AI RESEARCH
Neural Galerkin Normalizing Flow for Transition Probability Density Functions of Diffusion Models
arXiv CS.LG
•
ArXi:2603.18907v1 Announce Type: new We propose a new Neural Galerkin Normalizing Flow framework to approximate the transition probability density function of a diffusion process by solving the corresponding Fokker-Planck equation with an atomic initial distribution, parametrically with respect to the location of the initial mass.