Ein hochverfügbares KI-Servicezentrum für sensible und kritische Infrastrukturen (KISSKI)
The central approach for the KISSKI project is research into AI methods and their provision with the aim of enabling a highly available AI service centre for critical and sensitive infrastructures with a focus on the fields of medicine and energy. Due to their relevance to society as a whole, medicine and the energy industry are among the future fields of application-oriented AI research in Germany. Beyond technological developments, artificial intelligence (AI) has the potential to contribute significantly to social progress. This is particularly true in areas where digitalisation processes are increasingly gaining ground and complexity is high. For both medicine and the energy industry, the pressure to innovate, but also the potential, is immense due to the availability of more and more distributed information based on a multitude of new sensors and actuators. The increasing complexity of the tasks as well as the availability of very large data sets offer a high potential for the application of AI methods in both subject areas.
In addition, the Service and Competence Centre is open to requests from other subject areas and disciplines that are compatible with these mission goals and can benefit from the services and offerings provided.
Areas of application for AI methods in the energy industry include the challenges of feeding renewable energies (e.g. wind or solar energy), the coupling of sectors (electricity, heat, transport), intelligent management of controllable consumers and automated grid operation. Due to the large number of spatially distributed generators, consumers and active prosumers, the uncertainties but also the optimisation possibilities for an economic and secure energy supply increase enormously. A holistic control as well as a reliable monitoring of the system properties are no longer possible manually. The multitude of data streams from the individual energy grids, plants and markets that have been built up in recent years can only be meaningfully monitored, evaluated and integrated into innovative applications using intelligent data-driven methods. AI will be an essential enabler here.
Contact | Dr. Julian Kunkel | ||
Website | kisski.gwdg.de |
People from HPS
Funded Partners
- University Göttingen (Coordinator)
Consortium
- University Göttingen (Coordinator)
Goals
The main concern of the KISSKI project is to provide a highly available KI-servicecenter for critical and sensitive infrastructure. The focus is on the socially highly relevant fields of medicine and energy as these have high potential for future developments that benefit Germany.
With KISSKI we plan to provide services such as consulting with regard to the application of AI methods, access to hardware, software and data for the development and inference of AI models, development of proof of concept systems as well as training and teaching offerings even for school students.
Our consulting offerings aim to introduce newcomers to AI by explaining how it can be used to enhance existing processes or create entirely new products. The hardware provided through KISSKI will be hosted by the GWDG in Göttingen with a geo replication in Hannover handled by the Leibniz University. To enable customers to quickly try out and start developing AI solutions, we provide curated data and pre-trained models that can be further refined towards a specific use case. Customers that already have a concrete idea for an application of AI can receive support with the development of a first prototype, which can even be hosted on our infrastructure. The training materials consist of courses offerings as well as downloadable learning materials that can even be received by school students.
KISSKI also includes research efforts that aim to advance the possibilities of working with AI. The focus is on four aspects, (1) improving the scalability of training while accounting for sensibility of the data, (2) automating the scaling of AI inference with a focus on minimizing the response time, (3) enabling heterogeneous hardware for AI modells such as Graphcore, FPGA or Google TPU and (4) developing secure and efficient data management systems and processes that are able to account for the data protection requirements of our customers while also not losing out on performance.