A new local time-decoupled squared Wasserstein-2 method for training stochastic neural networks to reconstruct uncertain parameters in dynamical systems.

Xia M., Shen Q., Maini PK., Gaffney EA., Mogilner A.

In this work, we propose and analyze a new local time-decoupled squared Wasserstein-2 method for reconstructing the distribution of unknown parameters in dynamical systems from a finite number of observed temporal trajectories. Specifically, we show that a stochastic neural network model, which can be effectively trained by minimizing our proposed local time-decoupled squared Wasserstein-2 loss function, is an effective model for approximating the distribution of uncertain model parameters in dynamical systems. Through several numerical examples, we showcase the effectiveness of our proposed method in reconstructing the distribution of parameters in different dynamical systems.

DOI

10.1016/j.neunet.2025.107893

Type

Journal article

Publication Date

2026-01-01T00:00:00+00:00

Volume

193

Keywords

Dynamical systems, Stochastic neural network, Uncertainty quantification, Wasserstein distance, Neural Networks, Computer, Stochastic Processes, Uncertainty, Time Factors, Algorithms

Permalink More information Close