First-order asymptotic expansions of weak and Fréchet discretization errors in diffusion sampling are derived, explicit under Gaussian data through covariance geometry and robust to other data geometries.
Title resolution pending
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
A preconditioned regularized Wasserstein proximal sampling algorithm is introduced for particle-based approximation of Gibbs distributions, featuring a PDE-derived kernel formulation and non-asymptotic convergence analysis for quadratic potentials.
P-Flow solves linear inverse problems by optimizing the flow's source latent with a proxy gradient and a Gaussian-sphere projection, matching or beating prior restoration methods at far lower cost.
UTOPY trains unrolling algorithms for ill-posed inverse problems via a fidelity homotopy path from synthetic well-posed to real ill-posed sensing operators, yielding up to 2.5 dB PSNR gains.
citing papers explorer
-
Geometry-Aware Discretization Error of Diffusion Models
First-order asymptotic expansions of weak and Fréchet discretization errors in diffusion sampling are derived, explicit under Gaussian data through covariance geometry and robust to other data geometries.
-
Preconditioned Regularized Wasserstein Proximal Sampling
A preconditioned regularized Wasserstein proximal sampling algorithm is introduced for particle-based approximation of Gibbs distributions, featuring a PDE-derived kernel formulation and non-asymptotic convergence analysis for quadratic potentials.
-
P-Flow: Proxy-gradient Flows for Linear Inverse Problems
P-Flow solves linear inverse problems by optimizing the flow's source latent with a proxy gradient and a Gaussian-sphere projection, matching or beating prior restoration methods at far lower cost.
-
UTOPY: Unrolling Algorithm Learning via Fidelity Homotopy for Inverse Problems
UTOPY trains unrolling algorithms for ill-posed inverse problems via a fidelity homotopy path from synthetic well-posed to real ill-posed sensing operators, yielding up to 2.5 dB PSNR gains.