Using Reinforcement Learning for Per-Instance Algorithm Configuration on the TSP

Automated Algorithm Configuration (AAC) usually takes a global perspective: it identifies a parameter configuration for an (optimization) algorithm that maximizes a performance metric over a set of instances. However, the optimal choice of parameters strongly depends on the instance at hand and shou...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE Symposium on Computational Intelligence in Multi-Criteria Decision Making S. 361 - 368
Hauptverfasser: Vinzent Seiler, Moritz, Rook, Jeroen, Heins, Jonathan, Ludger Preub, Oliver, Bossek, Jakob, Trautmann, Heike
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 05.12.2023
Schlagworte:
Artificial neural networks
Benchmark testing
Computational modeling
Deep Reinforcement Learning
Measurement
Optimization
Per-Instance Algorithm Configuration
Prediction algorithms
Reinforcement learning
Traveling Salesperson Problem
ISSN:2472-8322
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract Automated Algorithm Configuration (AAC) usually takes a global perspective: it identifies a parameter configuration for an (optimization) algorithm that maximizes a performance metric over a set of instances. However, the optimal choice of parameters strongly depends on the instance at hand and should thus be calculated on a per-instance basis. We explore the potential of Per-Instance Algorithm Configuration (PIAC) by using Reinforcement Learning (RL). To this end, we propose a novel PIAC approach that is based on deep neural networks. We apply it to predict configurations for the Lin-Kernighan heuristic (LKH) for the Traveling Salesperson Problem (TSP) individually for every single instance. To train our PIAC approach, we create a large set of 100 000 TSP instances with 2 000 nodes each - currently the largest benchmark set to the best of our knowledge. We compare our approach to the state-of-the-art AAC method Sequential Model-based Algorithm Configuration (SMAC). The results show that our PIAC approach outperforms this baseline on both the newly created instance set and established instance sets.
AbstractList Automated Algorithm Configuration (AAC) usually takes a global perspective: it identifies a parameter configuration for an (optimization) algorithm that maximizes a performance metric over a set of instances. However, the optimal choice of parameters strongly depends on the instance at hand and should thus be calculated on a per-instance basis. We explore the potential of Per-Instance Algorithm Configuration (PIAC) by using Reinforcement Learning (RL). To this end, we propose a novel PIAC approach that is based on deep neural networks. We apply it to predict configurations for the Lin-Kernighan heuristic (LKH) for the Traveling Salesperson Problem (TSP) individually for every single instance. To train our PIAC approach, we create a large set of 100 000 TSP instances with 2 000 nodes each - currently the largest benchmark set to the best of our knowledge. We compare our approach to the state-of-the-art AAC method Sequential Model-based Algorithm Configuration (SMAC). The results show that our PIAC approach outperforms this baseline on both the newly created instance set and established instance sets.
Author Trautmann, Heike
Heins, Jonathan
Bossek, Jakob
Ludger Preub, Oliver
Vinzent Seiler, Moritz
Rook, Jeroen
Author_xml – sequence: 1
  givenname: Moritz
  surname: Vinzent Seiler
  fullname: Vinzent Seiler, Moritz
  email: moritz.seiler@uni-muenster.de
  organization: University of Münster,Data Science: Statistics and Optimization,Münster,Germany
– sequence: 2
  givenname: Jeroen
  surname: Rook
  fullname: Rook, Jeroen
  email: j.g.rook@utwente.nl
  organization: University of Twente,Data Management and Biometrics,Enschede,Netherlands
– sequence: 3
  givenname: Jonathan
  surname: Heins
  fullname: Heins, Jonathan
  email: jonathan.heins@tu-dresden.de
  organization: TU Dresden,Big Data Analytics in Transportation,Dresden,Germany
– sequence: 4
  givenname: Oliver
  surname: Ludger Preub
  fullname: Ludger Preub, Oliver
  email: oliver.preuss@uni-muenster.de
  organization: University of Münster,Data Science: Statistics and Optimization,Münster,Germany
– sequence: 5
  givenname: Jakob
  surname: Bossek
  fullname: Bossek, Jakob
  email: bossek@aim.rwth-aachen.de
  organization: Aachen University,Chair for AI Methodology RWTH,Aachen,Germany
– sequence: 6
  givenname: Heike
  surname: Trautmann
  fullname: Trautmann, Heike
  email: trautmann@wi.uni-muenster.de
  organization: University of Münster,Data Science: Statistics and Optimization,Münster,Germany
BookMark eNo1kM1KAzEUhaMoWGvfQDAvMDU3_1mWQetAweK065KZ3GkjbUYy48K3t6LC4TvwLc7i3JKr1Cck5AHYHIC5x7ouK8VBmjlnXMyBCcMZsxdk5owFrZUU7IxLMuHS8MIKzm_IbBjeGWOgQQsjJ2S9HWLa0zeMqetziydMI12hz-lHnxVdYy6qNIw-tUgXx32f43g40bJPXdx_Zj_GPtFzxgPSTb2-I9edPw44--sp2T4_bcqXYvW6rMrFqogAbixCYyFoFaRDMFwK3XWtbp1iSoFrbWuxaRgH5wMPwQmHIoBqZDDauiC9E1Ny_7sbEXH3kePJ56_d_wniG7qzUt0
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.1109/SSCI52147.2023.10372008
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP All) 1998-Present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Computer Science
EISBN 9781665430654
1665430656
EISSN 2472-8322
EndPage 368
ExternalDocumentID 10372008
Genre orig-research
GroupedDBID 6IE
6IF
6IK
6IL
6IN
AAJGR
AAWTH
ABLEC
ADZIZ
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
IEGSK
IPLJI
OCL
RIE
RIL
ID FETCH-LOGICAL-i119t-db81d65d49e172436ffc6c9505519c8c8ebb0219ad2dd939e3d15b4d7689d4a93
IEDL.DBID RIE
IngestDate Wed Aug 27 02:30:18 EDT 2025
IsPeerReviewed false
IsScholarly false
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i119t-db81d65d49e172436ffc6c9505519c8c8ebb0219ad2dd939e3d15b4d7689d4a93
PageCount 8
ParticipantIDs ieee_primary_10372008
PublicationCentury 2000
PublicationDate 2023-Dec.-5
PublicationDateYYYYMMDD 2023-12-05
PublicationDate_xml – month: 12
  year: 2023
  text: 2023-Dec.-5
  day: 05
PublicationDecade 2020
PublicationTitle IEEE Symposium on Computational Intelligence in Multi-Criteria Decision Making
PublicationTitleAbbrev SSCI
PublicationYear 2023
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0001616374
ssib053788446
Score 1.8530594
Snippet Automated Algorithm Configuration (AAC) usually takes a global perspective: it identifies a parameter configuration for an (optimization) algorithm that...
SourceID ieee
SourceType Publisher
StartPage 361
SubjectTerms Artificial neural networks
Benchmark testing
Computational modeling
Deep Reinforcement Learning
Measurement
Optimization
Per-Instance Algorithm Configuration
Prediction algorithms
Reinforcement learning
Traveling Salesperson Problem
Title Using Reinforcement Learning for Per-Instance Algorithm Configuration on the TSP
URI https://ieeexplore.ieee.org/document/10372008
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV07T8MwELagYmAqjyDe8sDqNKkfiUdUUdGlimiRulWxfSlB0FYh5fdjOwmIgQHJg-XBtvzQfWffdx9CdywCTZnISaTsbWJcFUQxGpGhjoUwTsLIFF5sIplO08VCZi1Z3XNhAMAHn0Hoqv4v32z0zj2VDRynbeipvftJkjRkre7wcJcYvfNtXhsoI2jC2piuOJKD2Ww04U6XJ3Sa4WHX2y9dFW9Wxv1_TugIBT8EPZx9m55jtAfrE9TvFBpwe2FPUeYjAvAT-Pyo2j8F4jal6grbJpxBRSYeItou799Wm6qsX96xG6dc7ZrjgW2xOBHPZ1mAnscP89EjaTUUSBnHsiZGWUAquGESLFRhVBSFFlpa3GOhm051CkpZMy9zMzRGUgnUxFwxY70QaVgu6RnqrTdrOEfYeYJ5buGLLBgDyB3eYspYE6-4LlJxgQK3QsttkyZj2S3O5R_tV-jQ7YOPDeHXqFdXO7hBB_qzLj-qW7-5X9s2o7M
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LSwMxEA5SBT3VR8W3OXhNu48k3RylWFqsZbEVeiubZLauaFfWrb_fJLurePAg5BBySEIezDfJfPMhdEM9UCHlCfGkuU2UyZRIGnokUD7n2koY6dSJTfSn02ixEHFNVndcGABwwWfQtVX3l69ztbFPZT3LaQsctXebURr4FV2rOT7MpkZvvJuXCszwsE_rqC7fE73ZbDBmVpmna1XDu01_v5RVnGEZtv85pX3U-aHo4fjb-BygLVgfonaj0YDrK3uEYhcTgB_BZUhV7jEQ10lVV9g04RgKMnYg0XR5-7rKi6x8fsN2nGy1qQ4INsUgRTyfxR30NLybD0akVlEgme-LkmhpIClnmgowYIWGPE0VV8IgHwPeVKQikNIYepHoQGsRCgi1zyTVxg8RmiYiPEatdb6GE4StL5gkBsCIlFKAxCIuKrUx8pKpNOKnqGNXaPleJcpYNotz9kf7NdodzR8my8l4en-O9uyeuEgRdoFaZbGBS7SjPsvso7hyG_0FkIKm-g
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=proceeding&rft.title=IEEE+Symposium+on+Computational+Intelligence+in+Multi-Criteria+Decision+Making&rft.atitle=Using+Reinforcement+Learning+for+Per-Instance+Algorithm+Configuration+on+the+TSP&rft.au=Vinzent+Seiler%2C+Moritz&rft.au=Rook%2C+Jeroen&rft.au=Heins%2C+Jonathan&rft.au=Ludger+Preub%2C+Oliver&rft.date=2023-12-05&rft.pub=IEEE&rft.eissn=2472-8322&rft.spage=361&rft.epage=368&rft_id=info:doi/10.1109%2FSSCI52147.2023.10372008&rft.externalDocID=10372008