Hello,

I am Robert Jankowski, a researcher working on network geometry and machine learning.

Postdoctoral Fellow at TU Delft.

Research interests

network geometry
hyperbolic embeddings
machine learning
graph neural networks
TDA

About

I am a Postdoctoral Fellow at TU Delft studying shortest paths in large incomplete networks. I hold a PhD in network geometry from the University of Barcelona, where I worked on network embeddings in multidimensional hyperbolic spaces.

Postdoctoral Fellow · TU Delft

Work

Selected Publications

Communications Physics

Mapping bipartite networks into multidimensional hyperbolic spaces

R. Jankowski, R. Aliakbarisani, M. À. Serrano, M. Boguñá

Communications Physics 9(36), 2026.

pdf code

arXiv

Network representations reveal structured uncertainty in music

L. B. Rosselló, R. Jankowski, H. Bersini, M. Boguñá, M. À. Serrano

arXiv preprint arXiv:2509.14053, 2025.

pdf

arXiv

Hyperbolic Benchmarking Unveils Network Topology-Feature Relationship in GNN Performance

R. Aliakbarisani, R. Jankowski, M. À. Serrano, M. Boguñá

arXiv preprint arXiv:2406.02772, 2024.

pdf code

Nature Communications

The D-Mercator method for the multidimensional hyperbolic embedding of real networks

R. Jankowski, A. Allard, M. Boguñá, M. À. Serrano

Nature Communications 14(7585), 2023.

pdf code

See full publication list

Projects

Projects / Software

D-Mercator

Multidimensional hyperbolic embedding

Method for mapping complex networks into multidimensional hyperbolic spaces.

network geometry
hyperbolic embeddings

paper code

B-Mercator

Mapping bipartite networks into the multidimensional hyperbolic spaces.

Hyperbolic embedding of bipartite networks

network geometry
hyperbolic embeddings

paper code

FiD-Mercator

Feature-aware ultra-low dimensional reduction of real networks

Hyperbolic embedding of complex networks with nodes' features.

network geometry
hyperbolic embeddings

paper code

HypBench

Hyperbolic Benchmark for GNNs

A comprehensive benchmarking framework for graph machine learning, focusing on the performance of GNNs across varied network structures.

GNNs
benchmarking
machine learning

paper code

Talks

Selected Talks

Network representations reveal structured uncertainty in music. XIII Polish Symposium on Physics in Economics and Social Sciences, Warsaw, Poland, 2025.
Task complexity shapes internal representations and robustness in neural networks. IAIFI PhD Summer School, Boston, USA, 2025.
Mapping bipartite networks into multidimensional hyperbolic spaces. NetSci 2025, Maastricht, The Netherlands, 2025.
Network geometry and multidimensional hyperbolic maps of real networks. Faculty of Physics, Warsaw University of Technology, Poland, 2024.

See full talks list

News

Jan. 2026
Article "Shortest-path percolation on scale-free networks" published in Physical Review E.
Jan. 2026
Article "Mapping bipartite networks into multidimensional hyperbolic spaces" published in Communications Physics.
Oct. 2025
I defended my PhD thesis, titled "Hyperbolic Cartography of Complex Networks."
Sep. 2025
New preprint: "Network representations reveal structured uncertainty in music" is available online (link).
Sep. 2025
New preprint: "Shortest-path percolation on scale-free networks" is available online (link).
Sep. 2025
New preprint: "Chordless cycle filtrations for dimensionality detection in complex networks via topological data analysis" is available online (link).
Aug. 2025
New preprint: "Task complexity shapes internal representations and robustness in neural networks" is available online (link).
Mar. 2025
New preprint: "Mapping bipartite networks into multidimensional hyperbolic spaces" is available online (link).