S³-PnP
A plug-and-play ADMM framework driven by self-supervised score priors for scientific imaging denoising. It mitigates the dual mismatch between the target distribution and intermediate-state distributions by learning score priors on the intermediate states of the optimization iterations.
Timeline (last 90 days)
The core operation of the S³-PnP framework is to learn a score prior directly on the observation-driven intermediate states of ADMM, requiring neither clean training targets nor external pretrained weights.
The score function learned in S³-PnP is used to correct the input of the analytic prior update, combining the observation-driven correction term with analytic regularization, so that the prior emerges naturally from the current observation and the ADMM iteration process rather than being borrowed externally.
S³-PnP's experiments cover three tasks: reference-guided hyperspectral restoration, single-image hyperspectral denoising, and MRI denoising, and per the paper's report achieve consistent improvements over competing methods.
All Facts (3)
The core operation of the S³-PnP framework is to learn a score prior directly on the observation-driven intermediate states of ADMM, requiring neither clean training targets nor external pretrained weights.
50%UnverifiedThe score function learned in S³-PnP is used to correct the input of the analytic prior update, combining the observation-driven correction term with analytic regularization, so that the prior emerges naturally from the current observation and the ADMM iteration process rather than being borrowed externally.
50%UnverifiedS³-PnP's experiments cover three tasks: reference-guided hyperspectral restoration, single-image hyperspectral denoising, and MRI denoising, and per the paper's report achieve consistent improvements over competing methods.
50%