Szczegóły publikacji
Opis bibliograficzny
Machine learning approach for detecting potential anomalous cosmic rays particles tracks in Earth-scale cosmic ray extremely distributed observatory / Jan Tyc, Marcin Zub, Tomasz HACHAJ // W: Parallel Processing and Applied Mathematics : 15th International Conference, PPAM 2024 : Ostrava, Czech Republic, September 8–11, 2024 : revised selected papers, Pt. 3 / eds. Roman Wyrzykowski, [et al.]. — Cham : Springer, cop. 2025. — (Lecture Notes in Computer Science ; ISSN 0302-9743 ; 15581). — ISBN: 978-3-031-85702-7; e-ISBN: 978-3-031-85703-4. — S. 302–315. — Bibliogr., Abstr. — Publikacja dostępna online od: 2025-04-01
Autorzy (3)
Dane bibliometryczne
| ID BaDAP | 159436 |
|---|---|
| Data dodania do BaDAP | 2025-04-28 |
| DOI | 10.1007/978-3-031-85703-4_21 |
| Rok publikacji | 2025 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Springer |
| Konferencja | International Conference on Parallel Processing and Applied Mathematics 2024 |
| Czasopismo/seria | Lecture Notes in Computer Science |
Abstract
In this paper, we present a novel contribution to the problem of the detection of potential anomalous cosmic rays particle tracks acquired by Earth-scale complex distributed observatory infrastructure. Detection of such signals provides an opportunity to analyze the performance of the entire research infrastructure of a complex system (for example, potential failures) and a chance to record unknown physical phenomena that can be recorded simultaneously with space radiation observations. The observatory comprises a worldwide network of CMOS detectors cooperating in the citizen science paradigm. We propose, evaluate, and compare several data processing, feature extractions, and potential anomaly discovery pipelines based on machine learning approaches. Although, to date, there have already been published works performing detection and analysis of anomalies in the Cosmic Ray Extremely Distributed Observatory Project (CREDO) dataset, they applied single methods without comparing the results with other possible approaches to analyze potential anomalies. In this work, for the first time, we have used and discussed many anomalous discovery approaches not previously applied to this issue to perform analysis of this extensive dataset (nearly 600,000 images). For this reason, the results published here are of significant cognitive importance to the problem under consideration. We have published both our research’s source codes and dataset so our results can be reproduced.
