Human-Centric AI & Security Seminar

Introducing graphons: a tool for modeling graphs

Our guest speaker:

Dr Sevvandi Kandanaarachchi, Senior Research Scientist, Statistical Machine Learning Group, CSIRO

Pricing

• Free

Dates and Times

Contact

Tina Wu

More information

Abstract:

Cyber security is intimately connected with graphs. From event graphs and attack graphs to networks of telecommunication scams, graphs are everywhere in a cyber context. Understanding and modeling their underlying structure is therefore important. In this talk, we introduce graphons—a non-parametric framework for modeling graphs—and explore recent advances, including their use in generative modeling. We discuss how this perspective could help us model and generate graphs relevant to cyber security.

Bio:

Sevvandi is a researcher currently in the Statistical Machine Learning Group. She completed her PhD in Differential Geometry and Partial Differential Equations at Monash University. After obtaining a Graduate Certificate in Data Science and Applications from Stanford, she moved into statistical machine learning research. Her interests include graphons, algorithm evaluation, Item Response Theory, and anomaly detection. She also works on applications where machine learning supports scientific discovery.