EUVIMEDDécouvrir EUVIMED
Crossref

Stable and efficient learning of degenerate stochastic differential equation in neural networks

Zhiding Wang, Ming Yi, Ting Gao iD, Xiaoli Chen iD

Chaos: An Interdisciplinary Journal of Nonlinear Science · 2026-03-01 · 36 (3)

DOI: 10.1063/5.0301167 ↗

At a glance

Journal article in Chaos: An Interdisciplinary Journal of Nonlinear Science, published online on 2026-03-03. Written by Zhiding Wang and 3 co-authors. Institution: School of Mathematics and Physics, China University of Geosciences, Wuhan 430074, China. 55 references. Funded by: National Natural Science Foundation of China, Natural Science Foundation of Hubei Province, Natural Science Foundation of Wuhan.

RésuméRéférences bibliographiquesCiteRelated studies

Abstract (original language)

Stochastic differential equations are fundamental for modeling complex dynamic systems subject to random noise. However, learning stochastic dynamics from empirical data remains challenging, particularly under degenerate noise conditions where the diffusion matrix is irreversible. Traditional approaches often suffer from numerical instability and prohibitive computational costs in such scenarios. To address these limitations, we propose three novel algorithms tailored for efficient and stable learning of SDEs with degenerate diffusion. Algorithm I integrates numerical techniques from second-order stochastic differential equation (SDE) solvers into a specialized learning framework, targeting the special case of a two-dimensional degenerate stochastic equation. This approach improves computational efficiency by taking advantage of the properties of the physical system. Algorithm II introduces a novel loss function that combines mean squared error and maximum likelihood estimation, specifically designed for systems with degenerate diffusion matrices, improving robustness by explicitly accounting for noise structure in the optimization process. Algorithm III incorporates a stabilizing auxiliary noise mechanism during training, which mitigates gradient instability without sacrificing convergence guarantees. Extensive numerical experiments demonstrate that the proposed methods significantly outperform conventional maximum likelihood estimation in terms of training stability, faster convergence, and higher accuracy. These advances enable reliable modeling of high-dimensional stochastic systems and offer promising tools for neural network learning tasks requiring robust noise adaptation.

Authors and institutions

  • School of Mathematics and Physics, China University of Geosciences, Wuhan 430074, China.
  • School of Mathematics and Statistics, Huazhong University of Science and Technology, Wuhan 430074, China.
  • Shenzhen Research Institute of China University of Geosciences, Shenzhen 518057, China.
  • Chongqing CUG Industrial Technology Research Institute Co., Ltd., Chongqing 400015, China.

ORCID profiles: Ting Gao, Xiaoli Chen

Funding

  • National Natural Science Foundation of China
  • Natural Science Foundation of Hubei Province
  • Natural Science Foundation of Wuhan
  • Guangdong Provincial Key Laboratory of Mathematical and Neural Dynamical Systems
  • Natural Science Foundation of Chongqing Municipality
  • Basic and Applied Basic Research Foundation of Guangdong Province
  • CUG Scholar Scientific Research Funds at China University of Geoscience
  • Fundamental Research Funds for the Central Universities, China University of Geoscience

Cite

APA

Wang, Z., Yi, M., Gao, T., & Chen, X. (2026). Stable and efficient learning of degenerate stochastic differential equation in neural networks. Chaos: An Interdisciplinary Journal of Nonlinear Science, 36(3). https://doi.org/10.1063/5.0301167

Vancouver

Wang Z, Yi M, Gao T, Chen X. Stable and efficient learning of degenerate stochastic differential equation in neural networks. Chaos: An Interdisciplinary Journal of Nonlinear Science. 2026;36(3). doi:10.1063/5.0301167

BibTeX
@article{Wang2026,
  author = {Wang, Zhiding and Yi, Ming and Gao, Ting and Chen, Xiaoli},
  title = {Stable and efficient learning of degenerate stochastic differential equation in neural networks},
  journal = {Chaos: An Interdisciplinary Journal of Nonlinear Science},
  year = {2026},
  volume = {36},
  number = {3},
  doi = {10.1063/5.0301167}
}
Start for free →