graph structure. Real-world data often exhibit a complex nature and contain abundant task-irrelevant noise, severely compromising UMGL's performance. Moreover, existing methods primarily rely on contrastive learning to maximize mutual information across different graphs, limiting them to multiplex graph redundant scenarios and failing to capture view-unique task-relevant information. In this paper, we focus on a more realistic and challenging task: to unsupervisedly learn a fused graph from multiple graphs that preserve sufficient task-relevant information while removing task-irrelevant noise. Specifically, our proposed Information-aware Unsupervised Multiplex Graph Fusion framework (InfoMGF) uses graph structure refinement to eliminate irrelevant noise and simultaneously maximizes view-shared and view-unique task-relevant information, thereby tackling the frontier of non-redundant multiplex graph. Theoretical analyses further guarantee the effectiveness of InfoMGF. Comprehensive experiments against various baselines on different downstream tasks demonstrate its superior performance and robustness. Surprisingly, our unsupervised method even beats the sophisticated supervised approaches. The source code and datasets are available at https://github.com/zxlearningdeep/InfoMGF.
Paper Structure (34 sections, 5 theorems, 43 equations, 8 figures, 6 tables, 2 algorithms)

This paper contains 34 sections, 5 theorems, 43 equations, 8 figures, 6 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Methodology
  4. Graph Structure Refinement
  5. Maximizing Shared Task-Relevant Information
  6. Maximizing Unique Task-Relevant Information
  7. Multiplex Graph Fusion
  8. Experiments
  9. Experimental Setups
  10. Effectiveness Analysis (RQ1)
  11. Robustness Analysis (RQ2)
  12. Ablation Study (RQ3)
  13. Conclusion and Limitation
  14. Acknowledgments and Disclosure of Funding
  15. Notations
  16. ...and 19 more sections

Key Result

Theorem 1

If $G_{i}^{\prime}$ is the optimal augmented graph of $G_{i}$, then $I(G_{i}^s;G_{i}^{\prime})=I(G_i^s;Y)$ holds.

Figures (8)

  • Figure 1: (a) and (b) illustrate that in a non-redundant multiplex graph, view-specific task-relevant edges exist in certain graphs. The color of nodes represents class, edges between nodes of the same class are considered relevant edges, and "unique" indicates that the edge exists only in one graph. (c) The unique relevant edge ratio = (the number of unique relevant edges) / (the total number of relevant edges in this graph). Each graph contains a significant amount of unique task-relevant information.
  • Figure 2: The overall framework of the proposed InfoMGF. Specifically, InfoMGF first generates refined graphs and the fused graph through the graph learner. Subsequently, it maximizes shared and unique task-relevant information within the multiplex graph and facilitates graph fusion. The learned fused graph and node representations are used for various downstream tasks.
  • Figure 3: Heatmaps of the subgraph adjacency matrices of the original and learned graphs on ACM.
  • Figure 4: Robustness analysis on ACM.
  • Figure 5: Additional experiments on sensitivity and robustness analysis.
  • ...and 3 more figures

Theorems & Definitions (11)

  • Definition 1
  • Definition 2
  • Theorem 1
  • Theorem 2
  • Theorem 3
  • Lemma 1
  • Proposition 1
  • proof : Proof of Proposition \ref{['Lb']}:
  • proof : Proof of Theorem \ref{['G_prime_Y']}
  • proof : Proof of Theorem \ref{['G_prime_G']}
  • ...and 1 more