Direct finite-horizon high-probability analysis of normalized two-point Gaussian zeroth-order gradient descent yields explicit query complexities for strongly convex, convex, and non-convex smooth objectives under smoothing-radius conditions.
A unified zeroth-order optimization framework via oblivious randomized sketching
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
math.OC 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Establishes high-probability bounds for zeroth-order GD showing logarithmic dependence on failure probability δ in deterministic case and specific query complexity in stochastic case.
citing papers explorer
-
High-Probability Guarantees for Random Zeroth-Order Gradient Descent on Smooth Functions
Direct finite-horizon high-probability analysis of normalized two-point Gaussian zeroth-order gradient descent yields explicit query complexities for strongly convex, convex, and non-convex smooth objectives under smoothing-radius conditions.
-
High-Probability Guarantees for Random Zeroth-Order (Stochastic) Gradient Descent
Establishes high-probability bounds for zeroth-order GD showing logarithmic dependence on failure probability δ in deterministic case and specific query complexity in stochastic case.