LCSR Seminar: Andreas Maier

Description

Andreas Maier, a professor and head of the Pattern Recognition Lab at the University of Erlangen-Nuremberg in Germany, will give a talk titled "Known Operator Learning–An Approach to Unite Machine Learning, Signal Processing, and Physics" for the Laboratory for Computational Sensing and Robotics.

Please attend the event by using the Zoom link.

Abstract:

We describe an approach for incorporating prior knowledge into machine learning algorithms. We aim at applications in physics and signal processing in which we know that certain operations must be embedded into the algorithm. Any operation that allows computation of a gradient or sub-gradient towards its inputs is suited for our framework. We derive a maximal error bound for deep nets that demonstrates that inclusion of prior knowledge results in its reduction. Furthermore, we show experimentally that known operators reduce the number of free parameters. We apply this approach to various tasks ranging from computed tomography image reconstruction over vessel segmentation to the derivation of previously unknown imaging algorithms. As such, the concept is widely applicable for many researchers in physics, imaging and signal processing. We assume that our analysis will support further investigation of known operators in other fields of physics, imaging and signal processing.