Core scientific interests
In my current research, I am interested in problems at the intersection of machine learning and domain sciences, particularly astrophysics. More specifically, I am curious to investigate how we can build ML models that make use of existing scientific domain knowledge about a given problem, such as symmetries, conserved quantities, or other known properties of the data which constrain the space of possible solutions. I hope that—under the right conditions—this may yield smaller, more interpretable models that need fewer data and are more robust towards domain shift. In my work, I aim to understand better how including domain knowledge can work in practice (for specific problems), the circumstances in which this is beneficial, and how we can adequately characterize and compare the resulting models.
Current research projects
In my current main project, we investigate how we can make use of the framework of half-sibling regression (HSR) and apply it to the problem of high-contrast imaging (HCI; also called direct imaging) of extrasolar planets. We are particularly interested to see how the HSR framework allows us to incorporate our prior domain knowledge about the problem and potentially also include additional information (such as the observing conditions) into the post-processing of HCI data.
Other academic interests
Besides my core research, I am also passionate about reproducibility in science. For instance, I am a strong advocate for making your research code publicly available. Maybe as a consequence of this, I also have a big interest in proper software development and writing good code. Finally I also really enjoy public outreach and science communication; for example, I have already successfully presented my research in multiple science slams.