Only the gradient component of score errors affects marginal distributions in diffusion models, so L2 error can be arbitrarily large with perfect match; this yields an impossibility result, a gradient-only KL bound, and a Sobolev estimator that correlates better with quality.
Convergence of denoising diffusion models under the manifold hypothesis
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3verdicts
UNVERDICTED 3representative citing papers
SiLD is a score-matching framework that learns both manifold projection and intrinsic density from a single objective, with proven sample complexity depending only on intrinsic dimension.
i-DEQ adds momentum to DEQ fixed-point iterations, yielding convergence guarantees, training stability, and halved inference time while matching state-of-the-art reconstruction quality on inverse problems.
citing papers explorer
-
Diffusion Models Observe Only Gradients: A Geometric Perspective on Score Matching Errors
Only the gradient component of score errors affects marginal distributions in diffusion models, so L2 error can be arbitrarily large with perfect match; this yields an impossibility result, a gradient-only KL bound, and a Sobolev estimator that correlates better with quality.
-
Provably Learning Diffusion Models under the Manifold Hypothesis: Collapse and Refine
SiLD is a score-matching framework that learns both manifold projection and intrinsic density from a single objective, with proven sample complexity depending only on intrinsic dimension.
-
i-DEQ: A stable inertial deep equilibrium model for image restoration
i-DEQ adds momentum to DEQ fixed-point iterations, yielding convergence guarantees, training stability, and halved inference time while matching state-of-the-art reconstruction quality on inverse problems.