EUVIMEDEUVIMED entdecken
Crossref

Diabatic quantum annealing for training energy-based generative models

Gilhan Kim, Ju-Yeon Gyhm, Daniel K Park

Physical Review E · 05.03.2026 · 113 (3)

DOI: 10.1103/2g6m-whm2 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Physical Review E, online erschienen am 05.03.2026. Verfasst von Gilhan Kim und 2 weiteren Autor:innen. Beteiligte Einrichtung: Yonsei University, Department of Statistics and Data Science, Seoul 03722, Republic of Korea. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 32 Literaturangaben. Gefördert von: Institute for Information and Communications Technology Promotion, National Research Foundation of Korea, Ministry of Trade, Industry and Energy.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Energy-based generative models, such as restricted Boltzmann machines (RBMs), require unbiased Boltzmann samples for effective training. Classical Markov chain Monte Carlo methods, however, converge slowly and yield correlated samples, making large-scale training difficult. We address this bottleneck by applying the analytic relation between annealing schedules and effective inverse temperature in diabatic quantum annealing. By implementing this prescription on a quantum annealer, we obtain temperature-controlled Boltzmann samples that enable RBM training with faster convergence and lower validation error than classical sampling. We further identify a systematic temperature misalignment intrinsic to analog quantum computers and propose an analytical rescaling method that mitigates this hardware noise, thereby enhancing the practicality of quantum annealers as Boltzmann samplers. In our method, the model's connectivity is set directly by the qubit connectivity, transforming the computational complexity inherent in classical sampling into a requirement on quantum hardware. This shift allows the approach to extend naturally from RBMs to fully connected Boltzmann machines, opening opportunities inaccessible to classical training methods.

Autor:innen und Institutionen

  • Yonsei University, Department of Statistics and Data Science, Seoul 03722, Republic of Korea.
  • Seoul National University, Department of Physics and Astronomy, Seoul 08826, Republic of Korea.
  • Yonsei University, Department of Applied Statistics, Seoul 03722, Republic of Korea.
  • Yonsei University, Department of Quantum Information, Seoul 03722, Republic of Korea.

Förderung

  • Institute for Information and Communications Technology Promotion
  • National Research Foundation of Korea
  • Ministry of Trade, Industry and Energy
  • Korea Health Industry Development Institute

Zitieren

APA

Kim, G., Gyhm, J. Y., & Park, D. K. (2026). Diabatic quantum annealing for training energy-based generative models. Physical Review E, 113(3). https://doi.org/10.1103/2g6m-whm2

Vancouver

Kim G, Gyhm JY, Park DK. Diabatic quantum annealing for training energy-based generative models. Physical Review E. 2026;113(3). doi:10.1103/2g6m-whm2

BibTeX
@article{Kim2026,
  author = {Kim, Gilhan and Gyhm, Ju-Yeon and Park, Daniel K},
  title = {Diabatic quantum annealing for training energy-based generative models},
  journal = {Physical Review E},
  year = {2026},
  volume = {113},
  number = {3},
  doi = {10.1103/2g6m-whm2}
}
Kostenlos starten →