Preprint
Jul 2026
Normalized First-Order Methods for Convex (L0, L1)-Smooth Optimization with Inexact Gradients
This work develops comparison-oracle variants of Normalized Gradient Descent and Gradient Descent with Polyak stepsizes and establishes explicit upper bounds on the approximation error that guarantee convergence and derive convergence rates for all proposed methods.
E. Kovalev, F. Stonyakin
· 0 citations