Professor Dr. Florian Bernard from the Institute of Computer Science II – Visual Computing has been awarded one of the renowned ERC Starting Grants. This grant is given by the European Research Council (ERC) to outstanding young scientists who want to develop pioneering projects. The ERC is an institution set up by the European Commission to fund top-level research. The starting grant supports innovative fundamental research with around 1.6 million euros for a maximum of five years. The aim is to support the scientific careers of young researchers and strengthen their independence. In addition to Bernard, six other researchers from the University of Bonn were awarded an ERC grant for their projects.
Combining machine learning and human knowledge for more sustainable algorithms
Bernards research area Visual Computing has the potential to be used in a broad range of different areas, from medical image analysis to autonomous driving. “The field of visual computing is making more and more use of machine learning methods,” explains Professor Florian Bernard from the Institute for Computer Science II—Visual Computing. “These approaches have already helped solve several problems over the past few years, such as the synthesis of photorealistic images and the automated creation of statistical shape models of 3D organs. However, a major drawback is that progress has mainly been achieved by increasing the amount of resources used, such as energy, data or hardware.”
Bernard wants to use the ERC Starting Grant to tackle this problem in the project “Harmonizing Observations and Underlying Principles for Visual Data Association” (Harmony for short).
Together with his working group, he is studying intelligent visual computing algorithms that combine machine learning techniques with human knowledge. Specifically, “Harmony” aims to understand commonalities in visual data, such as images, videos and 3D models. “I’d like to build upon what we already know about fundamental principles, such as physics or geometry, and combine it in an appropriate way with the collective knowledge that is implicitly present in large collections of visual data,” says Professor Bernard, who is also a member of the Modelling Transdisciplinary Research Area at the University of Bonn. “Combining these will help machine learning systems to avoid learning things that we already know from scratch.”
The grant will allow the computer scientist to broaden his research: “We’ll be able to look at more areas with exciting unanswered research questions while also linking existing and new strands of research more closely together.”
Professor Florian Bernard studied computer science at Trier University of Applied Sciences and went on to do his doctorate at the University of Luxembourg. He then worked as a postdoc there and at the Max Planck Institute for Informatics before joining the Technical University of Munich as a visiting professor in 2020. The following year, he came to the University of Bonn as an assistant professor in the Institute for Computer Science II—Visual Computing, being made professor in 2023.