Short CV

Lukas Einhaus, M.Sc. works since April 2020 as a researcher and PhD student at the Embedded Systems department of the University of Duisburg-Essen. After his Bachelors thesis about programming abstractions for concurrent embedded systems he received his Masters degree with a focus on distributed, reliable systems at the University Duisburg-Essen. In his Masters thesis he did research on quantizing neural networks.

Until March 2022 he worked on the research project “KI-Sprung: LUTNet – An Energy-Efficient AI Network Based on Elementary Lookup Tables,” which was funded by the Federal Ministry of Education and Research. This project investigated how pre-trained and pre-optimized neural networks can be used to develop AI solutions that that run with high efficiency on FPGAs.

Research

My research focuses on techniques for designing neural networks that allow for an efficient implementation in hardware, especially on FPGAs. My main interest herein lies on so called quantized or low precision neural networks. These differ from conventional full precision networks in the heavily reduced bit depth used for computational operations or the representation of the information flow.