Variance-Reduced Diffusion Sampling via Target Score Identity

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Duston, Alois, Bui-Thanh, Tan
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915749260427264
author Duston, Alois
Bui-Thanh, Tan
author_facet Duston, Alois
Bui-Thanh, Tan
contents We study variance reduction for score estimation and diffusion-based sampling in settings where the clean (target) score is available or can be approximated. Starting from the Target Score Identity (TSI), which expresses the noisy marginal score as a conditional expectation of the target score under the forward diffusion, we develop: (i) a plug-and-play nonparametric self-normalized importance sampling estimator compatible with standard reverse-time solvers, (ii) a variance-minimizing \emph{state- and time-dependent} blending rule between Tweedie-type and TSI estimators together with an anti-correlation analysis, (iii) a data-only extension based on locally fitted proxy scores, and (iv) a likelihood-tilting extension to Bayesian inverse problems. We also propose a \emph{Critic--Gate} distillation scheme that amortizes the state-dependent blending coefficient into a neural gate. Experiments on synthetic targets and PDE-governed inverse problems demonstrate improved sample quality for a fixed simulation budget.
format Preprint
id arxiv_https___arxiv_org_abs_2601_01594
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Variance-Reduced Diffusion Sampling via Target Score Identity
Duston, Alois
Bui-Thanh, Tan
Machine Learning
68T07, 65C05, 60J60, 62F15
G.3; I.2.6; I.5.1
We study variance reduction for score estimation and diffusion-based sampling in settings where the clean (target) score is available or can be approximated. Starting from the Target Score Identity (TSI), which expresses the noisy marginal score as a conditional expectation of the target score under the forward diffusion, we develop: (i) a plug-and-play nonparametric self-normalized importance sampling estimator compatible with standard reverse-time solvers, (ii) a variance-minimizing \emph{state- and time-dependent} blending rule between Tweedie-type and TSI estimators together with an anti-correlation analysis, (iii) a data-only extension based on locally fitted proxy scores, and (iv) a likelihood-tilting extension to Bayesian inverse problems. We also propose a \emph{Critic--Gate} distillation scheme that amortizes the state-dependent blending coefficient into a neural gate. Experiments on synthetic targets and PDE-governed inverse problems demonstrate improved sample quality for a fixed simulation budget.
title Variance-Reduced Diffusion Sampling via Target Score Identity
topic Machine Learning
68T07, 65C05, 60J60, 62F15
G.3; I.2.6; I.5.1
url https://arxiv.org/abs/2601.01594