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

AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz

AKPIK 2: Data Analytics & Machine Learning

AKPIK 2.8: Talk

Wednesday, March 23, 2022, 18:00–18:15, AKPIK-H13

Deep Learning Accelerated Maximum Likelihood Reconstruction of IACT Events — •Noah Biederbeck and Maximilian Nöthe for the CTA Consortium — Astroparticle Physics WG Elsässer, TU Dortmund University, Germany

The Cherenkov Telescope Array will be the next generation ground-based gamma-ray observatory, consisting of tens of Imaging Atmospheric Cherenkov Telescopes (IACTs) at two sites once its construction is finished.

In this talk we present a deep learning accelerated maximum likelihood reconstruction of gamma-ray events. A generative neural network predicts IACT camera images from a set of physical event parameters. These generated images are then compared to Monte Carlo simulated event images using a Poissonian likelihood loss in order to reconstruct the event properties, e.g. the energy of the primary particle and its direction.

First results on simulated single-telescope events will be presented and extensions to predictions of array events will be outlined.