Shared Prior Learning of Energy-Based Models for Image Reconstruction

We propose a novel learning-based framework for image reconstruction particularly designed for training without ground truth data, which has three major building blocks: energy-based learning, a pa...