Graph machine learning and graph neural networks (GNNs) have been generating enormous interest in both academia and industry. (The first conference dedicated to graph machine learning is taking place later this year.) You find graphs pop up in other corners of the broader data science and machine learning world, though; regardless of your specialty, expanding your knowledge of this foundational concept is a good idea. We’re here to help: here are three standout articles that demonstrate the power of graphs in practical and accessible ways.
- How graph modeling will shape the future of unsupervised learning. The starting premise of Cristiana de Azevedo von Stosch and Abhishek Singh’s article is that "feature determination importance is still a fundamental problem in Machine Learning." It goes on to explore how graph-modeling frameworks can help us address this challenge in the context of unsupervised learning applications.
- On directed acyclic graphs (DAGs) and why they matter. Who doesn’t like a thorough, well-illustrated explainer? Matteo Courthoud‘s debut TDS post walks us through the potential uses of DAGs, and shows how they provide visual intuition on the variables we need to include in causal analyses (as well as those we’re better off omitting).
- A new architecture for designing powerful GNNs. Maxime Labonne‘s latest post is both a patient walkthrough of a recent paper on Graph Isomorphism Networks (GINs) and a practical tutorial with a full implementation of a graph classification task. You’ll leave it with a firmer grasp of the process behind choosing the right GNN architecture.
We published some fantastic articles on other topics in the past week, from new and established authors alike; our team also shared a couple of original features we think you might enjoy. Give these a try whenever you have the time this weekend:
- How does data science inform the work of a psychology-trained management consultant? Read our Q&A with Hannah Roos to find out.
- In a fascinating new series, Ang Li-Lian and coauthors Ibukun Aribilola and Valdrin Jonuzi __ explore the challenges of measuring intergenerational mobility in the U.S. Start with Part 1, which lays out the stakes of this complex topic.
- Agile practices are common among software developers. Jenny Abramov introduces a framework that empowers AI practitioners to adapt Agile to the specific needs of their projects.
- How can we promote the responsible use of AI at a societal level? For her first post on TDS, Maya Murad explores potential frameworks and the challenges of regulation.
- If you’re in a tinkering mood, don’t miss Florent Poux, Ph.D.‘s excellent new tutorial, where you’ll learn how to create 3D semantic segmentation datasets.
- Cassie Kozyrkov continued her series on the qualities that make great data analysts, and shared a new post that combines career development with a generous dose of inspiration.
- Finally, we celebrated the arrival of May with our latest Monthly Edition, featuring tips and strategies for building a great data dashboard.
Thank you for reading, sharing, and engaging with the work we publish. If you’d like to explore another way to show your support, consider becoming a Medium member.
Until the next Variable,
TDS Editors
