AI & ML Breaks Assumption

Proves that image denoisers can be strictly contractive (robust to noise) without sacrificing state-of-the-art restoration quality.

March 30, 2026

Original Paper

Provably Contractive and High-Quality Denoisers for Convergent Restoration

Shubhi Shukla, Pravin Nair

arXiv · 2603.26168

The Takeaway

It challenges the long-held 'robustness-accuracy trade-off' in vision models by using Lipschitz-controlled refinements and unfolding techniques. This provides a roadmap for building 'Plug-and-Play' restoration algorithms that are both high-performing and mathematically guaranteed to converge.

From the abstract

Image restoration, the recovery of clean images from degraded measurements, has applications in various domains like surveillance, defense, and medical imaging. Despite achieving state-of-the-art (SOTA) restoration performance, existing convolutional and attention-based networks lack stability guarantees under minor shifts in input, exposing a robustness accuracy trade-off. We develop provably contractive (global Lipschitz $< 1$) denoiser networks that considerably reduce this gap. Our design co