DAAD postdoc AI tour: AI talent for Germany

How does the human brain work? How does the highly complex interaction of countless neurons create typical patterns that we perceive as smells, sounds or images? As is the case in many other fields, artificial intelligence has helped increasingly to find answers to complex questions, also in the area of neuroscience. AI technology meanwhile allows, for example, for processes that are still difficult to understand on a biological level, to be modelled on the neuronal level, successfully enough to make actual predictions possible.

Martin Schrimpf, a PhD student at the prestigious Massachusetts Institute of Technology (MIT) conducts research in this field. The idea behind this work is to focus on so-called spikes, which are the electrical impulses that provide for information exchange among neurons. ‘The most fascinating part is that by controlling the spikes, it is ultimately possible to immediately affect the visual perception apparatus,’ says Martin Schrimpf.

Exorbitant annual salaries

With his area of research, his prestigious employer and an impressive list of publications, Martin Schrimpf is part of the line-up of young AI talents that are currently being wooed by universities and industry more intensely than is the case in almost any other area. This is a challenge for a country such as Germany that sees itself as a location of top level AI research. ‘We are really struggling to fill our postdoc positions at the moment,’ points out Dr Daniel Cremers, Professor of Imaging and Artificial Intelligence at the Technical University of Munich (TUM).

Cremers goes on to explain that this is due in particular to the exorbitant salaries paid in the big tech environment with which the top doctoral candidates can be lured away from the universities before or right after obtaining their degrees. ‘We are talking about half a million dollars a year here. This is a lot more, of course, than German universities are able to offer for postdoc positions according to the pay scale.’ So what can higher education institutions and research facilities do to encourage young talent to come to Germany? An important first step is to make such talents aware of Germany as a top quality research location. ‘In the area of AI, we can certainly keep up with the elite universities around the world,’ says Daniel Cremers.

High level of research quality in Germany

This is exactly the goal of Postdoc-NeT-AI, a networking tour for your AI researchers, organised by the DAAD. Programme leader Luca Wettlaufer explains that regrettably, the first tour in November 2020 had to be held as a virtual format. ‘However, this enabled us to realise various personal conversations between postdocs and German researchers. This would not have been possible to the same extent in the context of a real-life tour.’ When selecting the candidates, the organisers looked out explicitly for highly-qualified talented people. The participating research institutions were also involved in the selection process. Wettlaufer points out that almost 100 individual conversations between fellows and researchers could be arranged in line with the respective areas of interest. 

During the week-long virtual tour, the 21 fellows had the opportunity to talk to representatives of several German AI hotspots. Following the initial virtual icebreaker event, at which the young researchers were able to get to know each other and exchange views, two top institutions per day introduced themselves and their research activities. The institutions are based in Stuttgart, Tübingen, Munich and Augsburg. After these presentations there was time for individual discussions. While Martin Schrimpf was particularly impressed by the high level of research quality at the University of Tübingen, Nooshin Ghavami, a postdoc student at Kings College London, was enthused by the German research system’s diversity.

Ghavami conducts research into the use of deep learning methods in medical imaging processes. ‘We develop applications that help doctors to identify certain anomalies more effectively, especially when they are not very experienced themselves,’ she explains. While she regretted not being able to be there on site, she felt the tour was still worthwhile in the virtual format. ‘I found out about many German research institutions that I had never heard about before.’

INFO BOX POSTDOC-NET-AI 04/2021

The next AI networking tour by the DAAD takes place from 26 to 30 April 2021. Excellent international researchers from the area of artificial intelligence are granted the opportunity to learn about important German AI hotspots and possible career options. The tour’s focus areas are big data, data analytics, data mining and information retrieval. The host institutes include the Center for Scalable Data Analytics and Artificial Intelligence (ScaDS.AI), the Berlin Institute for the Foundations of Learning and Data (BIFOLD), the Data and Web Science Group of the University of Mannheim and the Center for Artificial Intelligence and Data Science (CAIDAS) of the University of Würzburg.

Those interested in taking part can apply until 8 March 2021

To the application form

Research for the general public

Diego Marcos, a postdoc at the Dutch Wageningen University felt that the strongest components of the tour were the virtual one-on-one talks. ‘We had half an hour each to talk to the project leaders. This is pure luxury, you rarely get an opportunity like this, not even at a conference.’ Just like Nooshin Ghavami, Diego Marcos also works in the area of AI assisted image recognition, but in a more basic sense. ‘The systems for recognising particular patterns are very good by now, but they still make the occasional mistake. Due to the fact that they process information in a way so different from us humans, it is difficult to find out, what exactly the problem is,’ he explains. ‘This is what my work is all about.’

Marcos is aware of the fact that his research field that is known as interpretable or explainable AI among experts, is also interesting for industrial uses and that major technology companies are meanwhile responsible for more publications than universities, also in the area of basic research. ‘It simply feels better to me, to work for a public institution.’ 

Flexible recruiting

This is good news for professors at German universities who are looking for talent. Nevertheless, a specific job offer in Germany does not have to be the only option for attracting young researchers. Dr Zeynep Akata, a professor at the University of Tübingen has a very pragmatic approach. ‘I perceived the tour as a networking event in the first place, and not necessarily as a recruiting event for particular postdoc positions.’

Zeynep Akata believes this to be the exact right concept for being able to react flexibly in the ‘war for talent’ at the AI location Germany. ‘This first encounter may lead to a collaboration. A collaboration that in turn can lead to an application, not only as a postdoc candidate.’ A few weeks after their virtual meeting, she sent Martin Schrimpf a call for applications by the University of Tübingen. Not for a postdoc position, however, but for a professorship. Martin Schrimpf says that he can well imagine applying and returning to Germany.

author: Klaus Lüber

About the Postdoc-NeT-AI alumni

  • Martin Schrimpf Martin Schrimpf
  • Nooshin Ghavami Nooshin Ghavami
  • Diego Marcos Diego Marcos

Martin Schrimpf is fascinated by the interface between neuroscience and AI technology. He obtained his PhD at the prestigious MIT in Boston where he has been teaching and conducting research as a postdoc for the past four years. Regarding his future career, Schrimpf can well imagine returning to Germany.

Nooshin Ghavami is a Research Associate in the field of Deep Learning Fetal Imaging at Kings College London. She expects her future career to be in the area of teaching. As a football fan whose favourite team is Bayern München, she could well imagine working as a researcher near the Allianz Arena, at least for a while.

In his work, Diego Marcos focusses on the area of interpretable AI in AI assisted image recognition. He currently conducts research in this field as a postdoc student at Wageningen University in the Netherlands. Marcos is very familiar with Germany from his time as an Erasmus student. In summer 2020, he spent some time at the University of Tübingen where he was involved in a research project.

Are you interested in AI, and would you like to take part in a postdoc tour? Register here until 8 March 2021. 

Register here

February 2021