{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,9,25]],"date-time":"2025-09-25T18:07:18Z","timestamp":1758823638348,"version":"3.37.3"},"reference-count":38,"publisher":"IOP Publishing","issue":"2","license":[{"start":{"date-parts":[[2023,4,27]],"date-time":"2023-04-27T00:00:00Z","timestamp":1682553600000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2023,4,27]],"date-time":"2023-04-27T00:00:00Z","timestamp":1682553600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2023,6,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>To improve the performance-critical stability and brightness of the electron bunch at injection into the proton-driven plasma wakefield at the AWAKE CERN experiment, automation approaches based on unsupervised machine learning (ML) were developed and deployed. Numerical optimisers were tested together with different model-free reinforcement learning (RL) agents. In order to avoid any bias, RL agents have been trained also using a completely unsupervised state encoding using auto-encoders. To aid hyper-parameter selection, a full synthetic model of the beamline was constructed using a variational auto-encoder trained to generate surrogate data from equipment settings. This paper describes the novel approaches based on deep learning and RL to aid the automatic setup of a low energy line, as the one used to deliver beam to the AWAKE facility. The results obtained with the different ML approaches, including automatic unsupervised feature extraction from images using computer vision are presented. The prospects for operational deployment and wider applicability are discussed.<\/jats:p>","DOI":"10.1088\/2632-2153\/acce21","type":"journal-article","created":{"date-parts":[[2023,4,18]],"date-time":"2023-04-18T22:37:50Z","timestamp":1681857470000},"page":"025016","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":5,"title":["Towards automatic setup of 18 MeV electron beamline using machine learning"],"prefix":"10.1088","volume":"4","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7815-6011","authenticated-orcid":true,"given":"Francesco Maria","family":"Velotti","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Brennan","family":"Goddard","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Verena","family":"Kain","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Rebecca","family":"Ramjiawan","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Giovanni","family":"Zevi Della Porta","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Simon","family":"Hirlaender","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"266","published-online":{"date-parts":[[2023,4,27]]},"reference":[{"article-title":"AWAKE design report: a proton-driven plasma wakefield acceleration experiment at CERN","year":"2013","author":"","key":"mlstacce21bib1"},{"key":"mlstacce21bib2","doi-asserted-by":"publisher","first-page":"363","DOI":"10.1038\/s41586-018-0485-4","article-title":"Acceleration of electrons in the plasma wakefield of a proton bunch","volume":"561","author":"","year":"2018","journal-title":"Nature"},{"article-title":"Systematic optics studies for the commissioning of the awake electron beamline","year":"2019","author":"Bracco","key":"mlstacce21bib3"},{"author":"Gavalda","key":"mlstacce21bib4","article-title":"Multi-Objective genetic based algorithms and experimental beam lifetime studies for the synchrotron SOLEIL storage ring"},{"article-title":"Bayesian optimisation for fast and safe parameter tuning of SWISSFEL","year":"2019","author":"Kirschner","key":"mlstacce21bib5"},{"article-title":"Novel concepts for optimization of the CERN large hadron collider injection lines","year":"2011","author":"Fuchsberger","key":"mlstacce21bib6"},{"key":"mlstacce21bib7","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.124.124801","article-title":"Bayesian Optimization of a free-electron laser","volume":"124","author":"Duris","year":"2020","journal-title":"Phys. Rev. Lett."},{"key":"mlstacce21bib8","doi-asserted-by":"publisher","first-page":"2","DOI":"10.1109\/TNS.2016.2543203","article-title":"Neural networks for modeling and control of particle accelerators","volume":"63","author":"Edelen","year":"2016","journal-title":"IEEE Trans. Nucl. Sci."},{"article-title":"Neural network model of the PXIE RFQ cooling system and resonant frequency response","year":"2016","author":"Edelen","key":"mlstacce21bib9"},{"article-title":"First steps toward incorporating image based diagnostics into particle accelerator control systems using convolutional neural networks","year":"2018","author":"Edelen","key":"mlstacce21bib10"},{"key":"mlstacce21bib11","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.121.044801","article-title":"Demonstration of Model-independent control of the longitudinal phase space of electron beams in the linac-coherent light source with femtosecond resolution","volume":"121","author":"Scheinker","year":"2018","journal-title":"Phys. Rev. Lett."},{"key":"mlstacce21bib12","first-page":"pp 1241","article-title":"Feedback control for Cassie with deep reinforcement learning","author":"Xie","year":"2018"},{"article-title":"Solving Rubik\u2019s cube with a robot hand","year":"2019","author":"Akkaya","key":"mlstacce21bib13"},{"key":"mlstacce21bib14","doi-asserted-by":"publisher","first-page":"77","DOI":"10.1038\/s41586-020-2939-8","article-title":"Autonomous navigation of stratospheric balloons using reinforcement learning","volume":"588","author":"Bellemare","year":"2020","journal-title":"Nature"},{"key":"mlstacce21bib15","doi-asserted-by":"publisher","first-page":"414","DOI":"10.1038\/s41586-021-04301-9","article-title":"Magnetic control of tokamak plasmas through deep reinforcement learning","volume":"602","author":"Degrave","year":"2022","journal-title":"Nature"},{"article-title":"Model-free and bayesian ensembling model-based deep reinforcement learning for particle accelerator control demonstrated on the FERMI FEL","year":"2020","author":"Hirlaender","key":"mlstacce21bib16"},{"key":"mlstacce21bib17","doi-asserted-by":"publisher","DOI":"10.1088\/1748-0221\/16\/05\/P05012","article-title":"Surrogate modeling of the CLIC final-focus system using artificial neural networks","volume":"16","author":"Ogren","year":"2021","journal-title":"J. Instrum."},{"key":"mlstacce21bib18","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevAccelBeams.23.044601","article-title":"Machine learning for orders of magnitude speedup in multiobjective optimization of particle accelerator systems","volume":"23","author":"Edelen","year":"2020","journal-title":"Phys. Rev. Accel. Beams"},{"key":"mlstacce21bib19","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevAccelBeams.24.014601","article-title":"Multiobjective optimization of the dynamic aperture using surrogate models based on artificial neural networks","volume":"24","author":"Kranjcevic","year":"2021","journal-title":"Phys. Rev. Accel. Beams"},{"key":"mlstacce21bib20","first-page":"p WEAF040","article-title":"Neural network virtual diagnostic for the FAST low energy beam line","author":"Edelen","year":"2018"},{"key":"mlstacce21bib21","first-page":"p THYGBE2","article-title":"Neural network based approaches to the modeling and control of particle accelerators","author":"Edelen","year":"2018"},{"key":"mlstacce21bib22","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevAccelBeams.23.124801","article-title":"Sample-efficient reinforcement learning for CERN accelerator control","volume":"23","author":"Kain","year":"2020","journal-title":"Phys. Rev. Accel. Beams"},{"key":"mlstacce21bib23","doi-asserted-by":"publisher","DOI":"10.1063\/5.0003423","article-title":"Online multi-objective particle accelerator optimization of the AWAKE electron beam line for simultaneous emittance and orbit control","volume":"10","author":"Scheinker","year":"2020","journal-title":"AIP Adv."},{"author":"Burger","key":"mlstacce21bib24","article-title":"A new control system for the CERN TV beams observation"},{"article-title":"OpenAI Gym","year":"2016","author":"Brockman","key":"mlstacce21bib25"},{"key":"mlstacce21bib26","doi-asserted-by":"publisher","first-page":"102","DOI":"10.1016\/j.nima.2018.02.044","article-title":"The electron accelerators for the AWAKE experiment at CERN\u2014baseline and future developments","volume":"909","author":"Pepitone","year":"2018","journal-title":"Nucl. Instrum. Methods Phys. Res. A"},{"key":"mlstacce21bib27","article-title":"CERN python module pyjapc"},{"article-title":"Stable baselines, github repository","year":"2018","author":"Hill","key":"mlstacce21bib28"},{"article-title":"The BOBYQA algorithm for bound constrained optimization without derivatives","year":"2009","author":"Powell","key":"mlstacce21bib29"},{"article-title":"Addressing function approximation error in actor-critic methods","year":"2018","author":"Fujimoto","key":"mlstacce21bib30"},{"article-title":"Auto-encoding variational bayes","year":"2014","author":"Kingma","key":"mlstacce21bib31"},{"article-title":"Understanding disentangling in \u03b2-VAE","year":"2018","author":"Burgess","key":"mlstacce21bib32"},{"article-title":"Improved training of wasserstein GANs","year":"2017","author":"Gulrajani","key":"mlstacce21bib33"},{"key":"mlstacce21bib34","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevAccelBeams.24.072802","article-title":"Physics model-informed Gaussian process for online optimization of particle accelerators","volume":"24","author":"Hanuka","year":"2021","journal-title":"Phys. Rev. Accel. Beams"},{"article-title":"Online accelerator tuning with adaptive bayesian optimization","year":"2022","author":"Kuklev","key":"mlstacce21bib35"},{"key":"mlstacce21bib36","doi-asserted-by":"publisher","first-page":"182","DOI":"10.1109\/TAC.1966.1098285","article-title":"Extremum control of continuous systems","volume":"11","author":"Hamza","year":"1966","journal-title":"IEEE Trans. Autom. Control"},{"author":"Madysa","key":"mlstacce21bib37","article-title":"Automated intensity optimisation using reinforcement learning at LEIR"},{"key":"mlstacce21bib38"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,4,27]],"date-time":"2023-04-27T09:02:35Z","timestamp":1682586155000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acce21"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,4,27]]},"references-count":38,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2023,4,27]]},"published-print":{"date-parts":[[2023,6,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/acce21","relation":{},"ISSN":["2632-2153"],"issn-type":[{"type":"electronic","value":"2632-2153"}],"subject":[],"published":{"date-parts":[[2023,4,27]]},"assertion":[{"value":"Towards automatic setup of 18 MeV electron beamline using machine learning","name":"article_title","label":"Article Title"},{"value":"Machine Learning: Science and Technology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2023 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2022-12-09","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2023-04-18","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2023-04-27","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}