Probabilistic Simulation Methods for Machine Learning in Science and Technology

Explore a thought-provoking lecture on Probabilistic Numerics and its applications in machine learning for science and technology. Delve into the concept that computation itself can be viewed as a form of learning from electronically produced data, blurring the line between empirical and computational information. Discover how probabilistic numerical computation enables seamless inference across dynamical systems, potentially leading to significant efficiency gains in physics-informed machine learning approaches. Challenge the traditional view of solvers for PDEs, ODEs, and DAEs as immutable code, and instead consider them as interactive, adaptive components of the machine learning tool-chain. Learn from Prof. Philipp Hennig, Chair for the Methods of Machine Learning at the University of Tübingen, as he shares insights from his extensive research in the connection between computation and inference, supported by prestigious grants and culminating in the publication of "Probabilistic Numerics — Computation as Machine Learning."