I received my B.Sc. degree in biomimetics at the Westphalian University of Applied Science Bocholt in 2015. This line of studies focused on mechanical engineering with specification in light weight construction, as well as biology with specialization in sensory systems and bio-inspired information processing.
I deepened my knowledge in biological sensory information processing during my M.Sc. in Neurobiology at Bielefeld University. I worked together with Martin Egelhaaf and Elisabetta Chicca on the implementation of the visual information processing system of flying insects onto neuromorphic hardware and applied the abstracted artificial information processing network in the context of driving robots in order to avoid collisions with surrounding objects.
I am interested into transferring insights of biological information processing onto state-of-the-art soft- and hardware. The transfer is not a one-to-one copy of the biological model, but rather an abstraction of the underlying principle in order to implement the developed neural circuits in Spiking Neural Networks (SNN).
Currently I am exploring and developing SNN architectures, which directly operate on local spatio-temporal contrast changes (events) rather than frames. These networks inherently represent time in their nature of computation. I am investigating how we can introduce a conceptual understanding of spatio-temporal patterns in the context of driving robotic systems.
The process of learning these spatio-temporal features is completely unsupervised and event-driven. The key insight, which also differentiates my approach from conventional machine learning, is that the network is continuously learning using STDP and that the error is expressed as failed prediction of the future in strictly local neighborhood.