Principled Design of Diffusion-based Optimizers for Inverse Problems
For practitioners deploying diffusion models on inverse problems, this work reduces the need for per-task hyperparameter tuning and speeds up inference.
The paper proposes principled reparameterizations for diffusion-based inverse problem solvers that enable hyperparameter reuse across tasks without retuning, and introduces OptDiff, a pipeline that accelerates inference by integrating convex optimization tools. Experiments on image reconstruction, deblurring, and super-resolution achieve substantial speedups and improved image quality.
Score-based diffusion models achieve state-of-the-art performance for inverse problems, but their practical deployment is hindered by long inference times and cumbersome hyperparameter tuning. While pretrained diffusion models can be reused across tasks without retraining, inference-time hyperparameters such as the noise schedule and posterior sampling weights typically require ad-hoc adjustment for each problem setup. We propose principled reparameterizations that induce invariances, allowing the same hyperparameters to be reused across multiple problems without re-tuning. In addition, building on the RED-diff framework, which reformulates posterior sampling as an optimization problem, we further develop the OptDiff pipeline. OptDiff provides a simplified tuning framework that facilitates the integration of convex optimization tools to accelerate inference. Experiments on image reconstruction, deblurring, and super-resolution show substantial speedups and improved image quality.