Maresa Schröder

I am a final-year computer science PhD student at LMU Munich supervised by Prof. Stefan Feuerriegel. I hold a mathematics master’s degree from Technical University of Munich and bachelor degrees in economics and mathematics from University of Mannheim.

My research interest lies in causal machine learning with a special focus on reliable treatment effect estimation, such as the development of uncertainty quantification and robust estimation methods while ensuring ethical constraints. Furthermore, I am interested in developing advanced causal methods for medical applications and strategic decision-making. Overall, my work aligns with three important goals:

1. Robustness & reliability: Researchers often face high‑dimensional covariates, structured dependencies such as social or organizational networks, and potential model misspecification. Standard causal inference approaches can fail under these conditions, leading to biased estimates that undermine data‑driven decision‑making. I leverage statistical tools, such as orthogonal learning, to develop methods for robust treatment effect estimation.

2. Uncertainty quantification: Uncertainty in causal tasks arises not only from finite data or estimation procedures but also from potential violations of causal assumptions, such as unobserved confounding. To address these challenges, I develop partial identification methods and adapt existing uncertainty methods, such as conformal prediction, to the special requirements of causal inference.

3. Ethical & fair decision-making: Treatment effect estimates and algorithmic decisions can reinforce biases or lead to unfair or harmful outcomes. My work focuses on developing frameworks that detect, quantify, and mitigate such bias. This includes considering the impact of sensitive attributes on estimates or integrating privacy constraints into algorithmic decision‑making.