Geometric Embedding of Heterogeneous Networks

• Applicant: Professor Dr. Thomas Bläsius
• Subject Area: Theoretical Computer Science
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 562551722

Embeddings provide a unified representation of information by mapping high-dimensional numerical data like images or symbolic data like text and graphs to a low-dimensional numerical representation. The goal of this proposal is to engineer algorithms for computing such embeddings of graphs. This includes developing the theoretical foundations of geometric graph representations, questions of computational complexity, as well as careful implementations of devised algorithms. Our focus lies on what we view as the major obstacle to computing low-dimensional but accurate embeddings: heterogeneous graphs with a highly varying degree distribution. Embedding into hyperbolic or weighted spaces instead of Euclidean space has the potential to overcome this obstacle but also poses new algorithmic challenges, which we address in the proposed project.

• DFG Programme: Research Grants