Geometric Machine Learning

We study geometric structure in data and models and how to leverage such information

for the design of efficient machine learning algorithms with provable guarantees.

Our Research

Characterizing data and geometry and learning data representations in suitable non-Euclidean spaces. Non-Euclidean geometry of foundation models.

Leveraging geometric structure for efficient learning on graphs. Design of expressive and stable graph machine learning architectures.

Designing geometric architectures. Understanding learning-theoretic trade-offs in geometric settings.

Leverage geometric machine learning for resource-efficient and interpretable scientific machine learning.

Exploiting geometric structure in (non)convex (constrained) optimization.

Leveraging data and model geometry for more efficient and transparent foundation models.