Neural SDEs

Neural Stochastic Differential Equations (Neural SDEs) are a class of models using neural networks to parameterize the drift and diffusion terms of stochastic differential equations, enabling the modeling of continuous-time stochastic processes. Current research focuses on improving training efficiency through methods like finite dimensional matching and novel scoring rules, as well as developing stable architectures for handling irregular or missing data in real-world time series. These models find applications in diverse fields, including finance, healthcare (e.g., disease progression modeling), and physics, offering advantages in handling uncertainty and generating realistic simulations of complex systems. The ability to incorporate prior knowledge and quantify uncertainty makes Neural SDEs a powerful tool for various scientific and engineering problems.