Diffusion might be a better way to model randomness in probabilistic programming

A new paper uses diffusion models to approximate the probability distributions for faster, more accurate automated inference

automated diffusion diffusion models faster inference paper probability programming randomness