The application of weakly supervised learning in the search for heavy resonances at the LHC
| dc.contributor.author | Choma, Nalamotse Joshua | |
| dc.contributor.co-supervisor | Ruan, Xifeng | |
| dc.contributor.supervisor | Mellado, Bruce | |
| dc.date.accessioned | 2024-10-12T18:18:27Z | |
| dc.date.available | 2024-10-12T18:18:27Z | |
| dc.date.issued | 2023-06 | |
| dc.description | A Thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy, to the Faculty of Science, School of Physics, University of the Witwatersrand, Johannesburg, 2023. | |
| dc.description.abstract | The discovery of the Higgs boson at the Large Hadron Collider by the ATLAS and CMS experiments has made the search for new physics beyond the Standard Model a priority in the field of High Energy Particle Physics. New resonances have yet to be discovered using inclusive and model-dependent searches, which means they may be driven by subtle topologies. Rapid improvements in Machine Learning techniques have led to their increasing application in High Energy Particle physics. Unlike supervised learning, which is known to assume full knowledge of the underlying model, semi-supervised learning, in particular weakly supervised learning, allows the extraction of new information from data with partial knowledge. The goal of this study is to set up searches for heavy resonances at the electroweak scale with topological requirements performed in both inclusive and exclusive regions of phase-space tailored to a particular production mode. These resonances could be generated with different production mechanisms. In this work, we describe search procedures based on weakly supervised learning applied to mixed samples and used to search for resonances with little or no prior knowledge of the production mechanism. This approach has the advantage that sidebands or control regions can be used to effectively model backgrounds without relying on models. The effectiveness of this method is measured by the production of the Standard Model Higgs boson, which decays into a pair of photons in both inclusive and exclusive regions of phase-space at the LHC. Having confirmed the ability of the method to extract various Standard Model Higgs boson signal processes, the search for new phenomena in high mass final states will be set up at the LHC. Subsequently, the approach is used in the search for new resonances in the Zγ final state with Z → e +e − or Z → µ +µ −, using the Monte Carlo simulated signal samples for 139 fb−1 of integrated luminosity for Run 2 collected at the LHC. The weakly supervised learning approach is implemented and compared to the performance of the fully supervised approach, which is then used to calculate the production limit for Higgs-like particles for Zγ where the significance of the signal is maximal. | |
| dc.description.sponsorship | SA-CERN consortium. | |
| dc.description.sponsorship | National Institute for Theoretical and Computational Sciences. | |
| dc.description.submitter | MM2024 | |
| dc.faculty | Faculty of Science | |
| dc.identifier | 0000-0001-8848-211X | |
| dc.identifier.citation | Choma, Nalamotse Joshua. (2023). The application of weakly supervised learning in the search for heavy resonances at the LHC. [PhD thesis, University of the Witwatersrand, Johannesburg]. https://hdl.handle.net/10539/41532 | |
| dc.identifier.uri | https://hdl.handle.net/10539/41532 | |
| dc.language.iso | en | |
| dc.publisher | University of the Witwatersrand, Johannesburg | |
| dc.rights | ©2023 University of the Witwatersrand, Johannesburg. All rights reserved. The copyright in this work vests in the University of the Witwatersrand, Johannesburg. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of University of the Witwatersrand, Johannesburg. | |
| dc.rights.holder | University of the Witwatersrand, Johannesburg | |
| dc.school | School of Physics | |
| dc.subject | Weakly supervised learning | |
| dc.subject | Standard Model | |
| dc.subject | Fully supervised learning | |
| dc.subject | Resonances | |
| dc.subject | Topologies | |
| dc.subject | UCTD | |
| dc.subject.other | SDG-15: Life on land | |
| dc.title | The application of weakly supervised learning in the search for heavy resonances at the LHC | |
| dc.type | Thesis |