---
_id: '67161'
abstract:
- lang: eng
  text: <jats:p>Optimization algorithms contain parameters that greatly influence
    their behavior. Finding the right settings for parameters through automated algorithm
    configuration has become a critical component of designing competitive algorithms.
    While traditional offline configurators tackle this problem by finding one configuration
    that works well for a set of instances, instance-specific algorithm configuration
    utilizes features of the instances to provide configurations that are tailored
    to each instance to maximize performance. We propose the first instance-specific
    algorithm configurator based on deep reinforcement learning that can be used in
    general algorithm configuration settings. Our method is able to handle large,
    mixed, discrete and continuous search spaces and only requires a small number
    of instances for training. Not only does it select an individual configuration
    for every instance, it also selects configurations from a much broader range.
    We show that our configurator provides improvements over the state-of-the-art
    instance-specific configurators ISAC++ and Hydra on a wide range of problem domains.</jats:p>
article_number: '3844954'
author:
- first_name: Elias
  full_name: Schede, Elias
  last_name: Schede
- first_name: Moritz Vinzent
  full_name: Seiler, Moritz Vinzent
  last_name: Seiler
- first_name: Carolin
  full_name: Mensendiek, Carolin
  id: '75006'
  last_name: Mensendiek
- first_name: Kevin
  full_name: Tierney, Kevin
  last_name: Tierney
- first_name: Heike
  full_name: Trautmann, Heike
  id: '100740'
  last_name: Trautmann
  orcid: 0000-0002-9788-8282
citation:
  ama: 'Schede E, Seiler MV, Mensendiek C, Tierney K, Trautmann H. OPTICAT: A Deep
    Reinforcement Learning Framework for Instance-Specific Algorithm Configuration.
    <i>ACM Transactions on Evolutionary Learning and Optimization</i>. Published online
    2026. doi:<a href="https://doi.org/10.1145/3844954">10.1145/3844954</a>'
  apa: 'Schede, E., Seiler, M. V., Mensendiek, C., Tierney, K., &#38; Trautmann, H.
    (2026). OPTICAT: A Deep Reinforcement Learning Framework for Instance-Specific
    Algorithm Configuration. <i>ACM Transactions on Evolutionary Learning and Optimization</i>,
    Article 3844954. <a href="https://doi.org/10.1145/3844954">https://doi.org/10.1145/3844954</a>'
  bibtex: '@article{Schede_Seiler_Mensendiek_Tierney_Trautmann_2026, title={OPTICAT:
    A Deep Reinforcement Learning Framework for Instance-Specific Algorithm Configuration},
    DOI={<a href="https://doi.org/10.1145/3844954">10.1145/3844954</a>}, number={3844954},
    journal={ACM Transactions on Evolutionary Learning and Optimization}, publisher={Association
    for Computing Machinery (ACM)}, author={Schede, Elias and Seiler, Moritz Vinzent
    and Mensendiek, Carolin and Tierney, Kevin and Trautmann, Heike}, year={2026}
    }'
  chicago: 'Schede, Elias, Moritz Vinzent Seiler, Carolin Mensendiek, Kevin Tierney,
    and Heike Trautmann. “OPTICAT: A Deep Reinforcement Learning Framework for Instance-Specific
    Algorithm Configuration.” <i>ACM Transactions on Evolutionary Learning and Optimization</i>,
    2026. <a href="https://doi.org/10.1145/3844954">https://doi.org/10.1145/3844954</a>.'
  ieee: 'E. Schede, M. V. Seiler, C. Mensendiek, K. Tierney, and H. Trautmann, “OPTICAT:
    A Deep Reinforcement Learning Framework for Instance-Specific Algorithm Configuration,”
    <i>ACM Transactions on Evolutionary Learning and Optimization</i>, Art. no. 3844954,
    2026, doi: <a href="https://doi.org/10.1145/3844954">10.1145/3844954</a>.'
  mla: 'Schede, Elias, et al. “OPTICAT: A Deep Reinforcement Learning Framework for
    Instance-Specific Algorithm Configuration.” <i>ACM Transactions on Evolutionary
    Learning and Optimization</i>, 3844954, Association for Computing Machinery (ACM),
    2026, doi:<a href="https://doi.org/10.1145/3844954">10.1145/3844954</a>.'
  short: E. Schede, M.V. Seiler, C. Mensendiek, K. Tierney, H. Trautmann, ACM Transactions
    on Evolutionary Learning and Optimization (2026).
date_created: 2026-09-17T05:10:53Z
date_updated: 2026-09-17T05:11:28Z
department:
- _id: '819'
doi: 10.1145/3844954
language:
- iso: eng
publication: ACM Transactions on Evolutionary Learning and Optimization
publication_identifier:
  issn:
  - 2688-299X
  - 2688-3007
publication_status: published
publisher: Association for Computing Machinery (ACM)
status: public
title: 'OPTICAT: A Deep Reinforcement Learning Framework for Instance-Specific Algorithm
  Configuration'
type: journal_article
user_id: '15504'
year: '2026'
...