13ª Semana Integrada do Instituto de Física de São Carlos

Separation of iron and proton induced air showers using convolutional neural networks

21 de ago. de 2023 16:00

1h 30m

Salão de Eventos USP

Normal 16h00 - 17h30

Descrição

The Cherenkov Telescope Array (CTA) will be the next-generation gamma-ray observatory offering better flux sensitivity than the current generation. (1) Each telescope will detect the Cherenkov emission from air showers initiated by gamma and cosmic rays. State-of-the-art deep learning algorithms, such as convolutional neural networks (CNNs), are used to reconstruct the energy and classify the primary particle. (2-3) CNNs are commonly used for image classification because they extract and learn from the features and patterns within an image. In this work, we proposed an architecture to separate iron and proton induced showers using the image recorded in the camera (Img), the depth of the shower maximum (Xmax), and the particle energy (E) associated with the air shower. We tested the CNN performance with three different inputs: (i) Img + E, (ii) Img + Xmax, and (iii) Img + Xmax + E. In addition, we used reconstructed and true values (for the Xmax and E) during the training and the prediction. This architecture takes advantage of the image patterns and the air showers' mass-sensitive parameters. Therefore, we aim to obtain significant efficiency in identifying iron and proton events.

Referências

1 ACHARYA, B. S. et al. Science with the Cherenkov telescope array. Disponível em: https://arxiv.org/pdf/1709.07997.pdf. Acesso em: 23 jan. 2023.

2 NIETO, D. et al. Reconstruction of IACT events using deep learning techniques with CTLearn. Astronomical Society of the Pacific Conference Series, v. 532, p. 191-194, July 2022.

3 ASCHERSLEBEN, J. et al. Application of pattern spectra and convolutional neural networks to the analysis of simulated Cherenkov Telescope Array Data. Proceedigns of Science, v. 395, p. 697-1-697-14, Mar. 2022. DOI: https://dx.doi.org/10.22323/1.395.0697.