Press Release No. 20 | June 28, 2022

DFG to Fund Eight New Research Units in the Field of Artificial Intelligence

Strategic funding initiative boosts AI research in a wide range of fields / Total of around €31.4 million euros for first funding period

Strategic funding initiative boosts AI research in a wide range of fields / Total of around €31.4 million euros for first funding period

The DFG is to fund eight new Research Units under its strategic funding initiative in the field of artificial intelligence (AI). The DFG Joint Committee approved the new consortia on the recommendation of the Senate; the two bodies met in Freiburg im Breisgau, where the DFG’s Annual General Meeting is taking place from 27 to 29 June. The new Research Units will receive total funding of approximately €31.4 million, including a 22-percent programme allowance for indirect project costs. They will be funded for a maximum of two four-year periods.

“Artificial intelligence methods are crucial in all academic disciplines. For this reason, research into AI itself should be accompanied by its effective integration in fundamental scientific research,” said DFG President Professor Dr. Katja Becker. “In addition to the other newly established consortia, we’re now delighted to be able to fund eight Research Units in an area that is so important. The aim is for these Research Units to engage and network with each other and with other national and international actors in the field – not only at their universities but also at various events.”

The AI funding initiative was approved by the DFG in October 2019 with a total package of around €90 million. The initiative consists of two priority aspects: firstly, the funding of the eight new Research Units, each of which is to achieve a dovetailing of topics and personnel between a research priority at their university and research in the field of AI methodology. Research Units enable scholars to pursue current and pressing issues in their areas of research and take innovative directions in their work.

Secondly, a total of 15 Emmy Noether Junior Research Groups were established in two previous rounds of calls for proposals. The aim here is to attract the next generation of highly qualified academics with a research focus on AI methods. (A link to the list of funded Emmy Noether groups is provided under Further Information)

The eight new consortia in detail
(in alphabetical order of the spokespersons’ HEIs)

Whether images, genome sequences or time series: high-throughput measurements now provide the biomedical sciences with important structured data, but these are often characterised by heterogeneity, disturbance variables and sampling bias. By “integrating deep learning and statistics to understand structured biomedical data”, the Research Unit seeks to help eliminate uncertainties in the future, making statistical models more flexible and significantly improving the interpretability of the structured data situation by establishing a link to biomedical applications in general. (Spokesperson: Professor Dr. Sonja Greven, HU Berlin)

Achieving sustainable agriculture to fight hunger is a key goal of the UN 2030 Agenda. With its state-of-the-art measurement and monitoring technologies for data-driven optimisation, precision agriculture could open up a pathway here. The Research Unit Automation and Artificial Intelligence for Monitoring and Decision-Making in Horticultural Crops (AID4Crops) is looking for novel AI algorithms that can provide an optimum description and prediction of the condition of the plants, while at the same time automatically deciding which data needs to be collected for this purpose. This coupling of monitoring and decision-making will also provide the basis for proposing concrete AI-based management decisions. (Spokesperson: Professor Dr. Christopher McCool, University of Bonn)

Cartography and physical geodesy are two central areas of the science of surveying the Earth’s surface that deal with geometric abstractions of the real world. The Research Unit Algorithmic Data Analysis for Geodesy (AlgoForGe) focuses on these areas. Its main aim is to bridge the gap between current research in the fields of AI and geodesy, tapping into connections between the two disciplines to establish a more solid algorithmic basis for measuring and mapping the Earth’s surface. (Spokesperson: Professor Dr. Petra Mutzel, University of Bonn)

Nowadays, digital devices, sensors and technologies collect information biological, social and lifestyle information “on the side”. This digital trace of biographies can complement longitudinal epidemiological studies with the aim of achieving more personalised preventions and interventions in healthcare. The Research Unit Lifespan AI: From longitudinal health data to life course inference aims to further develop AI methods and tools in order to improve understanding of the individual development of diseases based on the data collected and using deep learning models, thereby allowing them to be used for empirical health research. (Spokesperson: Professor Dr. Tanja Schultz, University of Bremen)

The Research Unit Abstract Representations in Neuronal Architectures (ARENA) focuses on the emergence and coding of knowledge representations in functional networks of the human brain and in AI models at different levels of abstraction: starting with simpler phenomena such as seeing an object from different angles, moving onto neuronal categorisations and extending to abstraction models based on complex perceptual processes. The aim is to provide insights into the organisational processes in the brain – while at the same time providing inspiration for AI development. (Spokesperson: Professor Dr. Ingo Marzi, University of Frankfurt/Main)

The aim of the Research Unit Deep Learning on sparse chemical process data is to establish deep learning (DL) methods in chemical process engineering. Its central thesis is that DL opens up new avenues in anomaly analysis, state prediction, decision-making and autonomous processes in areas that are crucial to the field. To this end, the researchers involved aim to carry out specialised experiments to generate the large data sets required for the development of such methods, as these cannot normally be obtained from chemical plants. (Spokesperson: Professor Dr. Marius Kloft, TU Kaiserslautern)

Sophisticated processes are required for high-quality, cost-efficient production. Since experimentation has to be done in order to achieve this, development has been particularly costly up to now, especially when new materials and processes are used, the production process is highly complex, or no mature models are available. The systematic use of AI has the potential to be cheaper, faster and more efficient. For this reason, the Research Unit AI-based methodology for the rapid upgrading of immature production processes will set out to look for fundamentally new solutions in this area. (Spokesperson: Professor Dr. Jürgen Beyerer, KIT Karlsruhe)

According to the latest findings, most of the disciplines that deal with image information and 3D models (“visual computing”) would achieve significantly better results if complete data analysis could be carried out in a single step – and not in a staggered process, as has been the case so far. However, this requires basic research relating both to the sensors used and in the field of machine learning. It is this type of investigation that the Research Unit Learning Optimal Image Data Sensors wishes to conduct. In the long term, the aim is to develop a new methodology for the joint development of sensor systems and data-analysing networks for application. (Spokesperson: Professor Dr. Michael Möller, University of Siegen)

Further information

Media contact:

  • DFG Press and Public Relations
    Tel. +49 228 885-2109

Team spokespersons can also provide detailed information.

Contact at the DFG Head Office:

On the funding initiative in the field of artificial intelligence

Emmy Noether Junior Research Groups in the field of artificial intelligence methods

All press releases on the 2022 Annual General Meeting are also to be found in a digital press kit at , which is supplemented on an ongoing basis