% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@INPROCEEDINGS{Hirlaender:607018,
author = {Hirlaender, Simon and Pochaba, Sabrina and Lamminger, Lukas
and Kaiser, Jan and Xu, Chenran and Santamaria Garcia,
Andrea and Scomparin, Luca},
title = {{T}owards few-shot reinforcement learning in particle
accelerator control},
address = {Geneva, Switzerland},
publisher = {JACoW Publishing},
reportid = {PUBDB-2024-01723},
isbn = {978-3-95450-247-9},
pages = {1804 - 1807},
year = {2024},
note = {Literaturangaben;},
comment = {[Ebook] 15th International Particle Accelerator Conference,
Nashville, Tennessee : May 19-24, 2024, Nashville,
Tennessee, USA : proceedings / Pilat, Fulvia ; Andrian, Ivan
, [Geneva] : JACoW Publishing, [2024],},
booktitle = {[Ebook] 15th International Particle
Accelerator Conference, Nashville,
Tennessee : May 19-24, 2024, Nashville,
Tennessee, USA : proceedings / Pilat,
Fulvia ; Andrian, Ivan , [Geneva] :
JACoW Publishing, [2024],},
abstract = {This paper addresses the automation of particle accelerator
control through reinforcement learning (RL). It highlights
the potential to increase reliable performance, especially
in light of new diagnostic tools and the increasingly
complex variable schedules of specific accelerators. We
focus on the physics simulation of the AWAKE electron line,
an ideal platform for performing in-depth studies that allow
a clear distinction between the problem and the performance
of different algorithmic approaches for accurate analysis.
The main challenges are the lack of realistic simulations
and partially observable environments. We show how effective
results can be achieved through meta-reinforcement learning,
where an agent is trained to quickly adapt to specific
real-world scenarios based on prior training in a simulated
environment with variable unknowns. When suitable
simulations are lacking or too costly, a model-based method
using Gaussian processes is used for direct training in a
few shots only. The work opens new avenues for implementing
control automation in particle accelerators, significantly
increasing their efficiency and adaptability.},
month = {May},
date = {2024-05-19},
organization = {15th International Particle
Accelerator Conference, Nashville
(USA), 19 May 2024 - 24 May 2024},
keywords = {Accelerator Physics (Other) /
mc6-beam-instrumentation-controls-feedback-and-operational-aspects
- MC6: Beam Instrumentation, Controls, Feedback, and
Operational Aspects (Other) / MC6.D13 - MC6.D13 Machine
Learning (Other)},
cin = {MSK / KIT / CERN},
cid = {I:(DE-H253)MSK-20120731 / I:(DE-H253)KIT-20130928 /
I:(DE-H253)CERN-20181204},
pnm = {621 - Accelerator Research and Development (POF4-621) /
InternLabs-0011 - HIR3X - Helmholtz International Laboratory
on Reliability, Repetition, Results at the most advanced
X-ray Sources $(2020_InternLabs-0011)$},
pid = {G:(DE-HGF)POF4-621 / $G:(DE-HGF)2020_InternLabs-0011$},
experiment = {EXP:(DE-H253)ARES-20200101},
typ = {PUB:(DE-HGF)8 / PUB:(DE-HGF)7},
doi = {10.18429/JACOW-IPAC2024-TUPS60},
url = {https://bib-pubdb1.desy.de/record/607018},
}