LOGSMS

Learning on Graphs: Symmetry meets Structure

NWO VIDI project LOGSMS

Mathematical Foundations of Machine Learning on Graphs and Networks

  • How can we mathematically characterize the power and limitations of existing graph learning models?
  • How can we design provably powerful graph learning models based on these insights?
  • How can we extend the realm of classical graph theory to shed more light on graph learning models?

The LOGSMS project (2026 -- 2030) will contribute next-generation machine learning tools for graphs and networks.