Standard

Transfer Learning for Transformer-Based Modeling of Nonlinear Pulse Evolution in Er-Doped Fiber Amplifiers. / Bednyakova, A. E.; Gemuzov, A. S.; Mishevsky, M. S. et al.

In: Laser and Photonics Reviews, 12.12.2025.

Research output: Contribution to journal › Article › peer-review

Harvard

Bednyakova, AE, Gemuzov, AS, Mishevsky, MS, Saraeva, KP, Redyuk, AA, Mkrtchyan, AA, Nasibulin, AG & Gladush, YG 2025, 'Transfer Learning for Transformer-Based Modeling of Nonlinear Pulse Evolution in Er-Doped Fiber Amplifiers', Laser and Photonics Reviews. https://doi.org/10.1002/lpor.202502014

APA

Bednyakova, A. E., Gemuzov, A. S., Mishevsky, M. S., Saraeva, K. P., Redyuk, A. A., Mkrtchyan, A. A., Nasibulin, A. G., & Gladush, Y. G. (2025). Transfer Learning for Transformer-Based Modeling of Nonlinear Pulse Evolution in Er-Doped Fiber Amplifiers. Laser and Photonics Reviews, [e02014]. https://doi.org/10.1002/lpor.202502014

Vancouver

Bednyakova AE, Gemuzov AS, Mishevsky MS, Saraeva KP, Redyuk AA, Mkrtchyan AA et al. Transfer Learning for Transformer-Based Modeling of Nonlinear Pulse Evolution in Er-Doped Fiber Amplifiers. Laser and Photonics Reviews. 2025 Dec 12;e02014. doi: 10.1002/lpor.202502014

Author

Bednyakova, A. E. ; Gemuzov, A. S. ; Mishevsky, M. S. et al. / Transfer Learning for Transformer-Based Modeling of Nonlinear Pulse Evolution in Er-Doped Fiber Amplifiers. In: Laser and Photonics Reviews. 2025.

BibTeX

@article{c7db52d772574ed69cdfd85b96db57ae,
title = "Transfer Learning for Transformer-Based Modeling of Nonlinear Pulse Evolution in Er-Doped Fiber Amplifiers",
abstract = "A neural network model based on the Transformer architecture has been developed to predict the nonlinear evolution of optical pulses in Er-doped fiber amplifier under conditions of limited experimental data. To address data scarcity, a two-stage training strategy is employed. In the first stage, the model is pretrained on a synthetic dataset generated through numerical simulations of the amplifier's nonlinear dynamics. In the second stage, the model is fine-tuned using a small set of experimental measurements. This approach enables accurate reproduction of the fine spectral structure of optical pulses observed in experiments across various nonlinear evolution regimes, including the development of modulational instability and the propagation of high-order solitons.",
keywords = "fiber amplifier, machine learning, soliton, transfer learning, transformer",
author = "Bednyakova, {A. E.} and Gemuzov, {A. S.} and Mishevsky, {M. S.} and Saraeva, {K. P.} and Redyuk, {A. A.} and Mkrtchyan, {A. A.} and Nasibulin, {A. G.} and Gladush, {Y. G.}",
note = "A.G. and A.B. acknowledge the support of the Russian Science Founda-tion (No. 25-61-00010, https://rscf.ru/project/25- 61- 00010/). A.R. and K.S.acknowledge the support of the Ministry of Science and Higher Educationof the Russian Federation (Project No. FSUS-2021-0015).",
year = "2025",
month = dec,
day = "12",
doi = "10.1002/lpor.202502014",
language = "English",
journal = "Laser and Photonics Reviews",
issn = "1863-8880",
publisher = "Wiley-VCH Verlag",

}

RIS

TY - JOUR

T1 - Transfer Learning for Transformer-Based Modeling of Nonlinear Pulse Evolution in Er-Doped Fiber Amplifiers

AU - Bednyakova, A. E.

AU - Gemuzov, A. S.

AU - Mishevsky, M. S.

AU - Saraeva, K. P.

AU - Redyuk, A. A.

AU - Mkrtchyan, A. A.

AU - Nasibulin, A. G.

AU - Gladush, Y. G.

N1 - A.G. and A.B. acknowledge the support of the Russian Science Founda-tion (No. 25-61-00010, https://rscf.ru/project/25- 61- 00010/). A.R. and K.S.acknowledge the support of the Ministry of Science and Higher Educationof the Russian Federation (Project No. FSUS-2021-0015).

PY - 2025/12/12

Y1 - 2025/12/12

N2 - A neural network model based on the Transformer architecture has been developed to predict the nonlinear evolution of optical pulses in Er-doped fiber amplifier under conditions of limited experimental data. To address data scarcity, a two-stage training strategy is employed. In the first stage, the model is pretrained on a synthetic dataset generated through numerical simulations of the amplifier's nonlinear dynamics. In the second stage, the model is fine-tuned using a small set of experimental measurements. This approach enables accurate reproduction of the fine spectral structure of optical pulses observed in experiments across various nonlinear evolution regimes, including the development of modulational instability and the propagation of high-order solitons.

AB - A neural network model based on the Transformer architecture has been developed to predict the nonlinear evolution of optical pulses in Er-doped fiber amplifier under conditions of limited experimental data. To address data scarcity, a two-stage training strategy is employed. In the first stage, the model is pretrained on a synthetic dataset generated through numerical simulations of the amplifier's nonlinear dynamics. In the second stage, the model is fine-tuned using a small set of experimental measurements. This approach enables accurate reproduction of the fine spectral structure of optical pulses observed in experiments across various nonlinear evolution regimes, including the development of modulational instability and the propagation of high-order solitons.

KW - fiber amplifier

KW - machine learning

KW - soliton

KW - transfer learning

KW - transformer

UR - https://www.scopus.com/pages/publications/105024798599

UR - https://www.mendeley.com/catalogue/9cb75339-a599-351b-9604-eaa7cc9b9415/

U2 - 10.1002/lpor.202502014

DO - 10.1002/lpor.202502014

M3 - Article

JO - Laser and Photonics Reviews

JF - Laser and Photonics Reviews

SN - 1863-8880

M1 - e02014

ER -

ID: 72848596