Tackling big data

Digitalization & research / new graduate school for Data Science in Munich

November 05, 2018

As a result of digitalization, research is producing ever larger and more complex data sets. While these hold great potential for example for biomedicine, energy research, geo-research or robotics, they also need to be managed and interpreted. To address this need, the Munich School for Data Science @ Helmholtz, TUM & LMU (MuDS) has been established to train the next generation of researchers, who will tackle ‘big data’ problems. Over the next six years, the new graduate school will receive a total of twelve million euros in funding.

original

An example for machine learning: More than 200 detectors, which are arranged around the plasma, observe the X-ray light emitted by the plasma along different lines of sight. The large data volume recorded (left side) then is computationally assembled into a two-dimensional image of the plasma cross-section (right side). This allows to see exactly where the X-ray radiation was emitted in the hot plasma. For the fast computational transformation of the data a specially developed neural network was used.

The school was founded by the Helmholtz Zentrum München, the Max Planck Institute for Plasma Physics (IPP), the German Aerospace Center (DLR), the Technical University of Munich (TUM) and the Ludwig Maximilian University of Munich (LMU). The Leibniz Supercomputing Center (LRZ) and the Max Planck Computing & Data Facility (MPCDF), two major computing and data centers in the Munich region, are also associated to MuDS.

“Big challenges call for big solutions. We are delighted that we have managed to bring together these key players of the Munich metropolitan region for this project,” explains Professor Fabian Theis, Helmholtz Zentrum München/TUM, who will be the responsible coordinator of MuDS. The aim of MuDS is to combine training in methodological aspects with training in application domain areas, namely biomedicine, plasma physics, robotics and earth observation, to educate the next generation of data scientists. Data is increasing in volume and complexity, yet there is a shortage of experts to analyze it using the best methods possible, as the following sample illustrates: every single cell in our body contains about three billion DNA base pairs. That is equivalent to a library of 3,000 books, each with 1,000 pages, on each of which 1,000 letters are printed – and this is genetic information of only one cell.

Examples from other areas also show how great the demand for experts will be in the future. Take the latest generation of earth observation satellites, which generates petabytes of images and measuring data that is needed to research global change. One of the areas of interest at IPP is modeling future fusion power plants. For this purpose, model-based computer simulations are just as necessary as the evaluation of large data sets. “The new graduate school”, states Prof. Dr. Frank Jenko, IPP spokesperson in the MuDS, “will help to take these efforts to a new level by connecting leading plasma physicists and data scientists and enabling cross-fertilization between the four different application domains of MuDS”.

The Munich School for Data Science will offer joint projects for PhD students, each designed by two partners – a domain-specific application partner and a methodological partner. This will ensure that candidates receive methodological as well as application-specific training. In addition, participants will have the option of taking a course tailored to their needs, with a detailed onboarding phase followed up by advanced-level training. The training program will be integrated into existing courses provided by the universities as well as by the associated partners (LRZ and MPCDF), thus guaranteeing up-to-date, high-level training. MuDS will operate under the umbrella of the partner institutes’ highly successful graduate schools HELENA, HEPP, TUM-GS and Munich Aerospace.

Funding over a six-year period will total twelve million euros. Half of this amount will be provided by the Helmholtz Association of German Research Centers while the same amount is contributed by the participating institutes. The new Munich School for Data Science will be embedded in the Association’s strategy for digitalizing research (see Background). The first call for PhD students will be opened in December 2018 on www.mu-ds.de

Background:
The Helmholtz Association is creating four new innovative platforms to digitalize research and to this end will invest a total of 35 million euros a year. Each platform is located at one or at several Helmholtz Centers and will create an active network with other researchers. Specific funding lines will be established for this purpose. One such platform is the Helmholtz Information and Data Science Academy (HIDA) for up-and-coming scientists and graduate schools in this field. Within HIDA, five graduate schools will be set up at existing locations in Karlsruhe/Heidelberg, Jülich/Aachen/Cologne, Hamburg, Berlin and, of course, Munich.

Go to Editor View