Provable diffusion-based posterior sampling for linear inverse problems via DDIM

By Yuchen Jiao · Paper · cs.LG

Diffusion-based methods have achieved remarkable empirical success in solving inverse problems. However, many existing posterior samplers either lack rigorous theoretical guarantees or incur substantial computational overhead. We propose a simple and efficient algorithm, called \

Cs.lg

View original

HomeResourceLoading…