SMuK 2023 – scientific programme

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

T: Fachverband Teilchenphysik

T 12: Gamma Astronomy I

T 12.4: Talk

Monday, March 20, 2023, 17:15–17:30, POT/0151

Deep-learning-based gamma/hadron separation for IACTs — •Jonas Glombitza, Vikas Joshi, Benedetta Bruno, and Stefan Funk for the H.E.S.S. collaboration — Erlangen Centre for Astroparticle Physics, Erlangen, Germany

Ground-based gamma-ray observatories have opened in the last decades a new window to the non-thermal universe by studying air showers initiated by cosmic particles. Imaging Air Cherenkov Telescopes (IACTs), like the High Energy Stereoscopic System (H.E.S.S.), are utilized to image the distribution of Cherenkov light emitted during the development of air showers. For the rejection of the hadronic background, many algorithms rely on a high-level parameterization of these IACT images and exploit their correlation. Recently, deep-learning-based approaches showed promising results by explolting the full images, which overcomes the limitation of the elliptical modeling.

In this contribution, we present a new approach to reconstruct IACT images using deep learning. We model the images as a collection of triggered sensors that can be described by a graph and analyzed using graph convolutional neural networks. We describe our new algorithm, trained using H.E.S.S. simulations, examine its performance, and compare it to various classification algorithms.