Announcing the release of StellarGraph version 1.1 open source Python Machine Learning Library for graphs

StellarGraph is an open-source library implementing a variety of state-of-the-art graph machine learning algorithms. The project is delivered as part of CSIRO’s Data61.

Version 1.1 delivers new and improved demos and examples plus further overall performance and memory usage improvements. Get started with pip install stellargraph.

New algorithms include:

  • Unsupervised graph representation learning
  • Unsupervised RGCN with Deep Graph Infomax
  • Native Node2Vec using Tensorflow Keras, not the gensim library.

Some new algorithms and features are still under active development, but are available as an experimental preview:

  • RotatE: a knowledge graph link prediction algorithm that uses complex rotations (|z| = 1) to encode relations
  • GCN_LSTM (renamed from GraphConvolutionLSTM): time series prediction on spatio-temporal data (still experimental, but improved since last release).

Some of the performance enhancements in this release include:

  • The StellarGraph class continues to get smaller, faster and more flexible
  • Better demo notebooks and documentation to make the library more accessible to new and existing users
  • Significant improvements to support for the Neo4j graph database
  • Significant speed enhancements to various random walkers
  • Several bug fixes and other changes.

Jump into the new release on GitHub. StellarGraph is a Python 3 library. See full v1.1 release notes here.

We always welcome feedback and contributions.

With thanks, the StellarGraph team.