Enhancing Cyber-Physical Systems with Machine Learning - Drilling Holes into the Walls Between Disciplines Keynote Address

In the talk, we will introduce the notion of cyber-physical systems (CPS). For CPS, there is a complex interaction between the physical environment, hardware platforms, and the software running on them. We will briefly introduce CPS characteristics as well as challenges for their design. Mechanical, energy, thermal and dependability issues have to be considered, leading to the requirement of multidisciplinary design. Sensors may be providing huge amounts of data requiring sophisticated data analysis in order to generate reliable information. The collaborative research center CRC 876 at TU Dortmund (see http://sfb876.tu-dortmund.de) addresses the need for data analysis and multidisciplinarity. Its focus is on resource constrained machine learning and its applications. With this focus, it merges fundamentals of embedded systems and machine learning with applications in many scientific areas such as physics, medicine, logistics, and factory automation. In the second part of the talk, we will present exemplary results from cooperative work at the collaborative research center. These results will demonstrate the potential of interdisciplinary research using concrete examples. Final remarks will concern approaches for CPS education.