Uncertainty Quantification via Neural Posterior Principal Components

Uncertainty quantification is crucial for the deployment of image restoration models in safety-critical domains, like autonomous driving and biological imaging.