Background and Project Content

AI-related projects, particularly those involving computer vision, multimedia analysis, and the training of large-scale machine learning models, require a significant amount of storage space and are highly computation intensive. To this end, the St. Pölten University of Applied Sciences (UAS )is currently building up a large-scale computing and storage infrastructure funded by GFF Lower Austria. All relevant fields of research and research projects at St. Pölten UAS that are related to AI and data analysis will benefit from this novel infrastructure. By building upon the most recent hardware developments in high-performance computing and GPU computing, we will be able to further exploit and expand our existing research and to intensify and deepen our research in areas like computer vision, machine learning, human gait analysis and data science in general.

Demand and area of application

The following fundamental needs will be covered by the new computing infrastructure:

• Performance: the new high-performance computing infrastructure will boost our data-driven research projects and enable more comprehensive experimentation and evaluation than possible today. Furthermore, it helps us to run and extend the most recent (and future) foundation models, like large language models (LLMs) and Language-Vision Models on premise with several billions of parameters.
• New methods: New methods emerge with a high pace in AI research today, some of which with the potential to be a game-changer in specific fields of AI. Our computing infrastructure allows us to build upon the most recent methodology and thereby enables us to keeping up with the latest trends and innovations. We consider this also an essential prerequisite for providing the best AI-related education.
• Data size: Gains in the performance of deep learning algorithms go together with a higher demand for storage space. Our novel hardware infrastructure will comprise a large redundant storage with high-speed connection to the computing server to facilitate AI experiments with large-scale data.

Results

The new server infrastructure not only allows us to use more complex methods and larger datasets, but also to produce more comprehensive results, publish them faster, and as a result gain broader visibility. The infrastructure enables us to conduct research at the frontiers of AI, increasing the innovation potential of both basic and applied research projects. The new infrastructure also increases the attractiveness of our AI and data science course offerings by allowing our students to experiment with the most modern AI methods. To sum up, the new server infrastructure benefits our research while also strengthening the status of St.Pölten UAS as an educational institution and, on a larger scale, the state of Lower Austria as a main player in AI in Austria.