Europe is home to 5 of the top 10 universities for computer science in the world.
It comes as no surprise, then, that Europe is a hub for ground-breaking AI research.
Leading institutes are increasingly tackling real-world challenges using AI. And the trend is not limited to those schools traditionally focused on computer science; business-oriented schools are also recognizing the benefits AI can bring.
Those institutes at the cutting-edge of AI research are turning to NVIDIA’s DGX Systems, the world’s first portfolio of purpose-built AI supercomputers, to provide the computing power they need.
Inspired by the demands of deep learning and analytics, NVIDIA® DGX™ Systems is built on the NVIDIA Volta™ GPU platform. Combined with innovative GPU-optimized software and simplified management, these fully integrated solutions deliver groundbreaking performance and results. NVIDIA DGX Systems gives data scientists the most powerful tools for AI exploration.
Here is how some of the leading European research centers are revolutionizing research with DGX Systems:
DFKI
The German Research Center for Artificial Intelligence (DFKI), the leading research center in Germany in the field of innovative commercial software technology using AI,was the first institution in Europe to adopt the NVIDIA DGX-2.
DFKI is currently using the DGX-2 to analyse large-scale satellite and aerial imagery using image processing and deep neural network training.
From this analysis, the team are developing applications to support rescuers in disaster-response scenarios. The aim is to efficiently identify disaster zones and assess the accessibility of infrastructure during events such as floods. The final applications will enable rescuers to make more informed decisions, faster.
Recently, their research has also focused on demystifying “black box” of neural networks. They are the first to dive deeper into the hidden processes and hope that their work will help make decision-making processes more comprehensible and transparent.
Prof. Andreas Dengel, Site Head at DFKI in Kaiserslautern and Scientific Director of the Smart Data and Knowledge Services Research Department stated, “The DGX-2 has two main benefits for us. We are able to develop deeper networks thanks to having more available memory. And, thanks to NVSwitch, we can now perform cross-GPU computing at up to 512 GB GPU RAM. This is crucial for achieving our big applications.”
KU Leuven
KU Leuven is using their DGX-1 to transfer academic research into industrial applications. Their EAVISE research team develop customisable AI solutions with a focus on object detection projects. Some of their current work includes pedestrian tracking and detecting elderly people falling. By applying their AI expertise, the EAVISE team help to improve the efficiency and accuracy of these projects.
KU Leuven chose the DGX-1 as it enables them to build robust solutions with a lot of data. For them, the biggest benefit is the speed of inference. Before, it was almost impossible to simulate deep models on datasets like ImageNet because it took too long. Now, the speed boost has enabled many researchers to switch from the transfer learning principle to training complete networks. The DGX-1 has also improved the team's efficiency, giving them more time to try out new ideas.
University of St. Gallen
The University of St. Gallen, a business-focused school, recognized the increased demand for AI expertise and the benefits this expertise would bring to their students and staff.
In 2018, they established a new AI and Machine Learning lab, led by Professor Damian Borth.
The lab's research focuses on deep neural networks applied to four different areas:
- Architectures and models
- Multimedia analysis and synthesis
- Remote sensing and analyzing satellite data
- AI for financial markets
Aside from these, they are exploring the need to make AI usable at a broad scope. This involves working on regulation identity and the versioning of neural networks.
Professor Borth explained, "We will be using the DGX-2 for the training of multiple neural network models. We plan to investigate how we can guarantee that two networks, which come to the same conclusions, actually started at different points and are therefore copyright compliant. For this, we need huge computational power, we need multiple GPUs which we can train neural networks very quickly, so we can simulate the space accurately."
The foundation of the University of St. Gallen's research lab is indicative of the general acceptance that AI is becoming integral to every industry. If we are to make sense of the large amounts of data our society generates, we need the power of AI.
These are just some of the research centers who are powering their innovation with NVIDIA DGX Systems. Many more, including the University of Agder, University of Kaiserslautern, the Fraunhofer Institute, and the University of Aalborg, are using the world’s first portfolio of purpose-built deep learning systems for their research.
For more information, check out the resources below and keep up to date with the latest higher education and research news by signing up for our newsletter.
Resources:
- Discover how DFKI is using deep learning to analyse disaster zones.
- Learn how KU Leuven is at the forefront of artificial intelligence development.
- See how the University of St. Gallen is building their heritage in finance and business with their AI research.
- Check out other GTC Sessions featuring deep learning and AI.
