Anomaly detection in high-energy physics using machine learning: A case study with DELPHI Simulated Higgs Events
1 Department of Physics, P M college of Excellence, Gyanchand Shrivastav PG College, Damoh (M.P.) 470661 India.
2 Department of Physics, Government College, Banda, Sagar (M. P.) 470335 India.
3 Eklavya University, Sagar Road, Damoh (M.P.) India 470672.
Research Article
International Journal of Engineering Research Updates, 2025, 08(02), 001-009.
Article DOI: 10.53430/ijeru.2025.8.2.0033
Publication history:
Received on 05 September 2025; revised on 15 October 2025; accepted on 17 October 2025
Abstract:
Anomaly detection in high-energy physics (HEP) is critical for uncovering rare or novel physical phenomena beyond the Standard Model. This study explores machine learning techniques for identifying anomalous events in simulated Higgs boson data produced by the DELPHI experiment at CERN’s LEP collider. The data, originally stored in ZEBRA format, is preprocessed and converted into a machine-learning-compatible structure. We apply unsupervised and semi-supervised learning approaches, including autoencoders and isolation forests, to uncover rare signal-like events amidst large backgrounds. Our findings, based on dummy metrics, demonstrate that autoencoders can achieve an AUC of 0.86, with up to 40% of signal events detected using unsupervised methods. These results underscore the utility of ML tools in analyzing archival particle physics datasets.
Keywords:
High Energy Physics (HEP); Novel particle interactions; Machine Learning (ML); Deep Learning ; Anomaly detection
Full text article in PDF:
Copyright information:
Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0