Open Source Machine Learning Library for Large Data Graphs Gets Update

Last month open source, Python-based machine learning algorithm library for graphs created by data scientists and others at Australian-based research and technology company CSIRO Data 61, got a major update (0.80) with numerous new features.

Aside from various bug fixes, the update adds a number of new algorithms, including Directed GraphSAGE, an Attri2vec algorithm and GAT saliency maps -- support for which is another new feature of the update. Click on the links to find out more about each algorithm.

Other new features include improvements in activations and API support, multithreading improvements and a unified API for classes. "The 0.8.1 release of the library...further simplif[ies] graph machine learning workflows through standardized model APIs and arguments," the developers said in an announcement of the update.

There are also numerous refactoring-focused updates, including a change from using keras to tensorflow.keras and refactoring changes for the Ensemble class.

The project was originally released in June 2018. It focuses on combining entity resolution, predictive modeling and data visualization to "leverage the full graph relationships to enable predictions that are not currently possible with standard techniques."

More information on the project can be found here.

About the Author

Becky Nagel is the vice president of Web & Digital Strategy for 1105's Converge360 Group, where she oversees the front-end Web team and deals with all aspects of digital projects at the company, including launching and running the group's popular virtual summit and Coffee talk series . She an experienced tech journalist (20 years), and before her current position, was the editorial director of the group's sites. A few years ago she gave a talk at a leading technical publishers conference about how changes in Web browser technology would impact online advertising for publishers. Follow her on twitter @beckynagel.