Elamur Research⌕
☰
home / papers / 2026 / Large Language Models Can Help Mitigate Barren Plateaus in Quantum Neural Networks

Large Language Models Can Help Mitigate Barren Plateaus in Quantum Neural Networks

Paper · 2026 · Jun Zhuang and Chaowen Guan

Summary

Quantum neural networks (QNNs) are often untrainable because of barren plateaus (BPs). At initialization, gradient variance shrinks exponentially as qubits are added, so gradient-based optimizers stall almost immediately. Existing initialization fixes, such as GaInit and BeInit, sample parameters once from a fixed, hand-designed distribution. That approach relies on idealized assumptions and does not adapt to circuit size or data.

The authors propose AdaInit, a loop in which an LLM proposes candidate initial parameters. Its prompts include a dataset description and feedback on earlier gradient behaviour. Each candidate initializes a QNN that is trained briefly, and the resulting early gradient variance is scored with Expected Improvement. The prompts, the best variance so far and a pool of good candidates are then updated.

The authors model this loop as a submartingale. Using Doob's convergence and stopping-time results, they prove it almost surely finds parameters with non-negligible gradient variance in finite time. Experiments show AdaInit keeps gradient variance higher than baselines as QNNs grow, which positions LLMs as adaptive initializers for quantum models.

Key points

  • Theory: The paper proves a supremum and an expected hitting time for the iterative process. It assumes bounded gradients, meaning no gradient explosion occurs.
  • Baselines: AdaInit is compared with three classic initialization methods and two BP-oriented strategies, GaInit and BeInit.
  • Model sizes: Simulations cover QNNs with 2–20 qubits and 4–40 layers, under noise-free measurement.
  • Datasets: The experiments use the Iris, Wine, Titanic and MNIST benchmarks.
  • Limitations: BPs caused by the ansatz itself are out of scope. Future work targets faster convergence and using the framework to guide QNN architecture design.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Boise State UniversityBoise · United States4 works · 93 citations🎓University of CincinnatiCincinnati · United States6 works · 38,342 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksPhysics and Astronomy · 52 worksAtomic and Molecular Physics, and Optics · 4 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 3,894 works

Cite

Cite as Jun Zhuang and Chaowen Guan. 2026. Large Language Models Can Help Mitigate Barren Plateaus in Quantum Neural Networks. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.522

@inproceedings{zhuang2026large,
  title = {{Large Language Models Can Help Mitigate Barren Plateaus in Quantum Neural Networks}},
  author = {Jun Zhuang and Chaowen Guan},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.522},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.522}
}

References

38 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783.
  2. Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
  3. Anthropic. 2024. Introducing claude 3.5 sonnet. Accessed: 2025-02-15. find
  4. Kishor Bharti and Tobias Haug. 2021. find
  5. Quantum-Harper R Grimsley, George S Barron, Edwin Barnes, Sophia E Economou, and Nicholas J Mayhall. 2023. Adaptive, problem-tailored variational quantum eigensolver mitigates rough parameter landscapes and barren plateaus. npj Quantum Information, 9(1):19. find
  6. Valentin Heyraud, Zejian Li, Kaelan Donatella, Alexandre Le Boité, and Cristiano Ciuti. 2023. Efficient estimation of trainability for variational quantum circuits. PRX Quantum, 4(4):040335. find
  7. Jarrod R McClean, Sergio Boixo, Vadim N Smelyanskiy, Ryan Babbush, and Hartmut Neven. 2018. Barren plateaus in quantum neural network training landscapes. Nature communications, 9(1):4812. find
  8. Zoë Holmes, Kunal Sharma, Marco Cerezo, and Patrick J Coles. 2022. Connecting ansatz expressibility to gradient magnitudes and barren plateaus. PRX quantum, 3(1):010313. find
  9. Antonio A Mele, Glen B Mbeng, Giuseppe E Santoro, Mario Collura, and Pietro Torta. 2022. Avoiding barren plateaus via transferability of smooth solutions in a hamiltonian variational ansatz. Physical Review A, 106(6):L060401. find
  10. Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, et al. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276.
  11. Qingyue Jiao, Yongcan Tang, Jun Zhuang, Jason Cong, and Yiyu Shi. 2025. Mediq-gan: Quantum-inspired gan for high resolution medical image generation. arXiv preprint arXiv:2506.21015. — [arXiv:2506.21015](https://arxiv.org/abs/2506.21015) find
  12. Abhinav Kandala, Antonio Mezzacapo, Kristan Temme, Maika Takita, Markus Brink, Jerry M Chow, and Jay M Gambetta. 2017. Hardware-efficient variational quantum eigensolver for small molecules and quantum magnets. nature, 549(7671):242–246. find
  13. Muhammad Kashif and Saif Al-Kuwari. 2024. Resqnets: a residual approach for mitigating barren plateaus in quantum neural networks. EPJ Quantum Technology. find
  14. Muhammad Kashif, Muhammad Rashid, Saif Al-Kuwari, and Muhammad Shafique. 2024. Alleviating barren plateaus in parameterized quantum machine learning circuits: Investigating advanced parameter initialization strategies. In 2024 Design, Automation & Test in Europe Conference & Exhibition (DATE), pages… find
  15. Kosuke Mitarai, Makoto Negoro, Masahiro Kitagawa, and Keisuke Fujii. 2018. Quantum circuit learning. Physical Review A, 98(3):032309. find
  16. Carlos Ortiz Marrero, Mária Kieferová, and Nathan Wiebe. 2021. Entanglement-induced barren plateaus. PRX quantum, 2(4):040316. find
  17. Mateusz Ostaszewski, Edward Grant, and Marcello Benedetti. 2021. Structure optimization for parameterized quantum circuits. Quantum, 5:391. find
  18. Chae-Yeun Park and Nathan Killoran. 2024. Hamiltonian variational ansatz without barren plateaus. Quantum, 8:1239. find
  19. John Preskill. 2018. Quantum computing in the nisq era and beyond. Quantum, 2:79. find
  20. Han Qi, Lei Wang, Hongsheng Zhu, Abdullah Gani, and Changqing Gong. 2023. The barren plateaus of quantum neural networks: review, taxonomy and trends. Quantum Information Processing, 22(12):435. find
  21. Maqsudur Rahman and Jun Zhuang. 2025. Nqnn: Noise-aware quantum neural networks for medical image classification. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 433–442. Springer. find
  22. Ankit Kulshrestha and Ilya Safro. 2022. Beinit: Avoiding barren plateaus in variational quantum algorithms. In 2022 IEEE international conference on quantum computing and engineering (QCE), pages 197–203. IEEE. find
  23. Maqsudur Rahman and Jun Zhuang. 2026. Qapnet: A quantum-attentive patchwise network for robust medical image classification under noisy inputs. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 40, pages 25065–25072. find
  24. Guangxi Li, Zhixin Song, and Xin Wang. 2021. Vsql: Variational shadow quantum learning for classification. In Proceedings of the AAAI conference on artificial intelligence. find
  25. Sonny Rappaport, Gaurav Gyawali, Tiago Sereno, and Michael J Lawler. 2023. Measurement-induced landscape transitions in hybrid variational quantum circuits. arXiv preprint arXiv:2312.09135. — [arXiv:2312.09135](https://arxiv.org/abs/2312.09135) find
  26. Huan-Yu Liu, Tai-Ping Sun, Yu-Chun Wu, Yong-Jian Han, and Guo-Ping Guo. 2023. Mitigating barren plateaus with transfer-learning-inspired parameter initializations. New Journal of Physics, 25(1):013039. find
  27. Stefan H Sack, Raimel A Medina, Alexios A Michailidis, Richard Kueng, and Maksym Serbyn. 2022. Avoiding barren plateaus using classical shadows. PRX Quantum, 3(2):020365. find
  28. Xia Liu, Geng Liu, Hao-Kai Zhang, Jiaxin Huang, and Xin Wang. 2024. Mitigating barren plateaus of variational quantum eigensolvers. IEEE Transactions on Quantum Engineering. find
  29. Andrea Mari, Thomas R Bromley, Josh Izaac, Maria Schuld, and Nathan Killoran. 2020. Transfer learning in hybrid classical-quantum neural networks. Quantum, 4:340. find
  30. Antonio Sannia, Francesco Tacchino, Ivano Tavernelli, Gian Luca Giorgi, and Roberta Zambrini. 2024. Engineered dissipation to mitigate barren plateaus. npj Quantum Information, 10(1):81. find
  31. Marco Sciorilli, Lucas Borges, Taylor L Patti, Diego García-Martín, Giancarlo Camilo, Anima Anandkumar, and Leandro Aolita. 2024. Towards large-scale quantum optimization solvers with few qubits. arXiv preprint arXiv:2401.09421. — [arXiv:2401.09421](https://arxiv.org/abs/2401.09421) find
  32. Raja Selvarajan, Manas Sajjan, Travis S Humble, and Sabre Kais. 2023. Dimensionality reduction with variational encoders based on subsystem purification. Mathematics. find
  33. Yun Shang and Xiao Shi. 2025. Avoiding barren plateaus via gaussian mixture model. New Journal of Physics. find
  34. Myeongjin Shin, Seungwoo Lee, Mingyu Lee, Donghwa Ji, Hyeonjun Yeo, Harrison J Lee, and Kab-gyun Jeong. 2024. Layerwise quantum convolutional neural networks provide a unified way for estimating fundamental properties of quantum information theory. arXiv preprint arXiv:2401.07716. — [arXiv:2401.07716](https://arxiv.org/abs/2401.07716) find
  35. Samuel A Stein, Betis Baheri, Daniel Chen, Ying Mao, Qiang Guan, Ang Li, Bo Fang, and Shuai Xu. 2021. Qugan: A quantum state fidelity based generative adversarial network. In 2021 IEEE International Conference on Quantum Computing and Engineering (QCE), pages 71–81. IEEE. find
  36. Yudai Suzuki, Hiroshi Yano, Rudy Raymond, and Naoki Yamamoto. 2021. Normalized gradient descent for variational quantum algorithms. In 2021 IEEE International Conference on Quantum Computing and Engineering (QCE), pages 1–9. IEEE. find
  37. Gemini Team, Petko Georgiev, Ving Ian Lei, Ryan Burnell, Libin Bai, Anmol Gulati, Garrett Tanzer, Damien Vincent, Zhufeng Pan, Shibo Wang, et al. 2024. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context. arXiv preprint arXiv:2403.05530.
  38. Cenk Tüysüz, Giuseppe Clemente, Arianna Crippa, Tobias Hartung, Stefan Kühn, and Karl Jansen. 2023. Classical splitting of parametrized quantum circuits. Quantum Machine Intelligence. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhuang-2026-large-language-models-can-help/