AI RESEARCH
Distributional Shrinkage I: Universal Denoiser Beyond Tweedie's Formula
arXiv CS.LG
•
ArXi:2511.09500v4 Announce Type: replace-cross We study the problem of denoising when only the noise level is known, not the noise distribution. Independent noise $Z$ corrupts a signal $X$, yielding the observation $Y = X + \sigma Z$ with known $\sigma \in (0,1)$. We propose \emph{universal} denoisers, agnostic to both signal and noise distributions, that recover the signal distribution $P_X$ from $P_Y