Dr. Pantelis Elinas of Data61, CSIRO

We live in a connected world and generate a vast amount of connected data. Social networks, financial transaction systems, biological networks, and transportation systems are all examples. Machine learning researchers have proposed many new algorithms for tackling the challenge of building predictive models for network-structured data. These algorithms exploit relationships in the data and achieve state-of-the-art results on common tasks such as classification and regression.

In this talk, I will discuss fundamental concepts in modern network analysis and provide an overview of the new and fast developing graph machine learning specialization more generally known as geometric deep learning. Furthermore, I will introduce StellarGraph, an open source graph machine learning Python library under development at CSIRO's Data61. StellarGraph implements many state-of-the-art graph machine learning algorithms and it exposes a clean, consistent and easy-to-use API that integrates smoothly with the Python data science ecosystem. Lastly, I will discuss some of the challenges in graph machine learning especially in regard to practical applications.

Biography:

Dr. Pantelis Elinas is a Senior Research Engineer at CSIRO's Data61. His professional background spans the areas of Computer Science, Robotics and Machine Learning. In the past 15 years, Pantelis has tackled problems in robotics, computer vision, computer graphics, assistive living, and mining automation. Currently, as a member of the Investigative Analytics team at Data61, Pantelis is developing machine learning methods and tools for graph mining and analysis. He enjoys working on interesting problems, sharing knowledge, and developing useful software tools and applications.