Springer : Machine learning in high energy physics: a review of heavy-flavor jet tagging at the LHC

Mondal, Spandan; Mastrolorenzo, Luca

doi:10.1140/epjs/s11734-024-01234-y

Article
Report number	arXiv:2404.01071
Title	Machine learning in high energy physics: a review of heavy-flavor jet tagging at the LHC
Author(s)	Mondal, Spandan (Brown U.) ; Mastrolorenzo, Luca (CERN)
Publication	2024-07-19
Imprint	2024-04-01
Number of pages	38
Note	38 pages, 2 tables, 1 figure
In:	Eur. Phys. J. Spec. Top. 233, 15-16 (2024) pp.2657-2686
DOI	10.1140/epjs/s11734-024-01234-y (publication) 10.1140/epjs/s11734-024-01234-y (publication)
Subject category	physics.data-an ; Other Fields of Physics ; hep-ex ; Particle Physics - Experiment
Accelerator/Facility, Experiment	CERN LHC
Abstract	The application of machine learning (ML) in high energy physics (HEP), specifically in heavy-flavor jet tagging at Large Hadron Collider (LHC) experiments, has experienced remarkable growth and innovation in the past decade. This review provides a detailed examination of current and past ML techniques in this domain. It starts by exploring various data representation methods and ML architectures, encompassing traditional ML algorithms and advanced deep learning techniques. Subsequent sections discuss specific instances of successful ML applications in jet flavor tagging in the ATLAS and CMS experiments at the LHC, ranging from basic fully-connected layers to graph neural networks employing attention mechanisms. To systematically categorize the advancements over the LHC's three runs, the paper classifies jet tagging algorithms into three generations, each characterized by specific data representation techniques and ML architectures. This classification aims to provide an overview of the chronological evolution in this field. Finally, a brief discussion about anticipated future developments and potential research directions in the field is presented.
Copyright/License	publication: © 2024 The Author(s) (License: exclusive licence to EDP Sciences, Springer-Verlag GmbH Germany, part of Springer Nature) preprint: (License: CC BY 4.0)

$Approximate light-flavor jet rejection rates at a b jet efficiency of 70\% (y-axis) achieved by different ML-based single-pronged jet flavor tagging algorithms used in the CMS and ATLAS experiments. The x-axis shows the approximate years the algorithms were published by the respective experiments. The vertical color bands indicate the timespan of the runs at the LHC, while the different markers indicate different datasets reconstructed with different run conditions on which the taggers were trained and/or evaluated. Run-2 data reconstruction in CMS is split into two parts corresponding to before (Phase-0) and after the CMS Phase-1 pixel detector upgrade \citep{CMSTrackerGroup:2020edz}. It is important to note that (a) the CMS and ATLAS detectors, along with their subsystems, possess different detection and reconstruction capabilities, (b) there are differences in jet reconstruction techniques and quality criteria between the two experiments, (c) these techniques and criteria evolve over time, and (d) definitions of tagger scores, and hence the degree of discrimination specifically against light-flavor jets, vary between experiments (e.g. using tagger output scores versus employing transformations like that in eq. \ref{eq:DL1b}). Consequently, directly comparing tagger performance across experiments may not accurately reflect algorithm capabilities and must be interpreted carefully.$

Corresponding record in: Inspire

Back to search

Record created 2024-12-11, last modified 2024-12-12

Registos similares

Fulltext:

PDF

Add to personal basket
Export as BibTeX, MARC, MARCXML, DC, EndNote, NLM, RefWorks

CERN Document Server

Access articles, reports and multimedia content in HEP

Main menu

CERN Accelerating science