Arbeitsgruppe

Trained in computational linguistics, artificial intelligence, and
mathematics, my PhD resarch focussed on analyzing non-classical
reasoning mechanisms, specifically analogies and theory blending,
resulting in a category theoretic framework and a logic-based
formalization.

My current work aims at gaining a better understanding of
representations acquired by deep networks, mainly in the vision and
audio domain.  In our work group, we apply various methods from
statistics and information theory to investigate represntational
invariances as well as the learning behavior.

Lukas studied computer science in his bachelor and cognitive science in his master.
In his master thesis "Model-Based Leaf Instance Segmentation on 3D Point Clouds" he applied classical computer vision methods in combination with hyperparameter optimization algorithms to segment individual leafs.
He currently investigates the training behaviour of Kolmogorov-Arnold Networks on various initialization strategies.

Robin received his M.Sc. in 2022 from RWTH Aachen University, where he worked on improving the Recurrent Neural Aligner for speech recognition. 

His research interests include symbolic regression and reinforcement learning. 

Currently, he is focused on applying machine learning methods to formal systems.