Keywords: score-based generative modelling, diffusion model, reverse SDE, minimal data assumptions
TL;DR: Polynomial convergence results for denoising diffusion models with minimal data assumptions
Abstract: We give polynomial convergence guarantees for denoising diffusion models that do not rely on the data distribution satisfying functional inequalities or strong smoothness assumptions. Assuming a $L^2$-accurate score estimate, we obtain Wasserstein distance guarantees for any distributions of bounded support or sufficiently decaying tails, as well as TV guarantees for distributions with further smoothness assumptions.
Student Paper: No