Neural Diffusion Models

arXiv:2310.08337v2 Announce Type: replace
Abstract: Diffusion models have shown remarkable performance on many generative tasks. Despite recent success, most diffusion models are restricted in that they only allow linear transformation of the data distribution. In contrast, broader family of transformations can potentially help train generative distributions more efficiently, simplifying the reverse process and closing the gap between the true negative log-likelihood and the variational approximation. In this paper, we present Neural Diffusion Models (NDMs), a generalization of conventional diffusion models …

abstract arxiv contrast cs.lg data diffusion diffusion models distribution family gap generative linear performance process simplifying stat.ml success tasks train transformation type