Numerics for AI/AI for Numerics

Coordinators: Michael Feischl, Ansgar Jüngel

The SIG “Numerics for AI/AI for Numerics” focuses on the cross fertilization of the disciplines Numerical Analysis and Artificial Intelligence. On the one hand, numerical analysis is crucially required to understand the stability, robustness, and effectiveness of machine learning systems. Moreover, mathematical methods can be used to simulate and predict hardware properties (for novel semiconductor devices and electric circuits) for certain performance goals of AI applications.

On the other hand, AI can help to solve classical problems of numerical analysis, such as the approximation of solutions of partial differential equations, particularly in high-dimensional settings as well as optimization problems.

This group is dedicated to exploring and enhancing the application of advanced numerical methods within machine learning frameworks, aiming to optimize algorithms and improve computational efficiencies. By fostering collaboration among experts in mathematics, computer science, and artificial intelligence, the group seeks to push the boundaries of research and practical applications in these interdisciplinary areas. The SIG is supported by two projects of the European Research Council “Emerging network structures and neuromorphic applications” and “New Horizons in Optimal Adaptivity” with a combined funding of 4 million Euro.