Latent diffusion model guidance improves inverse problem reconstruction quality
ANaLOG: Anisotropic Native-Latent Operator Guidance for Solving Inverse Problems
Computer Vision and Pattern Recognition
Summary
Inverse problems involve figuring out the original data from a limited or noisy measurement, which is hard and slow to solve. The authors propose ANaLOG, a technique that uses smarter calculations in a compressed data space to guide the solution process more efficiently. They consider that some directions in this data space are more reliable than others and adapt their guidance accordingly. Their method improves the quality of reconstructed images while being faster than prior approaches.
What this means in practice
- •For medical imaging developers: Enhance image reconstruction from limited scan data by guiding latent diffusion-based models with uncertainty-aware weighting.
- •For industrial defect inspection teams: Improve reconstruction accuracy when diagnosing material defects with latent diffusion methods that adjust guidance according to measurement reliability.
Authors
Darshan Thaker, Lachlan Ewen MacDonald, René Vidal
Abstract
Native-latent guidance is a recent paradigm for solving inverse problems with latent diffusion models. It replaces repeated evaluations of the image-space forward model, each requiring a decoder pass, with efficient guidance computed using a learned latent-space surrogate. However, existing methods apply guidance uniformly across latent dimensions, ignoring that measurements are informative only along certain directions and that the reliability of model predictions varies across inputs and timesteps. We propose ANaLOG, a framework for efficient uncertainty-aware guidance with pretrained latent diffusion models. ANaLOG models uncertainty by learning an anisotropic, input- and time-dependent covariance that is integrated into the guidance mechanism to emphasize reliable directions and downweight uncertain ones. We theoretically analyze this framework in a linear model setting and prove that anisotropic, uncertainty-aware weighting is necessary for correct sampling, whereas isotropic guidance induces sampling errors. Experiments across five challenging inverse problems show that ANaLOG improves perceptual reconstruction quality over existing methods while preserving efficiency.