For Gaussian image priors, the exact-likelihood CGDM algorithm achieves the smallest exact 2-Wasserstein distance to the true conditional distribution, while DPS and PiGDM show measurable bias.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
baseline 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
baseline 1polarities
unclear 1representative citing papers
citing papers explorer
-
Exact Evaluation of the Accuracy of Diffusion Models for Inverse Problems with Gaussian Data Distributions
For Gaussian image priors, the exact-likelihood CGDM algorithm achieves the smallest exact 2-Wasserstein distance to the true conditional distribution, while DPS and PiGDM show measurable bias.