A Stability Benchmark of Generative Regularizers for Inverse Problems
For scientific and medical imaging, this work provides a critical stability benchmark for generative priors, highlighting their limitations under imperfect settings.
This paper evaluates the stability of generative (diffusion) priors for inverse problems in imaging, benchmarking them against optimization-based methods. Results show where generative priors achieve state-of-the-art reconstructions and where they fail or are problematic.
Generative (diffusion) priors demonstrate remarkable performance in addressing inverse problems in imaging. Yet, for scientific and medical imaging, it is crucial that reconstruction techniques remain stable and reliable under imperfect settings. Typical definitions of stability encompass the notion of ''convergent regularization'', robustness to out-of-distribution data, and to inaccuracies in the forward operator or noise model. We evaluate these properties numerically. Furthermore, we benchmark generative approaches against modern optimization-based methods inspired by the widely used variational techniques. Our results give insights for which settings and applications generative priors can deliver state-of-the-art reconstructions, and on those in which they fall short or may even be problematic.