PyG-SSL documentation

PyG-SSL is a Python library built upon PyTorch and PyTorch Geometric. It integrates state-of-the-art algorithms for Self-Supervised Learning on Graphs and offers simple APIs to call the algorithms from a variety of high-quality research papers. PyG-SSL can be helpful for many downstream tasks, depending on the dataset properties, as it contains many self-supervised algorithms:

Node-level Algorithms for General Graphs: DGI, GraphCL, MVGRL, GCA, SUGRL, ReGCL, BGRL, AFGRL

Algorithms for Heterogeneous/Multiplex/Multiview Graphs: HeCo

Graph-level Algorithms for Graph-level Representation Learning or Molecular Graphs: InfoGraph, GraphCL, GraphMAE

Note

This project is under active development.

💡 Get Started

• Introduction
• Starting Example

💯 Tutorial

• General Algorithms
  • DGI
  • GraphCL
  • MVGRL
  • GCA
  • SUGRL
  • BGRL
  • AFGRL
  • ReGCL
• Heterogeneous Algorithms
  • HeCo
• Molecular Graph Algorithms
  • InfoGraph
  • GraphCL
  • GraphMAE