Standard

Reinforcement learning-based optimal control for stochastic opinion dynamics. / Chen, Yajin; Gao, Hongwei; Mazalov, Vladimir V.; Liu, Yanshan.

в: Scientific Reports, Том 16, № 1, 12392 , 06.03.2026.

Результаты исследований: Научные публикации в периодических изданиях › статья › Рецензирование

Harvard

Chen, Y, Gao, H, Mazalov, VV & Liu, Y 2026, 'Reinforcement learning-based optimal control for stochastic opinion dynamics', Scientific Reports, Том. 16, № 1, 12392 . https://doi.org/10.1038/s41598-026-42646-1

APA

Chen, Y., Gao, H., Mazalov, V. V., & Liu, Y. (2026). Reinforcement learning-based optimal control for stochastic opinion dynamics. Scientific Reports, 16(1), [12392 ]. https://doi.org/10.1038/s41598-026-42646-1

Vancouver

Chen Y, Gao H, Mazalov VV, Liu Y. Reinforcement learning-based optimal control for stochastic opinion dynamics. Scientific Reports. 2026 Март 6;16(1). 12392 . https://doi.org/10.1038/s41598-026-42646-1

Author

Chen, Yajin ; Gao, Hongwei ; Mazalov, Vladimir V. ; Liu, Yanshan. / Reinforcement learning-based optimal control for stochastic opinion dynamics. в: Scientific Reports. 2026 ; Том 16, № 1.

BibTeX

@article{e174326ce604421d9c3ce7ef69241838,
title = "Reinforcement learning-based optimal control for stochastic opinion dynamics",
abstract = "This paper proposes a integrated framework for optimal control of opinion dynamics in social networks, addressing three progressively challenging scenarios: Model-based stochastic control, where agent interactions follow known probability distributions, enabling analytical optimal policies; Model-free Reinforcement Learning (RL), where interaction randomness has unknown distributions but system dynamics are preserved; Data-driven RL for unknown systems, where time-varying network dynamics (with stochasticity constraints) are fully unknown, requiring purely observational learning. By designing an RL control framework grounded in convex quadratic optimization, we bridge model-based control and data-driven learning, offering new insights for social network manipulation and multi-agent coordination. Numerical simulations demonstrate the framework{\textquoteright}s effectiveness.",
keywords = "Bellman equation, Opinion dynamics, Optimal control, Policy iteration, Reinforcement learning",
author = "Yajin Chen and Hongwei Gao and Mazalov, {Vladimir V.} and Yanshan Liu",
year = "2026",
month = mar,
day = "6",
doi = "10.1038/s41598-026-42646-1",
language = "English",
volume = "16",
journal = "Scientific Reports",
issn = "2045-2322",
publisher = "Nature Publishing Group",
number = "1",

}

RIS

TY - JOUR

T1 - Reinforcement learning-based optimal control for stochastic opinion dynamics

AU - Chen, Yajin

AU - Gao, Hongwei

AU - Mazalov, Vladimir V.

AU - Liu, Yanshan

PY - 2026/3/6

Y1 - 2026/3/6

N2 - This paper proposes a integrated framework for optimal control of opinion dynamics in social networks, addressing three progressively challenging scenarios: Model-based stochastic control, where agent interactions follow known probability distributions, enabling analytical optimal policies; Model-free Reinforcement Learning (RL), where interaction randomness has unknown distributions but system dynamics are preserved; Data-driven RL for unknown systems, where time-varying network dynamics (with stochasticity constraints) are fully unknown, requiring purely observational learning. By designing an RL control framework grounded in convex quadratic optimization, we bridge model-based control and data-driven learning, offering new insights for social network manipulation and multi-agent coordination. Numerical simulations demonstrate the framework’s effectiveness.

AB - This paper proposes a integrated framework for optimal control of opinion dynamics in social networks, addressing three progressively challenging scenarios: Model-based stochastic control, where agent interactions follow known probability distributions, enabling analytical optimal policies; Model-free Reinforcement Learning (RL), where interaction randomness has unknown distributions but system dynamics are preserved; Data-driven RL for unknown systems, where time-varying network dynamics (with stochasticity constraints) are fully unknown, requiring purely observational learning. By designing an RL control framework grounded in convex quadratic optimization, we bridge model-based control and data-driven learning, offering new insights for social network manipulation and multi-agent coordination. Numerical simulations demonstrate the framework’s effectiveness.

KW - Bellman equation

KW - Opinion dynamics

KW - Optimal control

KW - Policy iteration

KW - Reinforcement learning

UR - https://www.mendeley.com/catalogue/2d473d3b-3315-331e-a9ce-9ea698e9884f/

U2 - 10.1038/s41598-026-42646-1

DO - 10.1038/s41598-026-42646-1

M3 - Article

C2 - 41792364

VL - 16

JO - Scientific Reports

JF - Scientific Reports

SN - 2045-2322

IS - 1

M1 - 12392

ER -

ID: 150015116