13:00 - 13:30 Marco Virgolin (Life Sciences and Health), Symbolic regression & interpretable machine learning
The use of Machine Learning (ML) in high-stakes applications of, e.g., medicine and finance, requires ML models that are transparent and trustworthy. Recent years have seen a renowned interest in Symbolic Regression (SR), which is the task of learning ML models as interpretable analytical expressions, akin to physics' laws. In this talk, I wish to introduce you to (a) the type of ML problems where SR may be an interesting approach to try out, (b) what makes SR particularly challenging, and (c) recent methods and advances in the field. Hopefully, this talk will serve as an introductory overview for fellows curious to learn more about this particular subfield of interpretable ML and explainable AI.
13:30 - 14:00 Felix Lucka (Computational Imaging), Photoacoustic and Ultrasonic Tomography for Breast Imaging
New high-resolution, three-dimensional imaging techniques are being developed that probe the breast without delivering harmful radiation. In particular, photoacoustic tomography (PAT) and ultrasound tomography (UST) promise to give access to high-quality images of tissue parameters with important diagnostic value. However, the involved image reconstruction problems are very challenging from an experimental, mathematical and computational perspective. In this talk, we want to give an overview of these challenges and illustrate them with data from an ongoing clinical feasibility study that uses a prototype scanner for combined PAT and UST.
