<ul><li>Graph Representation Learning (GRL) aims to encode high-dimensional graph-structured data into low-dimensional vectors using Self-Supervised Learning (SSL) methods to avoid expensive human annotation.</li><li>A novel method called Subgraph Gaussian Embedding Contrast (SubGEC) is proposed in this work, featuring a subgraph Gaussian embedding module and optimal transport distances to measure similarity between subgraphs.</li><li>The approach ensures preservation of input subgraph characteristics while generating subgraphs with controlled distribution, enhancing the robustness of the contrastive learning process.</li><li>Extensive experiments show that SubGEC outperforms or competes effectively with existing state-of-the-art approaches, providing valuable insights into SSL methods for GRL.</li></ul>

Subgraph Gaussian Embedding Contrast for Self-Supervised Graph Representation Learning

Discover more