Conditioning diffusion models by explicit forward-backward bridging

2024/05/22 by Adrien Corenflos, Corenflos, Adrien, Zheng Zhao +7 · 3 citations
Computer Science · Physics and Astronomy · #Advanced Mathematical Modeling in Engineering #Computation (stat.CO) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Methodology (stat.ME) #Model Reduction and Neural Networks

paper · pdf · doi:10.48550/arxiv.2405.13794

openalex publication_date 2024/05/22 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Given an unconditional diffusion model targeting a joint model π(x, y), using it to perform conditional simulation π(x | y) is still largely an open question and is typically achieved by learning conditional drifts to the denoising SDE after the fact. In this work, we express exact conditional simulation within the approximate diffusion model as an inference problem on an augmented space corresponding to a partial SDE bridge. This perspective allows us to implement efficient and principled particle Gibbs and pseudo-marginal samplers marginally targeting the conditional distribution π(x | y). Contrary to existing methodology, our methods do not introduce any additional approximation to the unconditional diffusion model aside from the Monte Carlo error. We showcase the benefits and drawbacks of our approach on a series of synthetic and real data examples.

Cited by

• Efficient Zero-Shot Inpainting with Decoupled Diffusion Guidance
• Diffusion differentiable resampling
• Nonlinear filtering based on density approximation and deep BSDE prediction

Related

• ConDiSim: Conditional Diffusion Models for Simulation Based Inference
• Diffusion models as plug-and-play priors
• Conditional Simulation Using Diffusion Schrödinger Bridges
• Unveil Conditional Diffusion Models with Classifier-free Guidance: A Sharp Statistical Theory
• Non-asymptotic convergence bound of conditional diffusion models