My research focuses on data-efficient learning under uncertainty, including work on reinforcement learning, active learning and Bayesian deep learning. I am also interested in applying machine learning to problems in science and engineering.
My work revolves around two common themes:
Reasoning about uncertainty: I believe that obtaining accurate uncertainty estimates is crucial for efficient data acquisition and exploration. Therefore, I use the language of probability and Bayesian inference to express and reason about uncertainty.
Exploiting prior knowledge: Efficient learning requires to equip models with the right inductive biases. In particular, I am interested in exploiting prior knowledge about the symmetries of the underlying problem, such as translation- or rotation-equivariance.