Get Started
Research questionHow can DDPM sampling error be controlled for variance schedules beyond reverse-SDE discretizations?DDPM sampling is commonly analyzed by discretizing its reverse SDE, which can restrict the variance schedules and parameter choices considered. Relating DDPM sampling to the discretization of a Föllmer process may provide a broader framework for understanding these choices and their resulting sampling errors.
Diffusion Models
Statistical Machine Learning
Latest papersRecent research connected to this question, newest first.Connections between the Föllmer process and the denoising diffusion probabilistic modelThe source studies the Föllmer process as a Brownian motion conditioned on a prescribed terminal distribution and its connection to DDPM reverse-SDE sampling. It analyzes direct Föllmer-process discretizations, derives corresponding DDPM hyperparameter settings, accommodates a broader class of variance schedules, and recovers or slightly improves existing sampling-error bounds.research paper · Sep 2, 2026
Related questions
How can variance-inflated Thompson Sampling be analyzed for stochastic generalized linear bandits without tractable posterior approximations?How can random-feature spaces achieve spectral accuracy for multidimensional targets and elliptic PDEs while controlling ill-conditioning?How can federated PDE learning define and measure client heterogeneity across different equations?How can smooth density estimation preserve accuracy under relaxed local differential privacy?
Home
Topics
Search
Library