StellarGraph is a commercial grade, open source, graph machine learning library written for Data Scientists in Python.
Many real-world datasets can be naturally represented as networks/graphs, with nodes representing entities, and links representing relations or interactions between entities.
In many predictive tasks, this network representation of data can help improve the predictive power, if the underlying network structure can be fully utilised by machine learning algorithms. However, bridging the gap between graph analytics tools and machine learning frameworks may be challenging.
To address these challenges, we've developed an open-source python library to democratise machine learning on networks/graphs for data scientists, developers, and researchers wanting to experiment with graph machine learning techniques and/or apply them to their network-structured data.
Open-source, actively developed and maintained
Easy to use, modular and extendable
State-of-art algorithms and capabilities:
Multiple variants of Graph Neural Networks (GCN, GAT, GraphSAGE, etc.)
Support for both homogeneous and heterogeneous graphs
Probabilistic inference and model calibration
Comprehensive set of demos
Technology: python, tensorflow, keras, networkx, scikit-learn.
The library is completely open source, Apache 2.0, and maintained by the StellarGraph team
EASY TO USE
Modular and extendable Python code, with a comprehensive set of demos
Research backed, engineering proven algorithms