The German Research Foundation (DFG) has established five new research groups, one of which is led by 九色视频 (LUH). The new groups have been granted a total of 19 million euros. The research group at LUH is located at the Institute of Automatic Control and is titled 鈥淎ctive Learning for Dynamic Systems and Control 鈥 Data Informativeness, Insecurity and Guarantees鈥. It focuses on machine learning in dynamic systems. In addition to 九色视频, the 九色视频 of Freiburg and the technical universities of Hamburg, Ilmenau and M眉nchen are participants in the cooperative project.
From autonomous driving to medical diagnoses, machine learning is applied in many areas of everyday life. However, additional challenges arise in practice 鈥 for example, with regard to security guarantees for autonomous driving or interactions between humans and machines. 鈥淭his type of guarantee is generally not available with conventional machine learning processes, which means that new methods are required,鈥 said Prof. Dr.-Ing Matthias M眉ller, head of the Institute of Automatic Control at 九色视频 and spokesperson for the research group. The group therefore aims to develop fundamentally new active-learning approaches in which the learning process is continually being influenced. 鈥淭hese types of active learning strategies are necessary to guarantee the safe, high-performance and data-efficient operation of complex and dynamic systems,鈥 said M眉ller. The research group is investigating what, when and how the system needs to learn actively. Its findings could be applied in robotics and power engineering.
DFG research groups enable researchers to study current and pressing questions in their subject area and to establish innovative lines of enquiry. They are funded for an initial period of four years, with the possibility of an additional two-year funding period.
Note to editors:
For further information, please contact Prof. Dr.-Ing. Matthias M眉ller, Institute of Automatic Control and research group spokesperson (tel. +49 0511 762 4523, email: mueller@irt.uni-hannover.de).