Søren Hauberg
Also known as Soren Hauberg
Søren Hauberg researches geometric machine learning, Bayesian deep learning and representations. He became an associate professor at DTU in March 2017. In 2021, he coauthored a method that learns a geometric space from human demonstrations and uses paths in that space for robot motion. The experiments used a seven-degree-of-freedom manipulator, and the paper shared a RSS Best Student Paper Award. As of October 2026, he had been elected president of ELLIS Unit Denmark.
- Works
- 163
- Citations
- 2,009
- h-index
- 24
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
March 2017
A DTU career milestone
Hauberg became an associate professor at DTU’s Cognitive Systems group in March 2017. His research uses geometry both to design learning models and to represent constraints in observed data.
-
2021
Learning spaces for motion
Hauberg coauthored a robot-motion method that learns a Riemannian manifold from human demonstrations. A variational autoencoder models robot end-effector positions and orientations. The learned geometry provides paths for moving between points in the modeled space.
-
2021
Adapting and testing movement
The team tested the method on a seven-degree-of-freedom manipulator. The reported system reproduced motion patterns, avoided previously unseen obstacles and generated new movements in the tested settings. RSS gave the shared paper a Best Student Paper Award.
-
October 2026
Geometric learning and community work
Hauberg researches geometric machine learning, Bayesian deep learning and representations. As of October 3, 2026, he had been elected president of ELLIS Unit Denmark.
Sources and credits
Photo: orbit.dtu.dk; Søren Hauberg - Welcome to DTU Research Database; creator not stated. License: Not stated by source. Source: orbit.dtu.dk. Cropped to a square and resized. Ask us to remove or replace this photo.