SMuK 2023 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz

AKPIK 2: Applications in Particle and Astroparticle Physics

AKPIK 2.5: Talk

Tuesday, March 21, 2023, 18:00–18:15, ZEU/0118

Machine Learning based defect detection for large-scale electrodes — •Sebastian Vetter — Karlsruhe Institute of Technology, Institute for Astroparticle Physics

Like every piece of hardware produced in an industrial setting, detectors in physics experiments are subject to material defects, introduced during the production or handling of individual components. This can greatly influence detector behavior and lead to unexpected experimental results, depending on the affected part and the extent of the defect. Detection and quantification of such defects is therefore an important step in constructing a successful experiment.

It is still quite common for defect inspection to be done by eye. However, recent developments in computer-based inspection methods provide the opportunity to relieve humans from this tedious task, to remove the susceptibility of human error from the inspection step, and to objectively quantify the extent of detected defects.

In this talk, I present the defect inspection of a large-scale electrode mesh, as used for example in liquid noble gas Dark Matter experiments. This inspection was carried out first by hand and then compared to various Machine Learning approaches, ranging from simple decision trees to variational autoencoders.