CASUS Institute Seminar, Dr. Steve Schmerler, Helmholtz AI Consultant, Consultant Team “Matter”, Helmholtz-Zentrum Dresden-Rossendorf (HZDR)
Steve holds a PhD in theoretical solid state physics. Nowadays he is part of Helmholtz AI working at the interface between physics and machine learning. His projects cover machine learning surrogate models for complex simulations, uncertainty quantification methods to make neural network models more trustworthy as well as classical machine learning such as kernel methods and Gaussian processes.
Abstract of the talk // Kernel ridge regression (KRR) is an example of a kernel method. It represents the family of “classical” machine learning techniques (e.g. no neural networks) that is popular in application areas where data is not necessarily abundant and where one can build expressive models by using (physics-informed) engineered features. A closely related method is Gaussian process regression (GPR), whose foundations lie in Bayesian statistics and which can provide powerful uncertainty information. In this introduction, Steve will cover the mathematical foundations of those methods, shed light onto their close relation and discuss when they may be a useful parallel approach to more flexible models such as neural networks.
Steve will be in Görlitz. However, as the event is organized in a hybrid format that includes a videoconferencing tool by Zoom Inc., people interested in the topic have the chance to also join the talk remotely. Please ask for the login details via contact@casus.science.