danДоклады Национальной академии наук БеларусиDoklady of the National Academy of Sciences of Belarus1561-83232524-2431The Republican Unitary Enterprise Publishing House "Belaruskaya Navuka"10.29235/1561-8323-2026-70-5-364-369dan-1329Research ArticleФИЗИКАPHYSICSОптимизация параметров оптического нанотрансивера с использованием машинного обученияOptimization of optical nanotransceiver parameters using machine learningЕфимоваТ. А.EfimovaT. A.

Ефимова Таисия Адамовна – аспирант, мл. науч. сотрудник

пр. Независимости, 68-2, 220072, Минск 

Efimova Taisia A. – Postgraduate Student, Junior Researcher 

68-2, Nezavisimosti Ave., 220072, Minsk 

t.efimova@ifanbel.bas-net.byГапоненкоС. В.GaponenkoS. V.

Гапоненко Сергей Васильевич – академик, д-р физ.- мат. наук, профессор, гл. науч. сотрудник

пр. Независимости, 68-2, 220072, Минск 

Gaponenko Sergey V. – Academician, D. Sc. (Physics and Mathematics), Professor, Chief Researcher

68-2, Nezavisimosti Ave., 220072, Minsk 

s.gaponenko@ifanbel.bas-net.byИнститут физики имени Б. И. Степанова Национальной академии наук БеларусиБеларусьB. I. Stepanov Institute of Physics of the National Academy of Sciences of BelarusBelarus202613112026705364369Copyright © Ефимова Т.А., Гапоненко С.В., 20262026Ефимова Т.А., Гапоненко С.В.Efimova T.A., Gaponenko S.V.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://doklady.belnauka.by/jour/article/view/1329

Предложен подход к исследованию и оптимизации параметров оптического нанотрансивера на основе модели «металлическая наночастица – квантовый излучатель» с использованием методов машинного обучения. Для ускорения многомерного поиска построена суррогатная модель на основе градиентного бустинга (Gradient Boosting Machine, GBM), для чего было сгенерировано 60 000 точек методом латинского гиперкуба (Latin Hypercube Sampling, LHS) в шестимерном пространстве параметров: длина волны возбуждения и испускания, радиус наночастицы, расстояние частица–излучатель, показатель преломления среды и собственный квантовый выход излучателя. Достигнут коэффициент детерминации R2 = 0,947 на тестовой выборке. Проведена глобальная оптимизация по предсказаниям модели с последующим уточнением прямым расчетом и показано, что суррогатная модель эффективна для многопараметрического анализа и сужения области поиска параметров, необходимых для оптимизации эксперимента.

The paper proposes an approach to the study and optimization of optical nanotransceiver parameters based on the “metal nanoparticle – quantum emitter” model using machine learning methods. To speed up multidimensional search, a surrogate model based on Gradient Boosting Machine (GBM) was built, for which 60,000 points were generated using the Latin Hypercube Sampling (LHS) method in a six-dimensional parameter space: excitation and emission wavelengths, nanoparticle radius, particle–emitter distance, refractive index of the medium, and intrinsic quantum yield of the emitter. The coefficient of determination R2 = 0.947 was reached in the test sample. Global optimization based on model predictions was performed, followed by refinement by direct calculation, and it was shown that the surrogate model is effective for multiparametric analysis and narrowing the search area for parameters necessary for optimizing the experiment.

нанотрансиверплазмонное усилениеградиентный бустингмашинное обучениеnanotransceiverplasmon enhancementgradient boostingmachine learningReferencesPlasmonic enhancement of molecular fluorescence near silver nanoparticles: theory, modeling, and experiment / D. V. Guzatov, S. V. Vaschenko, V. V. Stankevich [et al.] // Journal of Physical Chemistry C. – 2012. – Vol. 116, N 19. – P. 10723–10733. https://doi.org/10.1021/jp301598wGuzatov D. V., Vaschenko S. V., Stankevich V. V., Lunevich A. Y., Glukhov Y. F., Gaponenko S. V. Plasmonic enhancement of molecular fluorescence near silver nanoparticles: theory, modeling, and experiment. Journal of Physical Chemistry C, 2012, vol. 116, no. 19, pp. 10723–10733. https://doi.org/10.1021/jp301598wGaponenko, S. V. Colloidal plasmonics for active nanophotonics / S. V. Gaponenko, D. V. Guzatov // Proceedings of the IEEE. – 2020. – Vol. 108, N 5. – P. 704–720. https://doi.org/10.1109/jproc.2019.2958875Gaponenko S. V., Guzatov D. V. Colloidal plasmonics for active nanophotonics. Proceedings of the IEEE, 2020, vol. 108, no. 5, pp. 704–720. https://doi.org/10.1109/jproc.2019.2958875Farsad, M. An extended finite element/level set method to study surface effects on the mechanical behavior and properties of nanomaterials / M. Farsad, F. J. Vernerey, H. S. Park // International Journal for Numerical Methods in Engineerin. – 2010. – Vol. 84, N 12. – P. 1466–1489. https://doi.org/10.1002/nme.2946Farsad M., Vernerey F. J., Park H. S. An extended finite element/level set method to study surface effects on the mechanical behavior and properties of nanomaterials. International Journal for Numerical Methods in Engineerin, 2010, vol. 84, no. 12, pp. 1466–1489. https://doi.org/10.1002/nme.2946Plasmonic nanoparticle simulations and inverse design using machine learning / J. He, C. He, C. Zheng [et al.] // Nanoscale. – 2019. – Vol. 11, N 37. – P. 17444–17459. https://doi.org/10.1039/c9nr03450aHe J., He C., Zheng C., Wang Q., Ye J. Plasmonic nanoparticle simulations and inverse design using machine learning. Nanoscale, 2019, vol. 11, no. 37, pp. 17444–17459. https://doi.org/10.1039/c9nr03450aDeep learning in photonic device development: nuances and opportunities / V. Iyer, B. A. Wilson, Y. Chen [et al.] // npj Nanophotonics. – 2026. – Vol. 3. – Art. 5. https://doi.org/10.1038/s44310-025-00097-yIyer V., Wilson B. A., Chen Y., Kildishev A. V., Shalaev V. M., Boltasseva A. Deep learning in photonic device development: nuances and opportunities. npj Nanophotonics, 2026, vol. 3, art. 5. https://doi.org/10.1038/s44310-025-00097-yPhysics-informed deep learning for plasmonic sensing of nanoscale protein dynamics in solution / C. Wu, S. Yang, K. Zeng [et al.] // Science Advances. – 2025. – Vol. 11, N 39. – Art. eadw0783. https://doi.org/10.1126/sciadv.adw0783Wu C., Yang S., Zeng K., Dai X., Duan Y., Zhang S., Ma P., Guo X., Zhang S., Yang X., Dai Q. Physics-informed deep learning for plasmonic sensing of nanoscale protein dynamics in solution. Science Advances, 2025, vol. 11, no. 39, art. eadw0783. https://doi.org/10.1126/sciadv.adw0783Гапоненко, С. В. Оптические нанотрансиверы в фотонике / С. В. Гапоненко, Т. А. Ефимова // Доклады Национальной академии наук Беларуси. – 2024. – Т. 68, № 4. – С. 288–295. https://doi.org/10.29235/1561-8323-2024-68-4-288-295Gaponenko S. V., Efimova T. A. Optical nanotransceivers in photonics. Doklady Natsional’noi akademii nauk Belarusi = Doklady of the National Academy of Sciences of Belarus, 2024, vol. 68, no. 4, pp. 288–295 (in Russian). https://doi.org/10.29235/1561-8323-2024-68-4-288-295Scikit-learn: Machine learning in Python / F. Pedregosa, G. Varoquaux, A. Gramfort [et al.] // Journal of Machine Learning Research. – 2011. – Vol. 12. – P. 2825–2830. – URL: https://jmlr.csail.mit.edu/papers/v12/pedregosa11a.html (date of access: 02.09.2026).Pedregosa F., Varoquaux G., Gramfort A., Michel V., Thirion B., Grisel O., Blondel M., Prettenhofer P., Weiss R., Dubourg V., Vanderplas J., Passos A., Cournapeau D., Brucher M., Perrot M., Duchesnay É. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 2011, vol. 12, pp. 2825–2830. Available at: https://jmlr.csail.mit.edu/papers/v12/pedregosa11a.html (accessed 02 September 2026).Shields, M. D. The generalization of Latin hypercube sampling / M. D. Shields, J. Zhang // Reliability Engineering & System Safety. – 2016. – Vol. 148. – P. 96–108. https://doi.org/10.1016/j.ress.2015.12.002Shields M. D., Zhang J. The generalization of Latin hypercube sampling. Reliability Engineering & System Safety, 2016, vol. 148, pp. 96–108. https://doi.org/10.1016/j.ress.2015.12.002Friedman, J. H. Greedy function approximation: a gradient boosting machine / J. H. Friedman // Annals of Statistics. – 2001. – Vol. 29, N 5. – P. 1189–1232. https://doi.org/10.1214/aos/1013203451Friedman J. H. Greedy function approximation: a gradient boosting machine. Annals of Statistics, 2001, vol. 29, no. 5, pp. 1189–1232. https://doi.org/10.1214/aos/1013203451Measurement precision enhancement of surface plasmon resonance based angular scanning detection using deep learning / K. Thadson, S. Sasivimolkul, P. Suvarnaphaet [et al.] // Scientific Reports. – 2022. – Vol. 12. – Art. 2052. https://doi.org/10.1038/s41598-022-06065-2Thadson K., Sasivimolkul S., Suvarnaphaet P., Visitsattapongse S., Pechprasarn S. Measurement precision enhancement of surface plasmon resonance based angular scanning detection using deep learning. Scientific Reports, 2022, vol. 12, art. 2052. https://doi.org/10.1038/s41598-022-06065-2Machine learning assisted plasmonic metascreen for enhanced broadband absorption in ultra-thin silicon films / W. W. Ahmed, H. Cao, C. Xu [et al.] // Light: Science & Applications. – 2025. – Vol. 14. – Art. 42. https://doi.org/10.1038/s41377-024-01723-8Ahmed W. W., Cao H., Xu C., Farhat M., Amin M., Li X., Zhang X., Wu Y. Machine learning assisted plasmonic metascreen for enhanced broadband absorption in ultra-thin silicon films. Light: Science & Applications, 2025, vol. 14, art. 42. https://doi.org/10.1038/s41377-024-01723-8Deep neural network for plasmonic sensor modeling / X. Li, J. Shu, W. Gu, L. Gao // Optical Materials Express. – 2019. – Vol. 9, N 9. – P. 3857–3862. https://doi.org/10.1364/ome.9.003857Li X., Shu J., Gu W., Gao L. Deep neural network for plasmonic sensor modeling. Optical Materials Express, 2019, vol. 9, no. 9, pp. 3857–3862. https://doi.org/10.1364/ome.9.003857

The authors declare that there are no conflicts of interest present.