Exploring End-to-end Deep Learning Applications for Event Classification at CMS
An essential part of new physics searches at the Large Hadron Collider (LHC) at CERN involves event classification, or distinguishing potential signal events from those coming from background processes. Current machine learning techniques accomplish this using traditional hand-engineered features li...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
EDP Sciences
2019-01-01
|
| Series: | EPJ Web of Conferences |
| Online Access: | https://www.epj-conferences.org/articles/epjconf/pdf/2019/19/epjconf_chep2018_06031.pdf |
| _version_ | 1818364713197633536 |
|---|---|
| author | Andrews Michael Paulini Manfred Gleyzer Sergei Poczos Barnabas |
| author_facet | Andrews Michael Paulini Manfred Gleyzer Sergei Poczos Barnabas |
| author_sort | Andrews Michael |
| collection | DOAJ |
| description | An essential part of new physics searches at the Large Hadron Collider (LHC) at CERN involves event classification, or distinguishing potential signal events from those coming from background processes. Current machine learning techniques accomplish this using traditional hand-engineered features like particle 4-momenta, motivated by our understanding of particle decay phenomenology. While such techniques have proven useful for simple decays, they are highly dependent on our ability to model all aspects of the phenomenology and detector response. Meanwhile, powerful deep learning algorithms are capable of not only training on high-level features, but of performing feature extraction. In computer vision, convolutional neural networks have become the state-of-the-art for many applications. Motivated by their success, we apply deep learning algorithms to low-level detector data from the 2012 CMS Simulated Open Data to directly learn useful features, in what we call, end-to-end event classification. We demonstrate the power of this approach in the context of a physics search and offer solutions to some of the inherent challenges, such as image construction, image sparsity, combining multiple sub-detectors, and de-correlating the classifier from the search observable, among others. |
| first_indexed | 2024-12-13T22:08:44Z |
| format | Article |
| id | doaj.art-9383ceee798d4e46ab9a2a40eda3d06e |
| institution | Directory Open Access Journal |
| issn | 2100-014X |
| language | English |
| last_indexed | 2024-12-13T22:08:44Z |
| publishDate | 2019-01-01 |
| publisher | EDP Sciences |
| record_format | Article |
| series | EPJ Web of Conferences |
| spelling | doaj.art-9383ceee798d4e46ab9a2a40eda3d06e2022-12-21T23:29:45ZengEDP SciencesEPJ Web of Conferences2100-014X2019-01-012140603110.1051/epjconf/201921406031epjconf_chep2018_06031Exploring End-to-end Deep Learning Applications for Event Classification at CMSAndrews MichaelPaulini ManfredGleyzer SergeiPoczos BarnabasAn essential part of new physics searches at the Large Hadron Collider (LHC) at CERN involves event classification, or distinguishing potential signal events from those coming from background processes. Current machine learning techniques accomplish this using traditional hand-engineered features like particle 4-momenta, motivated by our understanding of particle decay phenomenology. While such techniques have proven useful for simple decays, they are highly dependent on our ability to model all aspects of the phenomenology and detector response. Meanwhile, powerful deep learning algorithms are capable of not only training on high-level features, but of performing feature extraction. In computer vision, convolutional neural networks have become the state-of-the-art for many applications. Motivated by their success, we apply deep learning algorithms to low-level detector data from the 2012 CMS Simulated Open Data to directly learn useful features, in what we call, end-to-end event classification. We demonstrate the power of this approach in the context of a physics search and offer solutions to some of the inherent challenges, such as image construction, image sparsity, combining multiple sub-detectors, and de-correlating the classifier from the search observable, among others.https://www.epj-conferences.org/articles/epjconf/pdf/2019/19/epjconf_chep2018_06031.pdf |
| spellingShingle | Andrews Michael Paulini Manfred Gleyzer Sergei Poczos Barnabas Exploring End-to-end Deep Learning Applications for Event Classification at CMS EPJ Web of Conferences |
| title | Exploring End-to-end Deep Learning Applications for Event Classification at CMS |
| title_full | Exploring End-to-end Deep Learning Applications for Event Classification at CMS |
| title_fullStr | Exploring End-to-end Deep Learning Applications for Event Classification at CMS |
| title_full_unstemmed | Exploring End-to-end Deep Learning Applications for Event Classification at CMS |
| title_short | Exploring End-to-end Deep Learning Applications for Event Classification at CMS |
| title_sort | exploring end to end deep learning applications for event classification at cms |
| url | https://www.epj-conferences.org/articles/epjconf/pdf/2019/19/epjconf_chep2018_06031.pdf |
| work_keys_str_mv | AT andrewsmichael exploringendtoenddeeplearningapplicationsforeventclassificationatcms AT paulinimanfred exploringendtoenddeeplearningapplicationsforeventclassificationatcms AT gleyzersergei exploringendtoenddeeplearningapplicationsforeventclassificationatcms AT poczosbarnabas exploringendtoenddeeplearningapplicationsforeventclassificationatcms |