Entity · paper

Exact Posterior Score Estimation for Solving Linear Inverse Problems

paperactiveexact-posterior-score-estimation-for-solving-linear-inverse-problems-13c49ffe·1 events·first seen Jun 16, 2026

Aliases: Exact Posterior Score Estimation for Solving Linear Inverse Problems

Co-occurring entities

FFHQ ImageNet Exact Posterior Score

More like this (12)

Exact Posterior Score posterior predictive variance minimization imaging inverse problems finite-sample posterior sampling framework diffusion posterior sampling Logit-Contribution Scoring low-rank subspace projection Subspace Projection Posterior Refinement Hájek estimator Orthogonal Residual Projection Complexity Bounds and Approaches to Learning Projected Gradient Descent Solver Iterates

Recent events (1)

5arXiv · cs.LG·Jun 16, 2026·source ↗

Exact Posterior Score (EPS): Closed-form posterior sampling for linear inverse problems with diffusion models

A new arXiv preprint derives the exact posterior score in closed form for linear Gaussian inverse problems under general Gaussian interpolants, showing that posterior sampling reduces to a denoising problem at an operator-dependent shifted pivot under anisotropic noise covariance. The authors convert this identity into a training objective called Exact Posterior Score (EPS) that preserves the input/output structure of standard diffusion pretraining, enabling training from scratch or fine-tuning from a pretrained denoiser. EPS is evaluated on five linear inverse problems across FFHQ and ImageNet, outperforming both training-free and training-based baselines while requiring roughly an order of magnitude fewer denoiser evaluations than gradient-based posterior samplers.

Evaluation and Benchmarking Exact Posterior Score Estimation for Solving Linear Inverse Problems FFHQ ImageNet +1 more