TU Wien CAIML

CAIML Seminar: “Physics-Inspired Learning on Graphs”

Michael Bronstein talked about GNNs from a different viewpoint that gave new insights and linked to methods from differential geometry, algebraic topology, and physics.

Michael Bronstein
Michael Bronstein

CAIML Seminar with Michael Bronstein took place on October 24, 2024.

Abstract

Message passing has become the predominant paradigm of learning on graphs, underlying the majority of Graph Neural Network (GNN) architectures and allowing to theoretically analyse their expressive power due to the relation to the Wesifeiler-Lehman graph isomorphism test. In this talk, Michael Bronstein considers GNNs as dynamical systems governed by parametric systems of differential equations. This viewpoint yields insights into some of the common plights of GNNs and suggests ways to dealing with them, as well as provides links to methods from differential geometry, algebraic topology, and physics.

About the Speaker

Michael Bronstein is the DeepMind Professor of AI at the University of Oxford. He was previously Head of Graph Learning Research at Twitter, a professor at Imperial College London and held visiting appointments at Stanford, MIT, and Harvard. He has been affiliated with three Institutes for Advanced Study (at TUM as a Rudolf Diesel Fellow (2017-2019), at Harvard as a Radcliffe fellow (2017-2018), and at Princeton as a short-time scholar (2020)). Michael received his PhD from the Technion in 2007. He is the recipient of the EPSRC Turing AI World Leading Research Fellowship, Royal Society Wolfson Research Merit Award, Royal Academy of Engineering Silver Medal, five ERC grants, two Google Faculty Research Awards, and two Amazon AWS ML Research Awards. He is a Member of the Academia Europaea, Fellow of IEEE, IAPR, BCS, and ELLIS, ACM Distinguished Speaker, and World Economic Forum Young Scientist. In addition to his academic career, Michael is a serial entrepreneur and founder of multiple startup companies, including Novafora, Invision (acquired by Intel in 2012), Videocites, and Fabula AI (acquired by Twitter in 2019).

Slides

Here you can download the slides:
Download the slides.

Video

If you watch this video, data may be transmitted to third parties.

Physics-Inspired Learning on Graphs - Michael Bronstein
Watch on YouTube: https://youtu.be/HusHP2elwyY

Before this lecture, an introductory lecture with the title “What You Always Wanted to Know about Graph Learning (And Never Dared to Ask)” was given by Ismail Ceylan. Here you can find the recording of Ismail Ceylan’s lecture.