Application of deep learning methods to analysis of imaging atmospheric Cherenkov telescopes data

• Published in: Astroparticle physics Vol. 105; pp. 44 - 53
• Main Authors: Shilon, I., Kraus, M., Büchele, M., Egberts, K., Fischer, T., Holch, T.L., Lohse, T., Schwanke, U., Steppa, C., Funk, S.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.02.2019
• Subjects: Analysis technique; Convolutional neural networks; Deep learning; Gamma-ray astronomy; IACT; Recurrent neural networks; Deep learning; Analysis technique; Recurrent neural networks; IACT; Convolutional neural networks; Gamma-ray astronomy
• ISSN: 0927-6505; 1873-2852
• DOI: 10.1016/j.astropartphys.2018.10.003

Ground based γ-ray observations with Imaging Atmospheric Cherenkov Telescopes (IACTs) play a significant role in the discovery of very high energy (E > 100 GeV) γ-ray emitters. The analysis of IACT data demands a highly efficient background rejection technique, as well as methods to accurately determine the position of its source in the sky and the energy of the recorded γ-ray. We present results for background rejection and signal direction reconstruction from first studies of a novel data analysis scheme for IACT measurements. The new analysis is based on a set of Convolutional Neural Networks (CNNs) applied to images from the four H.E.S.S. phase-I telescopes. As the H.E.S.S. cameras pixels are arranged in a hexagonal array, we demonstrate two ways to use such image data to train CNNs: by resampling the images to a square grid and by applying modified convolution kernels that conserve the hexagonal grid properties. The networks were trained on sets of Monte-Carlo simulated events and tested on both simulations and measured data from the H.E.S.S. array. A comparison between the CNN analysis to current state-of-the-art algorithms reveals a clear improvement in background rejection performance. When applied to H.E.S.S. observation data, the CNN direction reconstruction performs at a similar level as traditional methods. These results serve as a proof-of-concept for the application of CNNs to the analysis of events recorded by IACTs.