Inference-Time Search using Side Information for Diffusion-based Image Reconstruction

ICLR 2026 Conference Submission15209 Authors

Keywords: Diffusion Models, Inverse Problems, Inference-Time Search, Side Information, Exploration–Exploitation, Gradient-Free Guidance, Ill-Posed Problems, Generative Priors

TL;DR: This work introduces an inference-time search framework that leverages side information to guide diffusion models for solving inverse problems, enabling more accurate and robust reconstructions than existing gradient-based approaches.

Abstract:

Diffusion models have emerged as powerful priors for solving inverse problems.
However, existing approaches typically overlook side information that could significantly improve reconstruction quality, especially in severely ill-posed settings.
In this work, we propose a novel inference-time search algorithm that guides the
sampling process using the side information in a manner that balances exploration
and exploitation. This enables more accurate and reliable reconstructions, providing an alternative to the gradient-based guidance that is prone to reward-hacking
artifacts. Our approach can be seamlessly integrated into a wide range of existing
diffusion-based image reconstruction pipelines. Through extensive experiments on
a number of inverse problems, such as box inpainting, super-resolution, and various deblurring tasks including motion, Gaussian, nonlinear, and blind deblurring,
we show that our approach consistently improves the qualitative and quantitative
performance of diffusion-based image reconstruction algorithms. We also show
the superior performance of our approach with respect to other baselines, including
reward gradient-based guidance algorithms.

Primary Area: generative models

Submission Number: 15209