The German Research Foundation (DFG) has approved a high-performance GPU system for Anhalt University of Applied Sciences. The positive scientific assessment makes the funding by the state of Saxony-Anhalt possible and paves the way for data-intensive science at a new level.
The new system will provide researchers with computing capacities that far exceed those of conventional computers. The German Research Foundation (DFG) has scientifically assessed and recommended the acquisition of a high-performance GPU system. The state of Saxony-Anhalt can thus provide the necessary funds from the DFG's large-scale equipment program. The new system offers computing power that can replace more than a hundred high-end PCs in the field of artificial intelligence and deep learning. Complex simulations are accelerated by orders of magnitude. GPU stands for Graphics Processing Unit - processors originally developed for graphics calculations that are particularly good for parallel calculations.
All departments benefit from centralized solution
The university is integrating the new GPU system into its existing Kubernetes infrastructure. Kubernetes is a platform that automates and scales the management of applications in data centers. In future, it will serve as a central computing unit for all departments at the university, which can use it for data-intensive research projects. "Previously, our researchers had to rely on external cloud service providers or work with insufficiently powerful infrastructure, which led to bottlenecks," explains Dr. Katharina Holstein from the university's KAT network. "The new system makes us more independent and efficient." The university is planning regular workshops in which researchers can learn how to use the new infrastructure in order to establish its use as widely as possible. The GPU system has also created additional staff capacity for technical support, headed by Prof. Dr. Christian Hänig. This centralized model saves resources: not every working group has to build up its own expertise or apply to funding bodies.
Data volumes are growing due to more research projects
The demand for high-performance computing has risen continuously in recent years. More and more research projects are generating large and complex amounts of data that can hardly be handled with conventional infrastructure. The new GPU system can support this data-intensive research in various areas: Environmental and nature conservation, AI-supported process optimization, biomedical image processing as well as deep learning and machine learning. Machine learning methods in particular, which evaluate large data sets and iteratively improve models, benefit from the parallel computing power of GPUs.
Concrete application in Agriculture
An example from remote sensing shows the practical benefits: Researchers here work with hyperspectral image data from Agriculture. These special hyperspectral cameras not only capture the three primary colors like normal cameras, but also up to 256 different color channels. The resulting data volumes are extremely detailed and correspondingly large.
The expensive special camera technology costs around 300,000 euros, which very few farmers can afford. The researchers are therefore developing algorithms that extract the crucial information from the complex raw data - such as the nitrogen requirements of plants, as well as disease infestation, water requirements and other nutrients. One goal: to enable cost-effective sensors, provide farmers with practical information and advance precision agriculture in the long term.
Students work with real data
The large-scale device is also used in teaching. Students in the fields of Data Science and Artificial Intelligence can work with real data sets during their project work. Instead of using standardized training data, they train on real problems - for example with maps of fields from the region. The practical experience makes a significant difference: "The students go through the entire process from data preparation to evaluation and gain valuable practical experience," explains Dr. Katharina Holstein, who herself supervises the lab in "Machine Learning" and the "Data Science" project at the Department of Computer Science and Languages.
Research strength confirmed
The DFG recommendation underlines the research strength of Anhalt University of Applied Sciences. It is one of the ten universities of applied sciences with the most third-party funding in Germany and is an active member of the Competence Network for Applied and Transfer-Oriented Research (KAT). KAT is an association of universities of applied sciences in Saxony-Anhalt.
Contact:
Prof. Dr. Christian Hänig
E-Mail: christian.haenig@hs-anhalt.de
