Active Diffusion-Based Solver for Inverse Problems with Incomplete Priors

This method uses a diffusion model to iteratively detect and correct model misspecification through posterior uncertainty. It enables robust parameter estimation even when initial training bounds exclude the true parameter space, providing a Bayesian framework for adaptive domain augmentation.