Score-based generative models break the curse of dimensionality in learning a family of sub-Gaussian distributions.

Frank Cole, Yulong Lu

Browse the full ICLR paper archive.