Plenary lecture

Day 2 (Wednesday 3 April) @ 10:00 – 10:45

Stefanie Jegelka (Massachusetts Institute of Technology)

Machine learning with symmetries: graphs, eigenvectors and generalization

In many applications, especially in the sciences, data and tasks have known structure, e.g., known symmetries. Encoding such invariances directly into a machine learning model can improve learning outcomes, while it also poses challenges on efficient model design. In this talk, we will look at the example of deep learning models on graphs and, in this context and beyond, on eigenvectors. For instance, eigenvectors are used as additional input, i.e., “positional encodings”, to graph neural networks, but also in applications related to point clouds and graphics. Graphs and eigenvectors possess symmetries such as invariance or equivariance to permutations and changes of basis. A first question is how to effectively and efficiently model such symmetries in neural network models. A second question is the approximation power of the resulting models, i.e., which functions they can approximate. A third question are empirical implications for various applications. Finally, we will address a more general question: empirically, encoding symmetries in a model often helps to learn with less data — is it possible to understand this theoretically? From a viewpoint of differential geometry, we show how encoding symmetries can help sample complexity of learning

Biography

Stefanie Jegelka is a Humboldt Professor at TU Munich and an Associate Professor in the Department of EECS at MIT. Before joining MIT, she was a postdoctoral researcher at UC Berkeley, and obtained her PhD from ETH Zurich and the Max Planck Institute for Intelligent Systems. Stefanie has received a Sloan Research Fellowship, an NSF CAREER Award, a DARPA Young Faculty Award, the German Pattern Recognition Award, a Best Paper Award at ICML and an invitation as sectional lecturer at the International Congress of Mathematicians. She has co-organized multiple workshops on (discrete) optimization in machine learning and graph representation learning, and has served as an Action Editor at JMLR and a program chair of ICML 2022. Her research interests span the theory and practice of algorithmic machine learning, in particular, learning problems that involve combinatorial, algebraic or geometric structure.