Subgraph Gaussian Embedding Contrast for Self-Supervised Graph Representation Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xie, Shifeng, Einizade, Aref, Giraldo, Jhony H.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912426612490240
author Xie, Shifeng
Einizade, Aref
Giraldo, Jhony H.
author_facet Xie, Shifeng
Einizade, Aref
Giraldo, Jhony H.
contents Graph Representation Learning (GRL) is a fundamental task in machine learning, aiming to encode high-dimensional graph-structured data into low-dimensional vectors. Self-Supervised Learning (SSL) methods are widely used in GRL because they can avoid expensive human annotation. In this work, we propose a novel Subgraph Gaussian Embedding Contrast (SubGEC) method. Our approach introduces a subgraph Gaussian embedding module, which adaptively maps subgraphs to a structured Gaussian space, ensuring the preservation of input subgraph characteristics while generating subgraphs with a controlled distribution. We then employ optimal transport distances, more precisely the Wasserstein and Gromov-Wasserstein distances, to effectively measure the similarity between subgraphs, enhancing the robustness of the contrastive learning process. Extensive experiments across multiple benchmarks demonstrate that \method~outperforms or presents competitive performance against state-of-the-art approaches. Our findings provide insights into the design of SSL methods for GRL, emphasizing the importance of the distribution of the generated contrastive pairs.
format Preprint
id arxiv_https___arxiv_org_abs_2505_23529
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Subgraph Gaussian Embedding Contrast for Self-Supervised Graph Representation Learning
Xie, Shifeng
Einizade, Aref
Giraldo, Jhony H.
Machine Learning
Artificial Intelligence
Graph Representation Learning (GRL) is a fundamental task in machine learning, aiming to encode high-dimensional graph-structured data into low-dimensional vectors. Self-Supervised Learning (SSL) methods are widely used in GRL because they can avoid expensive human annotation. In this work, we propose a novel Subgraph Gaussian Embedding Contrast (SubGEC) method. Our approach introduces a subgraph Gaussian embedding module, which adaptively maps subgraphs to a structured Gaussian space, ensuring the preservation of input subgraph characteristics while generating subgraphs with a controlled distribution. We then employ optimal transport distances, more precisely the Wasserstein and Gromov-Wasserstein distances, to effectively measure the similarity between subgraphs, enhancing the robustness of the contrastive learning process. Extensive experiments across multiple benchmarks demonstrate that \method~outperforms or presents competitive performance against state-of-the-art approaches. Our findings provide insights into the design of SSL methods for GRL, emphasizing the importance of the distribution of the generated contrastive pairs.
title Subgraph Gaussian Embedding Contrast for Self-Supervised Graph Representation Learning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2505.23529