High-Probability Guarantees for Random Zeroth-Order Gradient Descent

Haishan Ye · 2026

Read High-Probability Guarantees for Random Zeroth-Order Gradient Descent by Haishan Ye (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-23613.txt.

Description

Ye establishes convergence guarantees with high probability for random zeroth-order gradient descent, which optimises using only function values rather than gradients. For smooth, strongly convex problems the classic two-query method reaches near-optimality with high probability at a cost that adds only a logarithmic dependence on the confidence level over the average-case bound. A stochastic version under bounded noise needs only modest extra cost for the same reliability.

Catalog record

CREATORHaishan Ye
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE114110
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.23613

Browse the full library · About the Tower of Babel