Standard

Search for optimal deep neural network architecture for gamma ray search at KASCADE. / Kuznetsov, Mikhail; Petrov, Nikita; Plokhikh, Ivan et al.

In: Proceedings of Science, Vol. 423, 091, 14.12.2023.

Research output: Contribution to journal › Conference article › peer-review

Harvard

Kuznetsov, M, Petrov, N, Plokhikh, I, Sotnikov, V & Tsobenko, M 2023, 'Search for optimal deep neural network architecture for gamma ray search at KASCADE', Proceedings of Science, vol. 423, 091. https://doi.org/10.22323/1.423.0091

APA

Kuznetsov, M., Petrov, N., Plokhikh, I., Sotnikov, V., & Tsobenko, M. (2023). Search for optimal deep neural network architecture for gamma ray search at KASCADE. Proceedings of Science, 423, [091]. https://doi.org/10.22323/1.423.0091

Vancouver

Kuznetsov M, Petrov N, Plokhikh I, Sotnikov V, Tsobenko M. Search for optimal deep neural network architecture for gamma ray search at KASCADE. Proceedings of Science. 2023 Dec 14;423:091. doi: 10.22323/1.423.0091

Author

Kuznetsov, Mikhail ; Petrov, Nikita ; Plokhikh, Ivan et al. / Search for optimal deep neural network architecture for gamma ray search at KASCADE. In: Proceedings of Science. 2023 ; Vol. 423.

BibTeX

@article{c1a08478a0a1429483a0a60f9ec75af0,
title = "Search for optimal deep neural network architecture for gamma ray search at KASCADE",
abstract = "We present the first steps of a search for high-energy (> 1 PeV) gamma rays in archival data of the KASCADE experiment. With the data collected from 1996 to 2013 the KASCADE statistics is comparable with that of modern observatories. The data is provided by the KASCADE Cosmic ray Data Center (KCDC) and publicly available. We employ methods of machine learning to distinguish between air showers produced by hadronic and gamma-ray primaries. For that we design primary particle type classifiers and train them with the KASCADE Monte-Carlo simulations. We compare results of several deep learning methods: a graph neural network, a self-attention network and a compact convolutional transformer. The level of hadronic background suppression with respect to gamma-ray signal in the best of these methods exceeds that of the original KASCADE method by more than an order of magnitude.",
author = "Mikhail Kuznetsov and Nikita Petrov and Ivan Plokhikh and Vladimir Sotnikov and Margarita Tsobenko",
note = "The work is supported by the Russian Science Foundation grant 22-22-00883. We appreciate the contribution on the initial stages of this project made by our colleagues: Dmitry Kostunin, Vladimir Lenok and Victoria Tokareva.; 27th European Cosmic Ray Symposium ; Conference date: 25-07-2022 Through 29-07-2022",
year = "2023",
month = dec,
day = "14",
doi = "10.22323/1.423.0091",
language = "English",
volume = "423",
journal = "Proceedings of Science",
issn = "1824-8039",
publisher = "Sissa Medialab Srl",

}

RIS

TY - JOUR

T1 - Search for optimal deep neural network architecture for gamma ray search at KASCADE

AU - Kuznetsov, Mikhail

AU - Petrov, Nikita

AU - Plokhikh, Ivan

AU - Sotnikov, Vladimir

AU - Tsobenko, Margarita

N1 - Conference code: 27

PY - 2023/12/14

Y1 - 2023/12/14

N2 - We present the first steps of a search for high-energy (> 1 PeV) gamma rays in archival data of the KASCADE experiment. With the data collected from 1996 to 2013 the KASCADE statistics is comparable with that of modern observatories. The data is provided by the KASCADE Cosmic ray Data Center (KCDC) and publicly available. We employ methods of machine learning to distinguish between air showers produced by hadronic and gamma-ray primaries. For that we design primary particle type classifiers and train them with the KASCADE Monte-Carlo simulations. We compare results of several deep learning methods: a graph neural network, a self-attention network and a compact convolutional transformer. The level of hadronic background suppression with respect to gamma-ray signal in the best of these methods exceeds that of the original KASCADE method by more than an order of magnitude.

AB - We present the first steps of a search for high-energy (> 1 PeV) gamma rays in archival data of the KASCADE experiment. With the data collected from 1996 to 2013 the KASCADE statistics is comparable with that of modern observatories. The data is provided by the KASCADE Cosmic ray Data Center (KCDC) and publicly available. We employ methods of machine learning to distinguish between air showers produced by hadronic and gamma-ray primaries. For that we design primary particle type classifiers and train them with the KASCADE Monte-Carlo simulations. We compare results of several deep learning methods: a graph neural network, a self-attention network and a compact convolutional transformer. The level of hadronic background suppression with respect to gamma-ray signal in the best of these methods exceeds that of the original KASCADE method by more than an order of magnitude.

UR - https://www.scopus.com/record/display.uri?eid=2-s2.0-85181089625&origin=inward&txGid=4e322cbae4ba1c621f5724079018fb56

UR - https://inspirehep.net/literature/2634437

UR - https://www.mendeley.com/catalogue/9145bdd3-fb88-3740-90b3-72a3c82d345a/

U2 - 10.22323/1.423.0091

DO - 10.22323/1.423.0091

M3 - Conference article

VL - 423

JO - Proceedings of Science

JF - Proceedings of Science

SN - 1824-8039

M1 - 091

T2 - 27th European Cosmic Ray Symposium

Y2 - 25 July 2022 through 29 July 2022

ER -

ID: 59538832