ľ�ϸ���Ӱ��

Sjoerd Dirksen has been appointed Professor of Mathematics for Data Science at Utrecht ľ�ϸ���Ӱ��. His research focuses on high-dimensional probability theory and the mathematics underpinning data science. He searches for structure in a world with ever-increasing amounts of data.

How can we make weather forecasts more accurate? Can we make algorithms more efficient? And how do you find the proverbial needle in a digital haystack? Dirksen explains that mathematics often holds the answer to these questions.

Massive amounts of information

Dirksen applies mathematical methods to tackle concrete problems. A central question in his research is how to reduce massive amounts of data into something workable, without losing essential information. "Imagine a database with millions of records of plant species, for example," he says. "The more characteristics you have of each species, such as genetic information, the harder it becomes to quickly compare data."

Smart summaries

Mathematics can speed up that process. Dirksen: “The solution often lies in what we call dimensionality reduction: compressing data without losing its underlying structure.” To achieve this, he uses random matrices, mathematical constructions that are random in nature, but yield surprisingly reliable results in practice. “You create a new representation of your data. One that is more compact but still preserves the most important properties,” he explains. This technique makes it easier to quickly determine which plant species in a database most closely resembles a newly entered species.

Understanding AI

The chair in Mathematics for Data Science was established to explore exactly these kinds of fundamental mathematical insights and connect them to modern applications such as deep learning and artificial intelligence. “Many deep learning algorithms work well in practice, but we don’t yet fully understand why,” Dirksen notes. “We’re studying if and when these algorithms generalise, how robust they are, and whether they learn optimally.” His work touches on a central question in the AI debate: how transparent and reliable are the systems increasingly shaping our decisions? 

Many deep learning algorithms work well in practice, but we don’t yet fully understand why

More accurate weather forecasts

Other computer models, too, rely heavily on mathematics. Take weather forecasting, for example. Dirksen collaborates with the Royal Netherlands Meteorological Institute (KNMI), where he and his students work to improve forecast accuracy using statistical postprocessing. “Weather systems are chaotic: tiny disturbances can have big consequences. This method helps us better quantify forecast uncertainty, and in some cases even reduce it.”

From energy companies to forensic research

As a professor, Dirksen remains involved in teaching. He lectures first-year students in probability theory, developed a machine learning course for mathematics students, and now contributes to a new master’s course on high-dimensional probability. He also co-coordinates Utrecht ľ�ϸ���Ӱ��’s interdisciplinary project AI in Higher Education, which explores both the risks and opportunities of generative AI in teaching and learning.

Dirksen supervises bachelor’s and master’s students in internships and thesis projects. “What I find remarkable,” he says, “is how many different fields our students end up in, from energy companies to forensic research. With a strong mathematical foundation, you can go almost anywhere, I believe.”

Grades matter, but they don’t tell the whole story. What you actually learn and how you think, that’s what counts.

More than grades

Dirksen observes that students often focus heavily on results and grades, sometimes more than on genuine learning. “Grades matter, of course, but they don’t tell the whole story. They’re snapshots. What you actually learn and how you think, that’s what counts.” He encourages students to think beyond the obvious. “At Utrecht, there’s plenty of room to take courses outside your programme. Make use of that. It broadens your perspective and teaches you new ways of thinking.”