{"id":"arxiv/2605.09718","url":"https://arxiv.org/abs/2605.09718","tags":["stat.ML","cs.LG","math.PR","math.ST"],"score":0,"title":"Learning stochastic multiscale models through normalizing flows","domain":"stat.ML","source":"arxiv","authors":["Anan Saha","Arnab Ganguly"],"reviews":0,"subject":"stat.ML","summary":"Many systems in physics, engineering, and biology exhibit multiscale stochastic dynamics, where low-dimensional slow variables evolve under the influence of high-dimensional fast processes. In practice, observations are often limited to a single trajectory of the slow component, while the fast dynamics remain unobserved, making statistical learning challenging. Approaches based on partial differential equations (PDE), such as Fokker-Planck formulations, aim to characterize the evolution of probability densities, typically requiring dense space-time data or grid-based solvers. In contrast, we…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-05-10","reactions":{}}