Hometown: Herne, Germany
Educational and professional background: PhD student at Michigan State University, 2012-2017
How did you get into your field of research?
It all started with a class in statistical pattern classification and Bayesian learning. While the class introduced me to pattern classification on a more theoretical level, I could immediately see how the concepts learned in this class could map onto my research problems at that time.
I could think of an endless amount of problems that could be tackled with machine learning, which I found fascinating. And while I am personally more focused on the method development aspect, are successfully being used in several collaborations, which is very satisfying.
What attracted you to UW-Madison?
What I find particularly attractive is that UW-Madison is probably the best public university in the country when it comes to excellence in research and teaching. A large campus implies that there is a lot of diversity, regarding scientific and cultural backgrounds, which means that it’s a very stimulating atmosphere and the best environment for personal growth. Based on my many collaborations, you can probably tell that I like to work with people from many different backgrounds. Also, as an author of a bestselling textbook on machine learning, I am not only passionate about my research but also enjoy teaching.
What was your first visit to campus like?
I actually never visited Madison before I interviewed for this position and honestly didn’t expect too much from the campus geographically or infrastructure-wise. But I can remember that I was positively surprised about how bike-friendly this place was and that there is not only a beautiful downtown area nearby but also lots of nature (the various trails around the lakes) in such close proximity. While UW-Madison looked like a great place for me “on paper,” the visit showed me that it’s also a great environment, which I would enjoy — and having lived in Michigan for six years before, I am used to cold winters.
What’s one thing you hope students who take a class with you will come away with?
While it is impossible to cover all knowledge of a field in one single class, I want my students to walk away with the confidence that they now know enough to do something useful with the things learned in class — that next to absorbing knowledge, applications not only aid the learning process but are also a crucial aspect in becoming creative problem-solvers.
Do you feel your work relates in any way to the Wisconsin Idea? If so, please describe how.
My research is centered around the development of methods related to data science and machine learning; however, one important aspect in pursuing a research goal is to think about how these methods can be applied to real-world problem-solving. For example, funded by the Great Lakes Fishery Commission, a computational framework for hypothesis-driven inhibitor discovery I developed in the past led to the discovery of a powerful sea lamprey pheromone (3kPZS) inhibitor called 3PZs, which has then successfully been tested by our collaborators as a means for invasive species control in the Great Lakes, to restore and protect the native ecology and fishery.
What’s something interesting about your area of expertise you can share that will make us sound smarter at parties?
Since news media try so hard to make us believe the terms of artificial intelligence and machine learning are the same, it may be interesting to point out they are actually not. Machine learning is about teaching computers to learn from data. AI is concerned with teaching computers to solve tasks that humans are particularly good at (e.g., question answering, playing games, identifying objects in images or driving a car). Now, to “implement” AI, we can write computer programs by hand, e.g., think of set of “if-else” rules. However, composing a comprehensive set of rules is extremely tedious and for most tasks not feasible. Hence, the field of machine learning was developed as an alternative to handcrafted rules — it’s about letting computers learn by themselves from observing examples, instead. Machine learning has now grown into its own field, and while it does overlap with AI indeed, it can be considered more as a set of techniques that is used in the attempt to develop AI, and there are many different applications of machine learning that I wouldn’t call AI-related, and many machine learning have already been around for decades, for example check-parsing ATMs, passport scanners at international airports, email spam filters or internet search engines.
Long distance running, programming and participating in the open-source community.