Fynn Oliver Gerding
- Room number
- IDEG050
- Position
- Research Associate
- Other Profiles
Fynn Gerding received the M.Sc. degree in Intelligent Technology from Radboud University in the Netherlands in 2026. In the same year, Fynn joined the Signal Processing and Speech Communication Laboratory at Graz University of Technology as a PhD researcher, supervised by Bernhard Geiger. Fynn is also an ELLIS PhD student, with Giuseppe Durisi (Chalmers University of Technology) as external advisor and a research stay in Gothenburg planned for 2027.
Fynn works on the FWF project Information Planes and Decompositions (2025–2029). Fynn’s research sits between machine learning and information theory and asks how neural networks represent and process information, and in particular where information is duplicated across their many neurons, layers and feature groups. Using tools such as mutual information and partial information decomposition, the work focuses on principled ways to measure this redundancy, on how it emerges and evolves during training, and on what it reveals about the role of individual network components. The aim is a better understanding of the inner workings of neural networks, and with it more transparent and trustworthy AI. At a later stage, these insights will be used to explore more sustainable deep learning, for instance by removing redundant elements or encouraging efficient representations during training.
News, talks and publications are collected on Fynn’s academic website, and the code accompanying the publications is available on GitHub.