Sebastian Nowozin
Distinguished Scientist, Google DeepMind, London
PhD Student, Empirical Inference (Bernhard Schölkopf), Max-Planck-Institute for Intelligent Systems
What is your connection to Tübingen, Sebastian?
I joined the MPI in Tübingen in the summer of 2006 as PhD student and graduated in fall 2009. Since then I have benefitted not just from the knowledge gained and research done while there but also from the network of people I overlapped with in these three years.
Is there a memory from your time here that had a lasting positive impact on you?
During my whole time at the institute in Tubingen people were openly discussing new and sometimes fragile ideas. When I joined I liked to join the discussions but my mind would often try to prove itself in finding drawbacks or flaws in some idea. In my second months after joining, after one such discussion where I criticised another person's idea my supervisor took me aside and gave me some key advice of the form “Here we usually try to support each other, not shut each others’ ideas down”. That really hit a nerve and stuck with me; it changed both how I saw myself and how I interacted with other researchers with a lasting positive impact for myself and others.
I fondly remember the wide interdisciplinary discussions at Tübingen as a model of how exciting good research can be and have tried to instill the same excitement in my own research and the people I work with.
Sebastian Nowozin
Where has your career taken you since, and did your time in Tübingen play a role? If so, how?
When I graduated in 2009, I think it is fair to say that in many domains machine learning was not working well yet; computer vision was the area I worked in at the time and successful machine learning applications were in their infancy. But the main ideas that would make it work over the next decade were all there already and I was fortunate enough to observe this up close. I first joined Microsoft Research in Cambridge in England and continued with computer vision research over the next decade. The principles of good empirical research learned in Tübingen have been instrumental in my whole career. I also fondly remember the wide interdisciplinary discussions at Tübingen as a model of how exciting good research can be and have tried to instill the same excitement in my own research and the people I work with. I have been with Google DeepMind since 2022, focusing on applying AI to the natural sciences.