← Search

Julien Hermant

2 accepted papers

2026

Provably Accelerated Imaging with Restarted Inertia and Score-based Image Priors

ICLR 2026poster

Fast convergence and high-quality image recovery are two essential features of algorithms for solving ill-posed imaging inverse problems. Existing methods, such as regularization by denoising (RED), often focus on designing sophisticated image priors to improve reconstruction quality, while leaving…

Cited by 0SourcecodeScholar
2025

Gradient correlation is a key ingredient to accelerate SGD with momentum

ICLR 2025poster

Empirically, it has been observed that adding momentum to Stochastic Gradient Descent (SGD) accelerates the convergence of the algorithm. However, the literature has been rather pessimistic, even in the case of convex functions, about the possibility of theoretically proving this observation. We inv…