Noise2Same: Optimizing A Self-Supervised Bound for Image Denoising

Neural Information Processing Systems 

These studies can be divided into two categories according to the amount of extra information required. Methods in the first category requires the noise model to be known.