keyword

neural stochastic differential equations

Neural stochastic differential equations are continuous-time machine learning models that parameterize the drift and diffusion components of a stochastic differential equation using neural networks. As a stochastic extension of neural ordinary differential equations, they model complex dynamical systems by combining deterministic state evolution with random fluctuations driven by continuous noise processes such as Brownian motion. By using neural network architectures to learn both the expected trajectory and the magnitude of underlying randomness directly from data, these models are widely used for continuous-time generative modeling, simulating physical phenomena with inherent variability, and analyzing irregularly sampled time-series data while capturing system uncertainty.

1 item