Reproduce “Kinetic Langevin Meets Split Gibbs: Accelerated Posterior Sampling for Imaging Inverse Problems with Diffusion Priors” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dai Hai Nguyen, Duc Dung Nguyen
The authors introduce RED-KLwSGS, a split Gibbs sampler for image restoration that adds momentum to the prior-side update while retaining a Gaussian data-side update and one denoiser evaluation per iteration. They prove convergence under smoothness and strong-convexity assumptions. In tests, it improves iteration-wise progress over RED-LwSGS; one motion-deblurring example reaches 26.5 dB in 31.5 seconds, not a general timing guarantee.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.